Automated Design of 3D Printed Biomedical Products with Use of Knowledge-Based Engineering (KBE)

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 1
BIOMEDIX
E asmus+ s a egic pa ne ship o Highe Educa ion
BIOMEDICAL INNOVATIONS THROUGH DIGITAL TRANSFORMATION
Chap e 2:
Au oma ed Design o 3D P in ed Biomedical P oduc s wi h Use o
Knowledge-Based Enginee ing (KBE)
P ojec Ti le
Biomedical Inno a ions h ough Digi al T ans o ma ion
o Addi i e Technologies and Knowledge Exchange
KA220-HED-AA8A896B
Ou pu
De elopmen and Publica ion o an e-Book on Biomedical
Inno a ions h ough Digi al T ans o ma ion
Module
Chap e 2
Au oma ed Design o 3D P in ed Biomedical P oduc s
wi h Use o Knowledge-Based Enginee ing (KBE)
Da e o Deli e y
30.06.2025
Au ho s
Filip GÓRSKI, Magdalena ŻUKOWSKA, Na alia
WIERZBICKA, Radosław WICHNIAREK, Dan So in
COMSA, Emilia SMOLAREK
Ve sion
V1
This chap e is accompanied wi h an Augmen ed Reali y applica ion,
o And oid de ices (cellphones and able s). Use he a ached QR
code o download i .
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 2
Con en s
1 In oduc ion ............................................................................................................................. 3
2 Au oma ed Design o Biomedical P oduc s ............................................................................. 4
2.1 Ana omically Cus omized P oduc s ............................................................................. 4
2.2 Basic Design Me hodologies ....................................................................................... 7
Pa ame ic and non-pa ame ic modelling ............................................................ 7
Modelling o ana omical shapes ............................................................................. 9
Designing p os he ic and o ho ic de ices ........................................................... 11
Designing implan s and ope a i e models............................................................ 13
2.3 KBE P inciples ............................................................................................................ 15
2.4 Au oma ed Design Wo k low .................................................................................... 21
In elligen CAD Models ......................................................................................... 21
Au oma ed Design Sys ems .................................................................................. 24
T ansi ioning Be ween T adi ional and Au oma ed Wo k low ............................ 27
2.5 Fini e Elemen Analysis (FEA) o Au oma ically Designed O hopedic P oduc s ...... 29
S a e o he a in he ield o using he ini e elemen me hod (FEM) o
analyzing o hopedic p oduc s ......................................................................................... 29
Fini e elemen analysis o a he apeu ic o hosis manu ac u ed by 3D-p in ing . 39
3 Tools and Techniques ............................................................................................................ 48
3.1 CAD Sys ems wi h KBE Capabili ies ........................................................................... 48
3.2 Design Au oma ion Case S udies............................................................................... 52
Au oMedP in : A Sys em o O hopedic De ice Design ...................................... 52
Au oma ed Design o O hoses ............................................................................ 54
Au oma ed Design o P os heses ......................................................................... 57
3.3 Au oma ion in Addi i e Manu ac u ing Wo k lows .................................................. 61
3.4 Scaling Up: Mass Cus omiza ion o Biomedical P oduc s ......................................... 65
4 Conclusions ............................................................................................................................ 69
4.1 Challenges and Fu u e T ends ................................................................................... 69
4.2 Summa y and Recommenda ions ............................................................................. 70
Re e ences ...................................................................................................................................... 73
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 3
1 In oduc ion
The accele a ing usion o biomedical enginee ing and digi al echnologies has
undamen ally eshaped he de elopmen o medical p oduc s, especially hose equi ing
ana omical cus omiza ion. T adi ional design wo k lows, while e ec i e in gene ic
applica ions, inc easingly all sho in add essing he needs o pe sonalized heal hca e.
Eme ging demands o p ecision, ep oducibili y, and apid deli e y in pa ien -speci ic
ea men call o a pa adigm shi in how medical de ices a e concei ed, designed, and
manu ac u ed. This chap e explo es how such a shi can be achie ed h ough he in eg a ion
o Knowledge-Based Enginee ing (KBE) and addi i e manu ac u ing.
KBE ep esen s a ans o ma i e app oach ha embeds expe knowledge, design logic,
and decision-making p ocesses in o in elligen digi al sys ems. By o malizing and au oma ing
complex biomedical design wo k lows, KBE allows o scalable cus omiza ion— ans o ming
he p oduc ion o o ho ic de ices, p os he ics, implan s, and su gical aids om a isanal asks
in o epea able, knowledge-d i en enginee ing p ocesses. Combined wi h he laye -by-laye
eedom o 3D p in ing, his me hodology enables e icien c ea ion o ana omically
indi idualized solu ions ha ma ch he mo phological and unc ional nuances o each pa ien .
This chap e del es in o he p inciples and applica ions o Knowledge-Based Enginee ing
(KBE) in au oma ing he design o 3D p in ed biomedical p oduc s. I ocuses on
me hodologies, ools, and p ac ical wo k lows ha le e age enginee ing knowledge o
s eamline he design p ocess and ensu e e iciency, accu acy, and ep oducibili y. The main
objec i es o he chap e a e o explo e a ious design me hodologies in biomedical design,
especially using KBE, showcase applica ions and highligh syne gy be ween KBE, design
au oma ion, mass cus omiza ion and addi i e manu ac u ing.
By he end o his chap e , eade s will gain insigh in o he p inciples and p ac ice o
biomedical design au oma ion, he echnological enable s behind scalable mass
cus omiza ion, and he s a egic alue o knowledge o maliza ion in nex -gene a ion
heal hca e enginee ing. In doing so, he chap e unde sco es a cen al ene o mode n
biomedical inno a ion: ha au oma ion, when g ounded in deep domain expe ise, can
econcile indi iduali y wi h indus ial e iciency.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 4
2 Au oma ed Design o Biomedical P oduc s
2.1 Ana omically Cus omized P oduc s
In ecen yea s, he ield o biomedical p oduc design has unde gone a signi ican
ans o ma ion, d i en by he con e gence o digi al echnologies, pa ien -speci ic da a
acquisi ion, and ad anced manu ac u ing p ocesses - chie among hem, addi i e
manu ac u ing. A he hea o his ans o ma ion lies he concep o ana omical
cus omiza ion - he p ac ice o ailo ing medical p oduc s o ma ch he unique mo phological
and unc ional cha ac e is ics o indi idual pa ien s. These p oduc s, o en buil laye by laye
in 3D p in ing p ocesses, ma k a pa adigm shi om adi ional, one-size- i s-all solu ions
owa d highly indi idualized medical in e en ions (Gó ski 2025).
Ana omically cus omized p oduc s, equen ly desc ibed unde he b oade umb ellas o
cus omiza ion, pe sonaliza ion, o indi idualiza ion, a e medical de ices o ools ha
inco po a e p ecise ana omical da a o a pa ien in o hei geome y, s uc u e, o unc ion.
Unlike s anda d p os he ics o implan s, which a e ab ica ed in gene ic sizes and shapes,
ana omically ailo ed solu ions a e bo n om he digi al ep esen a ion o a speci ic human
body. This dis inc ion, hough seemingly sub le a i s , d ama ically al e s he en i e design
and p oduc ion wo k low, in using i wi h complexi y, e hical conside a ions, and egula o y
challenges (Gó ski 2025).
To unde s and he scope and impac o ana omically cus omized p oduc s, i is c ucial o
explo e hei de ining p inciples and con ex s o use. A he mos undamen al le el,
ana omical cus omiza ion le e ages ad anced imaging and scanning echnologies, such as CT,
MRI, o 3D su ace scanning o cap u e accu a e ep esen a ions o a pa ien ’s ana omical
egion o in e es . These da ase s a e hen ans o med in o 3D models using specialized
so wa e, se ing as he basis o he design o p oduc s ha mus in eg a e seamlessly wi h
he pa ien ’s body, whe he ex e nally (e.g., o hoses) o in e nally (e.g., join implan s).
Such indi idualized design is o en a medical necessi y. A join endop os hesis, o
ins ance, may equi e p ecise ma ching o a pa ien ’s bone geome y o ensu e biomechanical
compa ibili y, educe he isk o implan loosening, and op imize load dis ibu ion. Simila ly,
c anio acial implan s o auma o oncology pa ien s mus adhe e o he in ica e opog aphy
o a pa ien ’s skull o es o e symme y, unc ion, and aes he ic appea ance. In hese and
many o he scena ios, ana omical cus omiza ion ensu es ha he medical p oduc does no
me ely se e i s pu pose, i does so as i i we e na u ally in eg a ed in o he biological sys em
i is suppo ing (Gó ski 2025).
The bene i s o ana omical cus omiza ion a e mani old. Fi s ly, i signi ican ly enhances
com o and unc ionali y. De ices shaped o an indi idual’s ana omy a e mo e e gonomic,
educing he likelihood o p essu e poin s, mis i s, o mechanical ailu e. Secondly, i suppo s
be e clinical ou comes. Implan s designed o i p ecisely a e associa ed wi h lowe a es o
e ision su ge y and sho e eco e y imes. Thi dly, i enables inno a ions in ea men
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 5
s a egies: p oduc s ha would be impossible o p ohibi i ely expensi e o p oduce in a
adi ional sub ac i e manu ac u ing p ocess become easible when digi al design and
addi i e manu ac u ing a e employed.
Howe e , he shi om s anda dized o indi idualized p oduc de elopmen in oduces a
hos o new challenges. The ana omical da a used as he ounda ion o design mus be
cap u ed wi h excep ional p ecision and cla i y. This necessi a es high-quali y imaging, eliable
segmen a ion, and expe in e p e a ion. Equally, he design p ocess mus in eg a e
biomechanical and clinical insigh s o ensu e ha he inal p oduc is no jus an accu a e
eplica o a body pa , bu a unc ional subs i u e o enhancemen he eo . The ma e ial
selec ion mus also accoun o biocompa ibili y, s e ilizabili y, and s uc u al in eg i y,
ailo ed no jus o he gene al use-case, bu o he speci ic ana omical and unc ional con ex .
F om a p ocess s andpoin , ana omical cus omiza ion imposes signi ican demands on
wo k low s anda diza ion and epea abili y. T adi ional design wo k lows a e op imized o
ba ch p oduc ion; he cus omized ou e equi es wo k lows ha a e agile, modula , and o en
au oma ed o emain cos -e ec i e. Knowledge-Based Enginee ing (KBE), he subjec o his
book’s co e me hodology, plays a c i ical ole he e - o e ing ools o encapsula e expe
knowledge and design in en in o eusable logic ha can be apidly adap ed o a new pa ien 's
da ase . This app oach add esses one o he co e dilemmas o ana omical cus omiza ion: how
o combine uniqueness in shape wi h e iciency in p ocess.
F om a egula o y pe spec i e, ana omically cus omized medical p oduc s also inhabi a
g ay zone. On he one hand, hey a e p oduc s in ended o medical use, o en classi ied as
Class II o III de ices, demanding igo ous sa e y and e icacy alida ion. On he o he , hei
indi idualized na u e o en places hem ou side he con en ional pipeline o p oduc
ce i ica ion. This can esul in a pa adox: he p oduc is needed u gen ly and is likely supe io
o gene ic al e na i es, ye i s ou e o clinical use is bu dened by legal and bu eauc a ic
cons ain s. This issue is pa icula ly acu e in cases in ol ing implan s o umo pa ien s o
auma ic ims, whe e iming is c i ical and no o - he-shel solu ion exis s.
These egula o y hu dles a e compounded by issues o da a aceabili y, documen a ion,
and ep oducibili y. Each cus om-designed p oduc mus be ully documen ed - no jus in
e ms o i s design a ionale and ma e ial p ope ies, bu in e ms o i s speci ic manu ac u ing
s eps, quali y con ol checkpoin s, and usage condi ions. The equi emen o aceabili y
becomes e en mo e p onounced in he case o Class III implan s, whe e any ad e se e en can
ha e signi ican legal and medical consequences.
Despi e hese challenges, ana omically cus omized p oduc s con inue o gain ac ion,
pa icula ly as heal hca e sys ems inc easingly ecognize he alue o pa ien -speci ic
ea men s. Fu he mo e, he b oade end owa d pe sonalized medicine, d i en by
genomic da a and AI-assis ed diagnos ics, suppo s he a ionale o in eg a ing ana omical
cus omiza ion in o he medical p oduc de elopmen ecosys em. In many espec s,
ana omically ailo ed p oduc s ep esen he ma e ial embodimen o he pe sonalized
medicine pa adigm.

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 6
To cla i y he nuanced e minology used in his domain, he ollowing dis inc ions a e
help ul (Gó ski 2025). Cus omiza ion gene ally e e s o delibe a e adjus men s made wi h
some deg ee o use o clinician inpu - such as he choice o a ma e ial, moun ing mechanism,
o ex e nal con igu a ion. Indi idualiza ion, on he o he hand, e e s o he adap a ion o he
p oduc ’s geome y o unc ionali y o sui a speci ic pa ien , o en wi hou di ec use
in e ac ion in he design. I is his concep - especially ana omical indi idualiza ion - ha
ancho s he me hodologies desc ibed in his book. Finally, pe sonaliza ion e e s o aes he ic
o supe icial modi ica ions ha do no a ec he p oduc ’s p ima y unc ion bu may imp o e
pa ien accep ance and com o , such as adding colo s, ma kings, o deco a i e elemen s o
de ices like o hoses o wheelchai s.
Figu e 2.1. Cus omiza ion ypes o medical p oduc s
While all h ee ypes – cus omiza ion, indi idualiza ion, and pe sonaliza ion - may coexis
in a single p oduc , he ana omical indi idualiza ion p ocess is he mos echnically and
clinically demanding. I d aws di ec ly om ana omical da a, in eg a es medical expe ise, and
le e ages ad anced design me hodologies. This con e gence o biology, enginee ing, and
compu ing is emblema ic o he shi owa d ana omically awa e design - a design philosophy
ha p io i izes no only unc ionali y and aes he ics bu also he seamless in eg a ion o a
p oduc wi h he biological en i onmen in which i ope a es.
Ana omically cus omized p oduc s o e he po en ial o e olu ionize how medical ca e is
deli e ed - making i mo e p ecise, e ec i e, and pa ien -cen e ed. Howe e , ealizing his
po en ial demands new ools, new hinking, and a obus amewo k o design au oma ion.
I is wi hin his con ex ha Knowledge-Based Enginee ing p o ides an indispensable solu ion,
as i allows designe s o embed ana omical logic, clinical ules, and expe know-how di ec ly
in o digi al design en i onmen s - c ea ing sys ems ha can gene a e indi idualized p oduc s
swi ly, eliably, and sa ely.
The nex sec ions o his chap e will del e in o he echnical p inciples, so wa e
en i onmen s, and me hodological s uc u es equi ed o build such sys ems. Bu i is he
ana omical indi iduali y o each pa ien ha emains he ancho – de ining bo h he p oblem
and he p omise o biomedical p oduc design in he e a o digi al medicine.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 7
2.2 Basic Design Me hodologies
Pa ame ic and non-pa ame ic modelling
In he design o ana omically cus omized biomedical p oduc s, Compu e -Aided Design
(CAD) modelling me hods mus accoun o complex, o en i egula geome ies de i ed om
pa ien -speci ic ana omical da a. The wo p ima y app oaches o 3D modelling a e pa ame ic
modelling and non-pa ame ic modelling, each wi h dis inc me hodologies, ad an ages, and
limi a ions in e ms o geome ic con ol, ep oducibili y, and sui abili y o au oma ion.
Be o e discussing he modelling pa adigms, i is impo an o dis inguish be ween basic
CAD model ypes. The mos commonly used in biomedical p oduc de elopmen include:
• solid models – olume ic ep esen a ions wi h mass and ma e ial p ope ies, ypically
used in mechanical design (Cucos e al. 2018).
• su ace models – shell-like s uc u es de ined by bounda y su aces wi hou olume,
use ul o complex ee o m geome ies, especially in au omo i e b anch (Ha ies e
al. 2019).
• mesh models – composed o connec ed polygons ( ypically iangles), o en used o
scanned ana omical da a.
• wi e ame models – skele al ep esen a ions consis ing o edges and e ices; used
p ima ily o e e ence and layou .
• oxel models – olume ic ep esen a ions using disc e e spa ial elemen s; p ima ily
used in simula ions and medical imaging.
Each o hese model ypes (Fig. 2.2) can be c ea ed using ei he pa ame ic o non-
pa ame ic app oaches, al hough he compa ibili y and e ec i eness o each app oach a y
by applica ion.
Figu e 2.2. Model ypes
Pa ame ic modelling is based on he de ini ion o geome y h ough dimensional
pa ame e s, cons ain s, and a sequen ial se o ope a ions. The esul ing model has a his o y
ee, whe e each ea u e is s o ed and can be edi ed e oac i ely. The modelling p ocess is
ypically composed o :
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 8
• ske ch-based ea u es – c ea ion o p o iles (con ined in 2D o 3D space) ollowed by
ope a ions such as ex usion, e olu ion, lo ing, o sweeping;
• no-ske ch ea u es: ope a ions like ille ing, cham e ing, shelling, and pa e ning ha
do no equi e ske ches;
• ans o ma ion ea u es: mi o ing, ansla ing, o a ing, o scaling ea u es and
bodies;
• boolean ope a ions: union, sub ac ion, and in e sec ion o solid bodies (Gó ski 2025).
Pa ame ic modelling is widely adop ed in mechanical and medical de ice enginee ing due
o i s abili y o suppo design i e a ion, euse, and con olled a iabili y. Howe e , when
applied o highly complex o o ganic ana omical geome ies, pa ame ic modelling may
in oduce challenges. Excessi e pa ame e iza ion can lead o ins abili y, pa icula ly when
geome y changes in alida e downs eam ope a ions, esul ing in ailed model egene a ion
(Gó ski 2025).
Non-pa ame ic modelling (Fig. 2.3) does no ely on a ea u e his o y o pa ame e se s.
Ins ead, i is based on di ec geome y manipula ion, o en in he o m o “digi al sculp ing.”
(Alcaide-Ma zal e al. 2013). This is ypical o mesh-based o su ace-based models de i ed
om medical imaging o 3D scanning. Key cha ac e is ics include:
• no model his o y o o de ed ea u e ee,
• geome y is modi ied h ough di ec de o ma ion, displacemen , o smoo hing
ope a ions,
• local modi ica ions do no p opaga e ac oss he en i e model unless manually applied,
• common in ools suppo ing mesh edi ing (e.g., Blende , Meshmixe ) o di ec
modelling (e.g., Siemens NX, Au odesk Fusion’s di ec mode).
Figu e 2.3. Non-pa ame ic modelling (digi al sculp ing) – s ages o wo k
Non-pa ame ic me hods o e highe lexibili y when wo king wi h ana omical shapes bu
a e mo e di icul o s anda dize o au oma e. The absence o consis en pa ame e s makes
hem less sui ed o wo k lows equi ing ep oducibili y o ba ch design gene a ion.
Compa ison o he wo app oaches is p esen ed in Table 2.1.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 9
Table 2.1. Compa ison o pa ame ic and non-pa ame ic app oaches, based on (Gó ski
2025)
Fea u e
Pa ame ic Modelling
Non-Pa ame ic Modelling
Geome y con ol
High ( h ough cons ain s and
pa ame e s)
Local, in ui i e bu less p ecise
His o y acking
Full ea u e his o y (edi able)
No global his o y; local
modi ica ions only
Sui abili y o au oma ion
High (suppo s ule-based
design, empla es)
Low (manual, ha d o sc ip o
eplica e)
Compa ibili y wi h ana omical
da a
Mode a e ( equi es i ing
ana omical da a o clean
geome y)
High (can ope a e di ec ly on
scanned da a)
Pe o mance wi h complex
geome y
May become uns able o slow
wi h complexi y
E icien , eal- ime
manipula ion possible
Ou pu consis ency
High
Va iable, depends on use inpu
In eg a ion in enginee ing
wo k lows
Excellen (mechanical CAD
in eg a ion)
Limi ed; o en used in ea ly
concep o isualiza ions
Modelling o ana omical shapes
Modelling ana omical shapes o biomedical p oduc design p esen s dis inc challenges
no encoun e ed in s anda d mechanical enginee ing. Ana omical geome y is inhe en ly
complex, i egula , and o en pa ien -speci ic. E icien modelling equi es p ope s a egies
o inco po a ing ana omical da a in o CAD wo k lows, main aining ep oducibili y, and
ensu ing compa ibili y wi h egula o y s anda ds.
The modelling p ocess begins wi h he acquisi ion o ana omical da a, ypically in one o
he ollowing o ms (Gó ski 2025):
• An h opome ic measu emen s ( om a lases o manual ools),
• 2D medical images (X- ay, ul asound, pho og aphs),
• 3D meshes ob ained ia scanning o segmen a ion o medical imaging da a (CT, MRI),
• Ex ac ed geome ic ea u es (e.g. cu es, planes, landma ks) de i ed om 3D meshes.
Among hese, 3D iangle meshes a e mos commonly used in he design o cus omized
medical p oduc s. These meshes can be manipula ed di ec ly o used as a e e ence o
gene a ing pa ame ic geome y. Two p ima y modelling s a egies a e applied: he mesh-
based app oach and he pa ame ic app oach, wi h in e media e hyb id wo k lows also in use.
The mesh-based app oach uses ana omical meshes as he ounda ion o p oduc design.
This me hod can be subdi ided in o (Gó ski 2025):
• pu e mesh pa hway: he en i e model, including modi ica ions and de ices, is c ea ed
using mesh-edi ing ools (e.g., cu , o se , b idge), wi hou ansi ioning o solids o
su aces.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 16
and scalabili y. As cus omiza ion becomes he no m a he han he excep ion, especially in
pa ien -speci ic medical applica ions, he adi ional ial-and-e o o manually guided design
app oaches become insu icien . This is whe e Knowledge-Based Enginee ing (KBE) eme ges
as a c ucial pa adigm, acili a ing he embedding o expe knowledge di ec ly in o design
p ocesses h ough compu a ional sys ems (Zawadzki 2018, Gó ski e al. 2016).
In enginee ing, knowledge is mo e han aw da a o in o ma ion - i ep esen s an
o ganized s uc u e o ela ionships, ules, p ocedu es, and heu is ics buil on expe ience and
alida ed p ac ice. I is con ex ual, ac ionable, and o ms he ounda ion o decision-making
ac oss all s ages o p oduc de elopmen .
To dis inguish clea ly (Fig. 2.9):
• Da a a e aw, unp ocessed ac s (e.g., scanne ou pu , nume ical measu emen s).
• In o ma ion is da a in e p e ed wi hin con ex (e.g., iden i ied dimensions o a esidual
limb).
• Knowledge is he abili y o apply ha in o ma ion app op ia ely, o en codi ied as
ules, p ocedu es, o pa ame ic logic (e.g., i esidual limb leng h is below h eshold
X, use con igu a ion Y).
In biomedical p oduc de elopmen , knowledge a ises om mul iple disciplines -
enginee ing, medicine, ma e ial science, and manu ac u ing. The e o e, i s e ec i e cap u e
and ep esen a ion a e cen al o au oma ion and cus omiza ion.
Fig. 2.9. Da a, in o ma ion and knowledge wi hin con ex o medical p oduc design
Knowledge in enginee ing is di e se in o m and unc ion, and i s classi ica ion is c ucial
o e ec i e cap u e, ep esen a ion, and au oma ion. Wi hin he con ex o biomedical
enginee ing - especially he design o indi idualized medical de ices - di e en ypes o
knowledge play dis inc oles, om geome ic ules embedded in pa ame ic CAD models o
clinical decisions made in ui i ely by expe ienced specialis s. Unde s anding he na u e o his
knowledge is a p e equisi e o building obus and scalable Knowledge-Based Enginee ing
(KBE) sys ems. The below conside a ions a e based on au ho s’ knowledge, as well as on he
book (Rzydzik 2013).
1. Decla a i e Knowledge

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 17
Decla a i e knowledge e e s o s a ic ac s and ules abou a speci ic domain. I desc ibes
wha is, wi hou necessa ily indica ing how o use ha in o ma ion. In he con ex o
enginee ing design, decla a i e knowledge includes:
• Ana omical dimensions and ole ances,
• Ma e ial p ope ies,
• Logical cons ain s (e.g., "sc ew hole mus no in e sec ne e canal"),
• Clinical condi ions and hei de ini ions (e.g., “spas ici y in ol es inc eased muscle
one”).
This o m o knowledge is o en exp essed as pa ame e ables, design cons ain s,
enginee ing handbooks, o clinical guidelines. Decla a i e knowledge is ela i ely easy o
encode in da abases o knowledge ules and is cen al o he cons uc ion o CAD empla es
and design e i ica ion ou ines.
2. P ocedu al Knowledge
P ocedu al knowledge consis s o me hods, sequences, and ope a ions - i de ines how o
do some hing. In design enginee ing, i includes:
• The s eps o gene a ing a 3D model om a mesh,
• Manu ac u ing p ocess plans (e.g., p in o ien a ion, suppo s a egy),
• Simula ion wo k lows (e.g., se ing bounda y condi ions o FEA),
• Pos -p ocessing ope a ions o alida ion s eps.
P ocedu al knowledge is essen ial o au oma ion because i can be o malized in o logic
ees, p ocess chains, sc ip s, o CAD mac os. In KBE sys ems, p ocedu al knowledge o en
con ols he execu ion o design sequences o condi ional b anching in esponse o inpu
pa ame e s.
3. Episodic Knowledge
Episodic knowledge e e s o con ex ual o ime-dependen knowledge - knowledge o
e en s, expe iences, o speci ic cases. I is de i ed om eal-wo ld p ojec s and pas
occu ences and may include:
• A su geon’s ecall o a pa icula ly di icul implan placemen ,
• A eco d o p e ious design e isions and associa ed ou comes,
• Case-based easoning examples (e.g., “in a simila case, model B caused skin
i i a ion”).
While episodic knowledge is less s uc u ed han decla a i e o p ocedu al ypes, i is
c i ical o lea ning om expe ience and imp o ing sys em pe o mance. In enginee ing
con ex s, i o ms he basis o his o ical da abases, eedback sys ems, and design euse
s a egies.
4. On ological Knowledge
On ological knowledge conce ns he ca ego iza ion and ela ionships wi hin a domain. I
de ines how di e en en i ies ela e o one ano he , which is i al o s uc u ing knowledge
sys ema ically. Examples in biomedical design include:
• Classi ica ion o implan s by body egion (e.g., maxillo acial s. o hopedic),
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 18
• Rela ions be ween ana omical s uc u es (e.g., “ he adius is connec ed o he ulna”),
• G ouping o o ho ic ypes (e.g., passi e s. ac i e).
On ologies a e used o c ea e seman ic ne wo ks and knowledge g aphs, helping sys ems
“unde s and” he domain con ex and imp o ing knowledge e ie al, in e ope abili y, and
easoning. In complex KBE sys ems, on ological s uc u es suppo ule hie a chies and
decision-making logic.
Fig. 2.10. Types o knowledge
Complemen a y o he s uc u al classi ica ion (Fig. 2.10), enginee ing knowledge can also
be dis inguished by i s deg ee o o mali y and accessibili y - ha is, whe he i is explici o
aci (Zawadzki 2018).
1. Explici Knowledge
Explici knowledge is o malized, codi ied, and easily communica ed. I is accessible in
w i en o m and includes:
• Enginee ing d awings and CAD models,
• Design ules and checklis s,
• Medical imaging p o ocols and segmen a ion p ocedu es,
• Pa ame e ized empla es and sp eadshee s.
This o m o knowledge is he easies o in eg a e in o KBE sys ems. Fo example, a ule
s a ing “i socke diame e > 100 mm, use h ee suppo ibs” can be p og ammed di ec ly in o
a CAD au oma ion sc ip . Explici knowledge is also sha eable and ans e able be ween
enginee s, clinicians, and sys ems.
2. Taci Knowledge
Taci knowledge is expe ien ial, in ui i e, and di icul o a icula e. I esides in he minds
o expe s and is acqui ed h ough p ac ice a he han o mal ins uc ion. In he biomedical
con ex , examples include:
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 19
• A p os he is ’s skill in adjus ing a socke by eel,
• A su geon’s expe ience-based decisions in choosing incision poin s,
• An enginee ’s in ui ion abou whe e s ess concen a ions may occu .
Taci knowledge is essen ial bu ha de o sys ema ize. Cap u ing i o en in ol es:
• In e iews wi h expe s,
• Obse a ion o skilled asks,
• Use o collabo a i e wo kshops o design hinking sessions.
Al hough challenging o encode, e o s o ex ac aci knowledge a e i al o
ansi ioning om manual o au oma ed o semi-au oma ed p ocesses, pa icula ly in pa ien -
speci ic applica ions.
These wo classi ica ions - s uc u al (decla a i e, p ocedu al, episodic, on ological) and
unc ional (explici s. aci ) - a e no mu ually exclusi e. Ins ead, hey in e sec and
complemen each o he . Fo ins ance:
• A p ocedu al wo k low (explici knowledge) may ha e e ol ed om accumula ed
episodic expe iences ( aci knowledge).
• An on ological s uc u e (explici ) may be in o med by he expe ’s in ui i e
unde s anding o ana omical ela ionships ( aci ).
In KBE sys em de elopmen , i is c ucial o:
• S a wi h explici , decla a i e and p ocedu al knowledge (easie o o malize),
• G adually inco po a e episodic knowledge (e.g., ia da abases o case logs),
• Use in e iews and ield s udies o ex e nalize and o malize aci knowledge,
• In oduce on ological amewo ks o consis ency and logic s uc u ing.
By combining hese knowledge ypes wi hin he sys em’s a chi ec u e - ypically in he
o m o a ule base, pa ame e maps, and decision logic - design enginee s can achie e
au oma ion o complex, cus omizable wo k lows such as hose ound in he de elopmen o
o hoses, implan s, and su gical models.
The p ocess o in eg a ing enginee ing knowledge in o design sys ems in ol es se e al key
s ages – iden i ica ion, acquisi ion and ep esen a ion (Fig. 2.11), acco ding o ea lie wo k
(Zawadzki 2018, Gó ski e al. 2016).
1. Iden i ica ion
Knowledge sou ces mus i s be ecognized. In he con ex o biomedical enginee ing,
hey may include:
• Expe in e iews (clinicians, p os he is s, enginee s),
• Exis ing p oduc documen a ion,
• Simula ion o es esul s,
• Li e a u e and clinical s udies,
• Obse ed ou comes om p e ious p ojec s.
2. Acquisi ion
This phase in ol es ex ac ing and eco ding he knowledge using me hods such as:
• Manual documen a ion (e.g., decision ees, lowcha s),
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 20
• Case-based easoning (CBR),
• Re e se enginee ing om exis ing p oduc s,
• Da a mining and AI echniques o pa e n ecogni ion.
3. Rep esen a ion
Knowledge is s o ed in s uc u ed o ma s wi hin a Knowledge Base, which may include:
• Rule-based logic (e.g., IF limb leng h < X THEN use con igu a ion A),
• Seman ic ne wo ks o on ologies,
• Pa ame e ized CAD empla es,
• Tables o decision c i e ia o con igu a ion op ions,
• Anno a ed images o diag ams.
E ec i e knowledge ep esen a ion enables eusabili y, alida ion, and au oma ic
applica ion in KBE sys ems.
Fig. 2.11. In eg a ing knowledge in o enginee ing sys ems
The applica ion o KBE in biomedical enginee ing b ings measu able bene i s, especially
when dealing wi h la ge numbe s o cus omized p oduc s ha sha e a common design logic.
Key ad an ages include:
• design cycle ime educ ion: h ough au oma ion o epe i i e asks such as i ing,
empla e gene a ion, and documen a ion, design ime pe pa ien can be educed
om days o minu es.
• minimiza ion o human e o : by elying on alida ed ules and s uc u ed logic, e o s
ela ed o o e sigh , inexpe ience, o miscommunica ion a e signi ican ly dec eased.
• apid i e a ion and upda es: changes in design c i e ia o egula o y cons ain s can be
quickly implemen ed ac oss all empla es and con igu a ions.
• e icien documen a ion gene a ion: bill o ma e ials, echnical d awings, and CAM iles
can be gene a ed di ec ly om he knowledge base, ensu ing consis ency.
Fo ins ance, in o hosis o implan design, whe e ce ain geome ical ules epea ac oss
cases, a well-de eloped KBE sys em can gene a e a pa ien -adap ed CAD model om a limi ed
inpu da ase (e.g., scanned geome y and an h opome ic a ibu es), bypassing he need o
a ully manual modelling session.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 21
Scalabili y is a c i ical conce n in he shi om adi ional, hand-c a ed de ices o digi ally
manu ac u ed, pe sonalized p oduc s. KBE sys ems enable mass cus omiza ion by combining
ule-d i en au oma ion wi h lexible pa ame iza ion.
In p ac ice, a KBE-based wo k low o a p os he ic socke o su gical guide may ope a e as
ollows:
1. Inpu : pa ien -speci ic da a (e.g., limb scan, age, diagnosis).
2. P ocessing: au oma ed applica ion o ules and pa ame e s based on clinical and
design knowledge.
3. Ou pu : eady- o-manu ac u e CAD model, echnical d awings, and documen a ion.
Because he unde lying ules a e eusable and gene alizable, his p ocess can be epea ed
o hund eds o unique pa ien s wi h minimal manual in e en ion. Such scalabili y is i al o :
• clinics se ing la ge popula ions,
• manu ac u e s p o iding cus omizable de ices,
• esea ch cen e s de eloping lib a ies o case-adap able models.
KBE is pa icula ly well-sui ed o ana omically d i en design, whe e geome y is unique
bu logic and s uc u e a e epea ed. I ans o ms he knowledge o expe ienced
p o essionals in o sha eable, codi ied digi al asse s - ensu ing ha expe ise is no los , and
p oduc quali y is consis en ac oss la ge olumes.
Knowledge-Based Enginee ing se es as he backbone o digi al au oma ion in he
biomedical p oduc design wo k low. I allows o he codi ica ion o bo h s uc u ed
knowledge ( ules, p ocedu es) and p e iously aci expe ise in o in elligen design sys ems.
In he nex chap e s, his concep ual ounda ion will be expanded wi h p ac ical me hods and
implemen a ion s a egies in he con ex o biomedical enginee ing, including speci ic
applica ions o o ho ics, p os he ics, su gical guides, and implan s.
2.4 Au oma ed Design Wo k low
In elligen CAD Models
The de elopmen o ana omically cus omized biomedical de ices inc easingly elies on
design wo k lows ha a e capable o e icien ly handling high a iabili y wi hou
comp omising p ecision o epea abili y. To mee his demand, mode n enginee ing p ac ice
has in oduced in elligen CAD models a e highly s uc u ed, pa ame ic, and logic-d i en
design a i ac s capable o au oma ic geome y gene a ion based on case-speci ic inpu
(Ska ka e al. 2023, Gó ski 2025). These models o m he co e o many Knowledge-Based
Enginee ing (KBE) sys ems, suppo ing design au oma ion o o ho ic shells, p os he ic
socke s, implan s, su gical guides, and o he pa ien -adap ed solu ions.
An in elligen CAD model di e s undamen ally om a s a ic design. Ra he han encoding
a single shape o geome y, i ep esen s an en i e amily o po en ial designs go e ned by
embedded knowledge s uc u es. These may include dimensional pa ame e s, geome ic
cons ain s, logical condi ions, and p ocedu al ules. When p o ided wi h a new se o inpu
alues, such as ana omical measu emen s o diagnos ic classi ica ions, he model egene a es

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 22
a geome y a ian ha con o ms o all necessa y cons ain s and s anda ds. This egene a ion
is de e minis ic and epea able, o e ing a eliable al e na i e o manual edesign o each
case.
F om a echnical s andpoin , he a chi ec u e o an in elligen model is buil on h ee
in e ela ed laye s. The i s is he pa ame ic geome y, whe e each ea u e (such as cu es,
su aces, o solid bodies) is de ined in ela ion o a se o pa ame e s a he han ixed
dimensions. The second is he ela ional s uc u e, consis ing o cons ain s and dependencies
ha main ain in e nal consis ency - ensu ing, o ins ance, ha a sc ew hole always lies a ixed
dis ance om an edge, o ha he wall hickness scales wi h he wid h o an o ho ic ame.
The hi d laye is he ule logic, implemen ed using condi ional s a emen s o ex e nally linked
con ol iles. This laye enables model beha io o change dynamically based on inpu -
ac i a ing o supp essing ea u es, al e ing ea u e ypes, o es uc u ing he model wo k low
(Zawadzki 2018, Ska ka e al. 2023, Gó ski 2025).
The pa ame e s de ining a gi en a ian o a model a e usually bound oge he h ough
he design able – a sp eadshee o hype ex ile wi h a o malized s uc u e, whe e
pa ame e s – and o en hei ma hema ical and logical ela ions – a e s o ed and a e being
uploaded o he CAD model. Design able can be easily changed om he ou side, ei he
manually o au oma ically, ia a ious sys ems o con igu a o s and da a ex ac ion
mechanisms. An example is shown in Fig. 2.12, desc ibing a modula hand p os hesis model,
ully p esen ed in a p e ious publica ion (Gó ski e al. 2025).
Fig. 2.12. Design able o a modula hand p os hesis, depic ing ela ional s uc u e
be ween pa ame ic geome y o i s pa s, as desc ibed in (Gó ski e al. 2025)
The de elopmen o such models ollows a s uc u ed me hodology, which can be
gene alized in o he ollowing s eps:
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 23
1. Concep de ini ion and a iabili y analysis. The i s s age in ol es iden i ying he
p oduc amily o be au oma ed and analyzing he expec ed a iabili y ac oss cases.
In medical design, his o en includes an h opome ic anges, ana omical
classi ica ions, and modula op ions (e.g. size anges, limb ypes, ixa ion me hods).
This s ep also in ol es selec ing he e e ence geome y o design logic o be
p ese ed ac oss a ian s.
2. Pa ame e iden i ica ion. Based on he a iabili y analysis, he key design
pa ame e s a e selec ed and o mally de ined. These can be geome ic (leng hs,
adii, angles), logical (e.g. componen p esence o absence), o ca ego ical (e.g.
ype o ampu a ion o bone de ec ). Each pa ame e mus be mapped o one o
mo e ea u es in he model and assigned limi s, uni s, and dependencies.
3. Model cons uc ion in CAD en i onmen . Using a pa ame ic CAD sys em (e.g.
Siemens NX, SolidWo ks, Fusion 360), he model is cons uc ed using ske ch-based
and ea u e-based ope a ions, ensu ing all geome y is ully cons ained and linked
o he p ede ined pa ame e s. Ca e is aken o p ese e model s abili y ac oss he
expec ed pa ame e ange, and egene a ion e o s a e minimized h ough obus
cons ain schemes.
4. Embedding logic and condi ional beha io . Logic s uc u es a e in oduced o
allow dynamic changes in model opology. This may in ol e supp essing o
enabling ea u es based on dimensional h esholds o ca ego ical choices. Fo
example, an elbow o hosis may include a join mechanism only i he selec ed
a ian co e s he elbow a ea. Such beha io is ypically implemen ed using i -else
s a emen s, design ables, o sc ip ing (e.g. iLogic in In en o , D i eWo ks in
SolidWo ks).
5. Ex e nal da a in eg a ion. To suppo au oma ion, he model is linked o ex e nal
da a sou ces such as sp eadshee s, XML iles, o da abases. This allows non-CAD
use s (e.g. clinicians o echnicians) o inpu pa ien -speci ic alues in o a o m o
da a en y in e ace, which hen d i es he egene a ion o he CAD model
au oma ically, wi hou opening he model ile manually.
6. Valida ion and e o handling. The inal model is es ed ac oss i s pa ame e space
o ensu e alid egene a ion unde all easonable inpu scena ios. Bounda y cases
and e o condi ions a e iden i ied and mi iga ed h ough cons ain s, wa nings, o
allback beha io s. Documen a ion is c ea ed o de ine pa ame e oles,
dependencies, and usage p o ocols.
This me hodology is pa icula ly e ec i e in he biomedical ield, whe e base p oduc
con igu a ions o en epea , bu pa ien ana omy in oduces a iabili y. Fo example, a
ans adial p os he ic socke may ollow he same ana omical logic ac oss cases bu di e
signi ican ly in diame e , leng h, and suppo ib con igu a ion. An in elligen CAD model
allows hese di e ences o be cap u ed and gene a ed au oma ically, imp o ing design speed,
consis ency, and aceabili y.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 24
One o he key ad an ages o in elligen models is hei compa ibili y wi h au oma ion
sys ems. Once p ope ly s uc u ed, hey can be ope a ed by ex e nal in e aces, sc ip s, o
web-based pla o ms, allowing models o be gene a ed on demand di ec ly om clinical da a
o 3D scans. This opens he possibili y o in eg a ing in elligen CAD models in o semi-
au oma ed design pipelines o in-clinic cus omiza ion o indus ial-scale indi idualized
manu ac u ing.
Howe e , se e al echnical limi a ions mus be acknowledged. In elligen models a e
sensi i e o poo pa ame e managemen , and small changes in logic o inpu can lead o
opological e o s, especially in complex geome ies. Mo eo e , adap ing his me hodology o
o ganic shapes - such as acial ana omy o i egula bone su aces - equi es hyb id solu ions,
whe e scanned mesh da a is me ged wi h pa ame ic subs uc u es. Despi e hese challenges,
he in elligen CAD model emains one o he mos e ec i e ools o implemen ing scalable
cus omiza ion in biomedical design wo k lows.
Au oma ed Design Sys ems
As he demand o indi idualized biomedical p oduc s g ows - d i en by clinical need,
pa ien expec a ions, and echnological easibili y - manual design wo k lows quickly become
unsus ainable. The combina ion o high a iabili y, medical p ecision equi emen s, and ime-
sensi i e deli e y calls o s uc u ed sys ems capable o consis en ly gene a ing alid, pa ien -
speci ic designs. Au oma ed design sys ems p o ide his capabili y, unc ioning as so wa e-
d i en amewo ks ha ans o m ana omical o clinical da a in o manu ac u ing- eady digi al
models wi h minimal human in e en ion.
An au oma ed design sys em in eg a es mul iple echnical laye s: da a inpu in e aces,
ule-based decision logic, in elligen CAD modelling, and ou pu gene a ion modules
(Zawadzki 2018). These componen s a e o ches a ed o execu e a design p ocess end- o-end,
allowing clinicians, echnicians, o au oma ed se e s o igge geome y gene a ion wi hou
engaging in adi ional CAD modelling. The esul is a s eamlined, epea able, and scalable
solu ion o cus omized p oduc de elopmen .
A he co e o such sys ems lies he p inciple o design knowledge o maliza ion. Rules
ha we e once implici ly ollowed by expe ienced designe s - such as geome ic adjus men s
o com o , sa e dis ances om ana omical landma ks, o s anda d placemen s o suppo
ea u es - mus be codi ied in a o m ha can be p ocessed algo i hmically. Depending on
sys em a chi ec u e, his codi ica ion may ake he o m o pa ame e ables, decision ees,
sc ip -based au oma ion, o ull- ea u ed KBE logic modules embedded in CAD en i onmen s
(Zawadzki 2018, Gó ski e al. 2016).
A ypical au oma ed design sys em o a biomedical applica ion begins wi h pa ien -
speci ic da a acquisi ion. This can include nume ical alues (e.g. leng hs, diame e s, weigh ),
3D scans (mesh o poin cloud), o imaging-de i ed models (segmen ed STL iles om CT/MRI
da a). The da a is cap u ed using s uc u ed o ms, guided inpu in e aces, o di ec de ice
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 25
in eg a ion. I is hen alida ed agains p ede ined c i e ia o ensu e i alls wi hin accep able
limi s - bo h o pa ien sa e y and o sys em s abili y.
Following alida ion, he sys em p ocesses he da a h ough a se o de e minis ic ules.
These may include geome ic calcula ions, condi ional logic, and pa con igu a ion selec ion.
Fo ins ance, a lowe -limb o hosis design sys em may use leg leng h o scale he b ace shell,
oo wid h o de ine sole cu a u e, and a diagnosis code o selec among di e en hinge
mechanisms. Each ule is designed o igge speci ic ac ions in he model, and mul iple ules
may ope a e simul aneously o hie a chically o handle complex in e dependencies.
The in elligen CAD model a he hea o he sys em is con igu ed o accep hese inpu s,
egene a e acco dingly, and esol e all embedded ea u es and ela ionships. Depending on
he complexi y o he design, he model may be opened and egene a ed in a backg ound CAD
p ocess o ully embedded in a headless au oma ion amewo k. In bo h cases, he use is
shielded om he geome ic complexi y o he design p ocess, in e ac ing only wi h he inpu
and ou pu s ages.
The ou pu o he sys em ypically includes one o mo e o he ollowing:
• A ully de ined 3D CAD model in a neu al o ma (e.g. STEP, IGES),
• A mesh ile eady o addi i e manu ac u ing (e.g. STL, 3MF),
• Technical d awings o documen a ion (PDF, DXF),
• Associa ed me ada a o acking, labelling, o pos -p ocessing.
Some sys ems also p oduce pos -p ocessing ins uc ions (e.g. o suppo emo al, pa
labeling), bills o ma e ials, o au oma ically gene a ed epo s summa izing he design logic
and pa ien -speci ic inpu s used.
A concep ual o basic ope a ion o an au oma ed design sys em o design o cus omized
medical de ices is shown in Fig. 2.13. Based on da a inpu , ule-based decision logic go e ns
he CAD modelling, using also con igu a ion da a inpu , and gene a es a mesh model as an
ou pu o addi i e manu ac u ing.
Au oma ed design sys ems may be implemen ed in a ious a chi ec u es depending on
he wo k low equi emen s. In clinic-based sys ems, he au oma ion may un locally on a
wo ks a ion, di ec ly linked o scanning equipmen and 3D p in e s. In indus ial se ings,
sys ems a e o en cloud-based o se e -con olled, allowing ba ch p ocessing o mul iple
cases and in eg a ion wi h ERP o PLM sys ems. Web-based on -ends enable emo e da a
en y and cus ome in e ac ion, while backend au oma ion engines execu e model gene a ion
asynch onously.
One o he c i ical success ac o s in implemen ing au oma ed design sys ems is
obus ness. Unlike manual wo k lows, au oma ed sys ems mus ope a e wi hou supe ision,
handling a wide ange o inpu cases wi hou e o s. This places s ong emphasis on inpu
alida ion, model egene a ion es ing, and excep ion handling. I a pa ien ’s ana omical da a
is ou side he expec ed ange, he sys em mus espond app op ia ely - ei he by adjus ing he
model sa ely, lagging he case o manual e iew, o ejec ing i wi h a meaning ul
explana ion.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 32
cellula me ama e ials. Plesec and Ha ih p oposed he concep o a p os he ic line wi h a
gy oid-inspi ed la ice s uc u e, aiming o enhance com o by be e dis ibu ing p essu e
(Plesec & Ha ih, 2024). The gy oid uni -cell was chosen o i s iso opic mechanical p ope ies
and e iciency in load dis ibu ion. To model his ma e ial in FEA, he esea che s i s
pe o med uniaxial comp ession es s on sample la ices and used he da a o calib a e a
cus om mul ilinea ma e ial model. This calib a ed FE model, when compa ed agains a
con en ional solid silicone line , showed he cellula line could signi ican ly educe peak
con ac p essu es a he limb-p os hesis in e ace while ailo ing i s s i ness o use needs.
No ably, a so e la ice line yielded much lowe donning p essu es (ideal o sensi i e o
seden a y use s), whe eas a s i e la ice p o ided be e suppo and s abili y du ing high-
impac gai , all o which being cap u ed by he FEM simula ions. This is a clea example o how
bio-inspi ed ma e ial a chi ec u es, combined wi h p ope ma e ial modeling in FEA, a e
expanding he design space o p os he ic com o and pe o mance.
Composi e ma e ials emain a ho opic. Beyond adi ional ca bon- ibe , na u al ibe -
ein o ced composi es (NFRCs) a e being in es iga ed o sus ainable, ligh weigh p os he ic
componen s. Cas o-F anco and his cowo ke s (Cas o-F anco e al., 2024) no e ha NFRCs
(using ibe s like lax, hemp, e c.) ha e lowe cos and en i onmen al impac , and discuss hei
ad an ages and d awbacks in p os he ic design. Thei e iew emphasizes ha compu a ional
biomechanical models used in FE simula ions a e essen ial o e alua e he e ec i eness o
new ma e ials in p os he ic designs. Fo ins ance, FEM can simula e how changing he ibe
o ien a ion o laye ing in a p os he ic socke a ec s i s s i ness and he p essu e on he
esidual limb. O e all, he end is owa d mul i unc ional ma e ials (ligh weigh , s ong,
complian whe e needed) and using FEM o cap u e hei complex beha io . Whe he i is a
ca bon- ibe oo keel o a 3D-p in ed nylon elbow o hosis, accu a e inpu da a (elas ic
moduli, yield s eng hs, damping cha ac e is ics, e c.) and ad anced ma e ial models in FEM
a e enabling enginee s o push he bounda ies o design while ensu ing sa e y.
3. Valida ion echniques (in i o & in i o)
Wi h he inc easing complexi y o FE models, igo ous alida ion has become mo e
impo an han e e . O e he las i e yea s, esea che s ha e e ined bo h in i o (lab-based)
and in i o alida ion me hods o ensu e FEM p edic ions ma ch eal-wo ld beha io .
In i o alida ion o en in ol es mechanical es ing o p os he ic/o ho ic componen s
unde con olled condi ions, hen compa ing he esul s o simula ion. Fo example, o
alida e an FE model o a p os he ic oo , Bala amak ishnan and his cowo ke s
(Bala amak ishnan e al., 2020) scanned a comme cially a ailable oo (O obock SACH oo ),
de i ed ma e ial p ope ies h ough ensile and comp ession es s, and hen measu ed he
shape o he oo (e ec i e cu a u e du ing s ance) using a cus om es ig. The FE model
inco po a ed he same geome y and a hype elas ic ma e ial law, and ema kably, he
di e ence be ween he simula ed and expe imen al oll-o e cu a u e was only abou 7.5%.
Such a small e o demons a ed he accu acy o he model in eplica ing he beha io o he

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 33
oo , gi ing con idence ha he FE model could be used o explo e design weaks i ually. In
o he cases, esea che s use s anda d ma e ial es ing machines o eco d o ce–displacemen
cu es o an o ho ic de ice (e.g., de lec ing an AFO o wis ing a knee b ace) and hen check
ha he FE simula ion p oduces he same s i ness o de o ma ion pa e n. S ain gauge
measu emen s on de ice s uc u es a e also common – o example, bonding gauges on a
p os he ic pylon o b ace od o eco d s ain unde load, and e i ying he s ain ou pu o
he FE model a he same loca ions. This helps ca ch any disc epancy in ma e ial modeling o
bounda y condi ions in he simula ion.
In i o alida ion echniques ha e ad anced, especially o aspec s like in e ace p essu e
and i ness which a e ha d o eplica e in bench es s. A no able me hod is using p essu e
senso s o p essu e ma s be ween he pa ien ’s limb and he de ice. Recen socke s udies
placed hin esis i e p essu e senso s on he esidual limb inside a p os he ic socke and
measu ed he p essu e dis ibu ion while he use s ood o walked, hen compa ed i o FE
p edic ions (Shah & Rehman, 2025). In one case, he p essu e p edic ed by he FE model a
key poin s di e ed by only abou 8.5 kPa om he senso eadings, con i ming ha he
simula ion closely cap u ed eali y (Shah & Rehman, 2025). Mo ion cap u e and gai analysis
p o ide ano he laye o alida ion: esea che s eco d a subjec walking wi h he
p os hesis/o hosis o ge g ound eac ion o ces, limb kinema ics, and e en muscle ac i a ion
pa e ns. These da a can d i e he FE model (as bounda y condi ions o loading inpu s), and
he ou pu o he model – such as he de o ma ion o a p os he ic oo o he s ess in a b ace
du ing gai – can be indi ec ly alida ed i he o e all beha io (e.g., s ide leng h, ene gy
e u n) ma ches expe imen al da a.
In he ealm o spinal o hoses (like scoliosis b aces), imaging and de ailed measu emen s
a e used o alida ion. Techniques such as digi al image co ela ion (DIC) o lase -based
me hods (e.g., Elec onic Speckle Pa e n In e e ome y) ha e been applied o b aces o
measu e how much hey de o m o how much load hey apply o he body (Guan e al., 2020).
Guy and Aubin c ea ed a pa ien -speci ic FE model o a Bos on scoliosis b ace and hen
pe o med in i o measu emen s o he p essu e applied by he b ace on he o so and he
co ec ion achie ed in spine cu a u e (Guy and Aubin, 2023). The p elimina y esul s showed
good ag eemen wi h FE bu also highligh ed he need o accoun o pa ien b ea hing and
muscle one in he model o uly accu a e p edic ions. Such mixed alida ion app oaches
(combining senso da a, imaging, and mo ion analysis) a e becoming he gold s anda d o
build ealis ic FE models.
I is wo h no ing ha alida ion is an i e a i e p ocess. Disc epancies be ween FEM and
expe imen lead o model e inemen s – o example, upda ing ma e ial p ope ies o con ac
de ini ions. O e he las i e yea s, he communi y has also published guidelines and e iews
on alida ion bes p ac ices. Al-Fakih and his cowo ke s (Al-Fakih e al., 2016) e iewed
decades o p os he ic socke in e ace s udies and ca alogued senso echnologies o
measu ing s esses in socke s, p o iding a baseline o cu en esea che s. Mo e ecen ly,
s anda ds o epo ing FE alida ion like including e o quan i ica ion, as done in he
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 34
p os he ic oo oll-o e s udy (Bala amak ishnan e al., 2020) ha e imp o ed he
anspa ency and eliabili y o simula ion in his ield.
4. Clinical applica ions and case s udies
FEM ad ancemen s a e no jus heo e ical – hey a e inc easingly ansla ing in o clinical
p ac ice and de ice de elopmen . One majo applica ion is in he design o pa ien -speci ic
de ices. Cus om p os he ic socke s a e a p ime example: clinicians now use 3D scans o an
ampu ee’s limb o c ea e a digi al model, and FEM is employed o assess he socke i and
p edic p essu e poin s be o e manu ac u ing. Shah and Rehman (Shah & Rehman, 2025)
gene a ed a To al Su ace Bea ing socke om a CT scan and e alua ed i s biomechanical
beha io unde load wi h FEA, hen ab ica ed i ia addi i e manu ac u ing. The esul was a
well- i ed socke ha exhibi ed a uni o m s ess dis ibu ion and closely ma ched p essu e
senso da a on he pa ien , demons a ing how FEM can s eamline he i ing p ocess. This
app oach educes i e a i e physical adjus men s, po en ially cu ing down he nume ous clinic
isi s o en needed o weak socke i in he i s yea a e he ampu a ion (Plesec & Ha ih,
2024).
Ano he p omising a ea is he use o FEM o op imiza ion and pe sonaliza ion o o hoses.
Sophis ica ed wo k lows now combine FE simula ions wi h op imiza ion algo i hms o achie e
ailo ed de ice designs. Fo ins ance, in scoliosis ea men , Ka dash and he cowo ke s
(Ka dash e al., 2022) ha e de eloped pe sonalized simula ion models o a pa ien ’s o so and
an au oma ic design loop o op imize b ace geome y. By making he FE model di e en iable
(so ha he op imize can e icien ly see how design changes a ec clinical ou comes), ha
s udy managed o compu a ionally design b aces ha imp o ed spinal alignmen me ics by
an a e age o 45% in simula ion, while main aining pa ien com o wi hin accep able limi s.
Such pa ien -speci ic op imized b aces we e gene a ed o mul iple subjec s, showing he
easibili y o a semi-au oma ed design p ocess. While hese b aces s ill need o be physically
es ed on pa ien s (cu en ly he esul s a e based on simula ions), his indica es a u u e
whe e o ho ic de ices could be algo i hmically adap ed o each pa ien ’s ana omy and
condi ion.
FEM is also accele a ing inno a ion in de ice concep s. In p os he ic ee , o example,
designe s a e mo ing beyond s a ic design cha s o using FE models as i ual es beds.
Ene gy-s o ing p os he ic ee (which lex and e u n ene gy du ing walking) ha e bene i ed
om FEM-d i en design – allowing enginee s o y ou new geome ies o composi e layups
and immedia ely see he impac on de lec ion and ene gy e u n. Bala amak ishnan and his
cowo ke s (Bala amak ishnan e al., 2020) eplaced he adi ional ial-and-e o p o o yping
wi h a pa ame ic FE model o design a mul i-axial p os he ic oo . By adjus ing pa ame e s in
simula ion (ankle s i ness in a ious planes), hey achie ed an op imal balance ha was la e
buil and e i ied expe imen ally. Simila ly, in o hoses, we see 3D-p in ed w is splin s and
ankle b aces whose la ice s uc u es we e e ined ia FEA o maximize en ila ion and
minimize weigh , all while ensu ing he suppo unc ion is main ained (Plesec & Ha ih, 2024).
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 35
These case s udies unde sco e an impo an poin : FEM is enabling apid inno a ion by se ing
as a b idge be ween concep and clinic.
Finally, FEM is inc easingly used in clinical decision-making and educa ion. Clinicians can
isualize s ess dis ibu ions o po en ial p oblem a eas in a pa ien -speci ic model, which
helps in explaining o pa ien s why a ce ain o ho ic design is chosen ( o example, showing
whe e a scoliosis b ace will apply p essu e on he ibs). Some esea ch eams ha e e en
c ea ed eal- ime o nea - eal- ime FE analysis ools so ha a clinician could, in p inciple,
adjus design pa ame e s on he compu e and immedia ely see he e ec on i ness o
alignmen . While no ye con en ional, his in eg a ion o FEM in o he clinical wo k low is on
he ho izon as compu a ional ools become as e and mo e use - iendly.
5. Combining a i icial in elligence (AI) wi h FEM
FEM combined wi h AI has seen signi ican ad ancemen s in he las yea s, enabling as e
simula ions, sma e designs, and mo e pe sonalized o ho ic and p os he ic de ices. Below,
we analyze s a e-o - he-a de elopmen s ac oss key a eas, wi h e e ence o ecen academic
wo k.
a) Su oga e modeling wi h machine lea ning
Da a-d i en su oga es a e ained on FEM esul s o emula e de ailed simula ions a a
ac ion o he cos . Fo example, K iging eg ession models ha e been i ed o p os he ic
socke FE analyses, p o iding eal- ime p essu e and s ain p edic ions wi hin milliseconds
(S ee e al., 2020a).
b) AI in op imiza ion loops
AI echniques a e inc easingly embedded in FEM-d i en design op imiza ion. Su oga es
(ANNs, K iging, e c.) ac as apid e alua o s in i e a i e op imiza ion algo i hms (like gene ic
algo i hms o pa icle swa m op imiza ion) o ind op imal de ice designs. This has been
applied o opology and shape op imiza ion o p os he ic componen s. Fo example, a mul i-
objec i e gene ic algo i hm was combined wi h a su oga e FEM model o gene a e pe son-
speci ic p os he ic socke designs, spanning a spec um om pa ella - endon-bea ing o o al-
su ace-bea ing shapes (S ee e al., 2020b). The AI-assis ed op imize p oduced a sui e o
candida e socke geome ies ha s a egically edis ibu e p essu e o imp o e com o . Such
me hods d as ically educe manual ial-and-e o in design.
c) FEM applica ions enhanced by AI
• S ess analysis
In o ho ic design, homogeniza ion-based su oga es we e used o e alua e la ice
de lec ions much as e han explici FE, enabling quicke s ess- elie uning o la oo
insoles (Moeini e al., 2023).
• Fa igue li e and ailu e p edic ion
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 36
AI has imp o ed p edic ions o long- e m pe o mance such as a igue and s ess shielding
in p os he ic de ices. FEM is adi ionally used o e alua e a igue li e by simula ing epea ed
loads (e.g., gai cycles), bu his can be ime-consuming. Machine lea ning now helps es ima e
a igue pe o mance di ec ly om design pa ame e s and single- un FE esul s. Fo example,
in he design o semi-po ous hip p os heses, esea che s calcula ed sa e y ac o s and s esses
ia FEA o a ious po osi y con igu a ions, hen ained ML models o p edic hose ou pu s
o new designs ins an ly (Akkad e al., 2023). This app oach can iden i y designs ha minimize
s ess shielding in bone and ex end implan li e wi hou exhaus i e cyclic simula ions.
• Topology and shape op imiza ion
AI accele a es opology op imiza ion and design e inemen o cus om p os he ic
componen s. T adi ional opology op imiza ion i e a i ely calls FEM sol e s o ind an op imal
ma e ial layou , bu AI can guide o eplace some o hese i e a ions. Recen s udies use AI-
assis ed design o achie e ligh weigh ye s ong p os heses. One app oach combined pa ien
gai da a, mul i-body dynamics, FEM, and machine lea ning o op imize he shape o a hip
implan o each pa ien (Milone e al., 2024). The algo i hm adap s he p os hesis geome y
o he pa ien ’s weigh , heigh , and walking pa e n, imp o ing a igue s eng h and educing
he s ess peaks by up o 40% compa ed o he o iginal design. In gene al, neu al ne wo ks
ha e been ained o p edic op imal ma e ial dis ibu ions o o e ine designs a e a ew
ini ial FEM uns, d ama ically speeding up he design cycle. While AI does no ully eplace
physics-based opology op imiza ion, i se es as a powe ul helpe – sugges ing nea -op imal
layou s o p o iding an excellen s a ing poin o inal FEA ine- uning (Kulka ni e al., 2024).
• Ma e ial beha io and cons i u i e modeling
FEM accu acy depends on good ma e ial models o biological issues and de ice ma e ials.
AI is used o de elop mo e accu a e ma e ial beha io p edic ions. Fo example, AI has been
employed o p edic e ec i e p ope ies o new 3D-p in ed o ho ic ma e ials (e.g. la ice
in ills) wi hou exhaus i e physical es ing. These ad ances help enginee s e alua e no el
ma e ials and composi es, accele a ing he adop ion o high-pe o mance ma e ials in
p os heses design (Milone e al., 2024).
• Pa ien -speci ic modeling
A majo ad an age o AI-enhanced FEM is he abili y o handle pa ien a iabili y
e icien ly. S a is ical shape models and AI allow he c ea ion o pa ien -speci ic FE models on
he ly. Recen wo k combined p incipal componen analysis (PCA) o limb shapes wi h a
K iging su oga e o accoun o ana omical a iabili y in socke design (S ee e al., 2020a).
The esul was a su oga e ha could ake a new ampu ee’s esidual limb shape and socke
design pa ame e s as inpu and ins an ly p edic in e ace p essu es and issue s ains,
some hing ha would no mally equi e a leng hy FE se up o each pa ien .
• Clinical applica ions o AI-FEM in eg a ion
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 37
A mul i-objec i e AI op imize gene a ed a sui e o “bes i ” socke shapes o an ampu ee,
le ing he clinician choose an op imal s a ing design om da a-d i en op ions (S ee e al.,
2020b). This educes he i e a i e guesswo k in achie ing a com o able i . These AI-assis ed
design pla o ms ac as decision suppo ools, helping clinicians and enginee s explo e a la ge
design space quickly and base hei choices on quan i a i e p edic ions a he han ial-and-
e o . The esul is a as e design p ocess and de ices ha a e ailo ed o he pa ien ’s needs
om he ou se . One o he mos p omising clinical bene i s o AI-enhanced FEM is eal- ime
o nea - eal- ime simula ion, which opens he doo o in e ac i e i ing and uning o de ices.
Su oga e models can un nea ly ins an aneously, enabling eal- ime wha -i analyses du ing
pa ien consul a ions. Fo ins ance, S ee and his cowo ke s achie ed socke in e ace
p essu e p edic ions in 1.6 milliseconds using a su oga e, essen ially p o iding ins an
eedback on design adjus men s (S ee e al., 2020a). This capabili y means a p os he is could
i ually adjus an o hosis (s ap posi ion, s i ness, e c.) and immedia ely see he p edic ed
e ec on p essu e dis ibu ion o alignmen , e en du ing an appoin men . In he u u e, eal-
ime FEM p edic ions could be combined wi h mo ion cap u e o gai analysis: as a pa ien
mo es, an AI model could con inuously es ima e s esses o iden i y ins abili ies, allowing on-
he- ly uning o sma p os he ic componen s. Ea ly s eps owa d his ision a e seen in
mecha onic win amewo ks, whe e a physical p os hesis es e ( obo ic pla o m) and a
i ual FEM model un in pa allel (Chen e al., 2022). Such se ups ha e been used o ain deep
lea ning algo i hms o ecognize dynamic load pa e ns in socke s. Ul ima ely, his could lead
o closed-loop sys ems whe e senso da a om a pa ien ’s de ice is ed in o an AI model
(in o med by FEM), which hen sugges s eal- ime adjus men s o imp o e com o o
pe o mance. Beyond design op imiza ion, AI+FEM a e being le e aged o p edic clinical
ou comes and guide pa ien ca e. Machine lea ning algo i hms can mine simula ion esul s
and pa ien da a o iden i y pa e ns ha co ela e wi h success o complica ions. Fo
example, la ge da ase s o p os he ic use (e.g. senso eadings om sma p os heses) can be
c oss-analyzed wi h FEM-based s ess calcula ions o p edic issues like socke pain o isk o
skin b eakdown. One in e disciplina y e iew highligh ed he de elopmen o ML models ha
analyze p os he ic senso da a alongside o he pa ame e s o ind ends in pa ien ou comes
(Kulka ni e al., 2024). In p ac ice, his means AI migh de ec ha a ce ain pa e n o p essu e
dis ibu ion (p edic ed by FEM and con i med by senso s) o en p ecedes a skin so e, he eby
ale ing clinicians o in e ene ea lie . AI-d i en p edic i e models a e also being explo ed o
ehabili a ion: using simula ions o di e en alignmen o componen choices, algo i hms
could o ecas a pa ien ’s gai s abili y o ene gy cos and ecommend he op imal
con igu a ion o ha indi idual. This kind o pe sonalized ou come p edic ion ep esen s a
s ep owa d e idence-based, cus omized p os he ic ca e. As Kulka ni and his cowo ke s
emphasized, by iden i ying pa e ns in big da a om p os he ic usage, AI can enable uly
pe sonalized ea men plans ailo ed o each pa ien ’s li es yle and physiology (Kulka ni e
al., 2024). The in eg a ion o AI and FEM s ongly aligns wi h he goals o pe sonalized
medicine. E e y pa ien ’s ana omy and needs a e unique, and AI-enhanced FEM makes i

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 38
easible o accoun o hose di e ences in de ice design and uning. Pa ien -speci ic FE can
be gene a ed apidly wi h AI assis ance, as discussed. Clinically, his leads o cus om
o hoses/p os heses ha a e op imized o he indi idual a he han a one-size- i s-all
app oach. In he con ex o o hoses, o ins ance, an AI-FEM pipeline can ake a 3D scan o a
pa ien ’s oo and au oma ically ou pu a 3D-p in able o ho ic insole design ha edis ibu es
p essu e acco ding o ha pa ien ’s a ch geome y and gai load p o ile. Such wo k lows a e
beginning o appea in esea ch p o o ypes. The use o cloud compu ing u he helps in
pe sonaliza ion: la ge da abases o p io pa ien s’ FE models and ou comes can be s o ed, and
AI can le e age his collec i e knowledge o design a new pa ien ’s de ice wi h simila
success ul ea u es (an analogy o case-based easoning). O e all, AI is enabling a shi om
manual cus omiza ion o model-d i en pe sonaliza ion, whe e da a and simula ions ensu e
each de ice is he bes possible i o ha one pa ien .
• Valida ion echniques and hyb id modeling
A common alida ion s ep is compa ing AI su oga e p edic ions wi h adi ional FE esul s
o expe imen al measu emen s. Many s udies epo he e o ma gins o hei AI models o
es ablish c edibili y. Fo example, he su oga e ans ibial socke model de eloped by S ee
and his cowo ke s (S ee e al., 2020a) was alida ed agains ull FE solu ions, showing
p essu e e o s < 4 kPa and s ain e o s < 3%, which was deemed accep able o clinical use.
Ul ima ely, AI-enhanced FEM models mus be p o en in eal-wo ld scena ios. Resea che s a e
inc easingly designing p o ocols o compa e AI-FEM p edic ions wi h eal de ice pe o mance
on pa ien s. Fo example, an ins umen ed p os he ic socke (wi h embedded p essu e
senso s) can p o ide g ound u h da a on load dis ibu ion, which is hen compa ed o he
model’s p edic ions o he same pa ien walking. Ka amousadakis and his cowo ke s
(Ka amousadakis e al., 2021) de eloped a senso -based sys em o moni o ans emo al
socke p essu es and used an FE model as a e e ence o alida ion. Valida ion e o s also
ocus on he eliabili y and sa e y o AI in heal hca e con ex s. One ecognized conce n is ha
AI models a e only as good as hei aining da a, and may exhibi bias o e o s in scena ios
no well ep esen ed in ha da a (Kulka ni e al., 2024). Resea che s a e hus ca e ul o ain
on di e se da ase s (e.g., mul iple ana omies, load cases) and o quan i y unce ain y in
p edic ions. Some ecen FEM-AI s udies inco po a e Bayesian me hods o es ima e
con idence in e als o a p edic ed s ess o li e span, ale ing use s i a p edic ion is
ex apola ing beyond he known domain. Mo eo e , he communi y emphasizes ha AI
should assis , no o e ide, enginee ing judgmen – any c i ical design p edic ed by AI o en
unde goes a inal FE e i ica ion o a sa e y ac o bu e . By anspa en ly epo ing model
accu acy and unce ain ies, and by alida ing agains bo h simula ions and expe imen s,
enginee s aim o build us in hese AI-augmen ed ools. The goal is a alida ed hyb id
amewo k whe e AI p o ides speed and insigh , while FEM and physical es ing ensu e
accu acy and sa e y.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 39
FEM is now commonly used in o ho ic and p os he ic inno a ion. Compu a ional
modeling is now mo e powe ul and pa ien -speci ic, ma e ials om composi es o
me ama e ials a e being igo ously simula ed, and alida ion e o s gi e con idence in hese
i ual models. FEM is ac i ely d i ing clinical solu ions – om imp o ed socke i s o
op imized scoliosis b aces – and gains e en mo e capabili y h ough AI in eg a ion. These
ad ancemen s ul ima ely con e ge on a common goal: o c ea e p os heses and o hoses ha
a e ligh e , s onge , mo e com o able, and pe ec ly adap ed o each use , all achie ed
h ough he syne gy o simula ion and expe imen a ion.
Fini e elemen analysis o a he apeu ic o hosis manu ac u ed by 3D-p in ing
The p esen a ion below is ocused on e alua ing he s eng h cha ac e is ics o a
he apeu ic o hosis by simula ing a h ee-poin bending es (Łukaszewski e al., 2020). The
p inciple o he es is shown in Figu e 2.14. As one may no ice, a e being placed on wo
suppo blocks, he o hosis is loaded by a downwa d e ical o ce ac ing on he ed su ace
pa ch. This load g adually inc eases om 0 (ze o) o a maximum alue depending on he
ma e ial used o 3D-p in ing. The con ac be ween he o hosis and he suppo blocks akes
place along pe ec ly ma ching su aces.
Fig. 2.14. P inciple o he h ee-poin bending es simula ed o e alua ing he s eng h
cha ac e is ics o he he apeu ic o hosis
The ollowing assump ions ha e been made when p epa ing he ini e elemen model o
he h ee-poin bending es wi h he SOLIDWORKS Simula ion module o he SOLIDWORKS
compu e -aided design so wa e package:
a) The o hosis is made om PLA o PA12 exhibi ing an iso opic linea elas ic beha io . Table
2.4 lis s he physical and mechanical p ope ies o PLA and PA12 ha a e ele an o he
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 40
ini e elemen model o he h ee-poin bending es . These pa ame e s ha e been s o ed
in he cus om lib a y BIOMEDIX FEA Ma e ials.sldma p o ided o he s uden s oge he
wi h he 3D model shown in Figu e 2.14.
b) The suppo blocks a e pe ec ly igid bodies.
c) The o hosis is allowed o slide along i s con ac su aces wi h he suppo blocks. The
ic ional componen o his con ac in e ac ion is neglec ed.
Table 2.4. Physical and mechanical p ope ies o PLA (Fa ah e al., 2016) and PA12
Mass densi y
ρ [kg/m3]
Elas ic modulus
E [MPa]
Poisson’s a io
ν [-]
Yield s eng h
Y [MPa]
PLA
1252
3500
0.36
59
PA12
1010
1800
0.45
48
The ollowing s eps ha e been pe o med o p epa e he ini e elemen model o he
h ee-poin bending es :
a) De ining he suppo blocks as pe ec ly igid bodies
b) Associa ing he PLA/PA12 ma e ial o he o hosis by accessing he BIOMEDIX FEA
Ma e ials.sldma lib a y (see he example in Figu e 2.15)
c) Speci ying he con ac in e ac ion be ween he suppo blocks and he o hosis:
ic ionless sliding con ac (Fig. 2.16)
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 41
Fig. 2.15. Associa ing he PLA ma e ial o he o hosis by accessing he BIOMEDIX FEA
Ma e ials.sldma lib a y
Fig. 2.16. Speci ying he con ac in e ac ion be ween he suppo blocks and he o hosis
( ic ionless sliding con ac )
d) En o cing a ull locking kinema ic cons ain on he bo om aces o he suppo blocks (Fig.
2.17)
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 48
3 Tools and Techniques
3.1 CAD Sys ems wi h KBE Capabili ies
The design o ana omically cus omized biomedical p oduc s inc easingly equi es mo e
han basic 3D modelling ools. E icien , scalable p oduc ion o indi idualized designs - such as
o hoses, p os heses, implan s, o su gical models - demands sys ems capable no only o
pa ame ic modelling bu also o embedding design knowledge di ec ly in o he modelling
p ocess. These a e e e ed o as KBE-capable CAD sys ems, meaning hey suppo Knowledge-
Based Enginee ing unc ionali y h ough in elligen modelling, ule-based au oma ion, and
ex e nal da a in eg a ion.
KBE-capable CAD pla o ms combine adi ional solid and su ace modelling ools wi h
logic-d i en model gene a ion, sc ip ing in e aces, and pa ame iza ion amewo ks. This
enables he c ea ion o in elligen CAD models ha can adjus geome y based on pa ien -
speci ic da a o p e-se clinical ules, o ming he co e o au oma ed design sys ems. In his
sec ion, se e al commonly used KBE-capable CAD pla o ms a e b ie ly p esen ed and
compa ed wi h ega d o hei capabili ies, limi a ions, and sui abili y o implemen a ion in
biomedical design au oma ion wo k lows.
Se e al p o essional CAD sys ems p o ide ma u e en i onmen s o bo h mechanical
modelling and knowledge-based au oma ion. The mos p ominen in his space include
Au odesk In en o , Fusion 360, SolidWo ks, and CATIA V5. Each o hese sys ems suppo s
in elligen model de elopmen , bu hey di e in e ms o au oma ion capabili ies, modelling
pa adigms, accessibili y, and a ge use base. The ollowing analysis is based on ea lie wo k
(Gó ski 2025, Zawadzki 2018).
Au odesk In en o
In en o is a obus mechanical design so wa e widely used in p oduc de elopmen . I
o e s s ong suppo o pa ame ic and ea u e-based modelling, wi h in eg a ed modules
o pa and assembly c ea ion, simula ion, and documen a ion. I s p ima y ad an age in KBE
wo k lows is he iLogic sys em (Fig. 3.1), a sc ip ing en i onmen ha allows use s o embed
condi ional logic, pa ame e ela ionships, and ule-based beha io s wi hin CAD models. iLogic
also enables linking o ex e nal da a sou ces (e.g., Excel iles), allowing model egene a ion
based on pa ien -speci ic da a se s. Fo mo e complex au oma ion, Visual Basic o
Applica ions (VBA) can be used o sc ip ing ad anced ou ines. In en o suppo s hyb id solid-
su ace wo k lows, making i sui able o de eloping componen s ha need ana omical
adap a ion wi h s uc u ed ea u es such as langes o moun s.

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 49
Fig. 3.1. Au odesk In en o and iLogic sc ip ing sys em – modula p os hesis
Au odesk Fusion 360
Fusion 360 is a ligh weigh , cloud-based CAD/CAM pla o m designed o in eg a ed
p oduc de elopmen . I suppo s pa ame ic modelling, assembly design, FEA simula ion, and
gene a i e design. Al hough i s sc ip ing capabili ies ( h ough Py hon and Ja aSc ip ) a e mo e
limi ed han In en o ’s iLogic, i emains highly sui able o au oma ed wo k lows, especially
o smalle -scale applica ions. Fusion’s open a chi ec u e and na i e cloud connec i i y make
i a s ong candida e o dis ibu ed o emo e design wo k lows, and i s ee licensing op ions
o educa ional o s a up use s u he enhance accessibili y. I s s eng h lies in i s seamless
use expe ience and ull in eg a ion o design- o-manu ac u ing ools.
SolidWo ks
SolidWo ks, de eloped by Dassaul Sys èmes, is one o he mos widely used CAD sys ems
in enginee ing. I ea u es a comp ehensi e pa ame ic modelling en i onmen , powe ul
assembly ools, and a b oad ecosys em o plugins. Fo KBE applica ions, SolidWo ks o e s
buil -in au oma ion h ough equa ions, design ables, and mac os, as well as sc ip ing h ough
VBA. I suppo s con igu a ion managemen , enabling mul iple design a ian s o be managed
wi hin a single ile. While SolidWo ks does no o e a dedica ed KBE module compa able o
CATIA’s Knowledgewa e, i s en i onmen is ma u e enough o suppo in elligen modelling
when combined wi h sc ip ing and modula design logic.
CATIA V5
CATIA is a highly ad anced CAD/CAE sys em buil o complex enginee ing, pa icula ly in
au omo i e, ae ospace, and la ge-scale indus ial design. I o e s ull-spec um modelling
ools - om A-class su acing and ee o m design o mechanical simula ion and PLM
in eg a ion. I s Knowledgewa e module allows use s o c ea e pa ame ic empla es, embed
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 50
logic-based ules, build p oduc con igu a o s, and au oma e design asks using i s own
sc ip ing language. CATIA handles hyb id modelling qui e well and can wo k di ec ly wi h
impo ed meshes coming om 3D scans o medical imaging, making i ideal o ana omical
modelling a scale. Howe e , he so wa e is e y expensi e and no easy o lea n. Typically i
is used by la ge en e p ises and esea ch ins i u ions, a he han small heal hca e p o ide s,
SMEs o independen specialis s.
To acili a e selec ion o p ac ical implemen a ion, a compa a i e summa y is p esen ed
in he able 3.1, emphasizing KBE- ele an ea u es.
Tab. 3.1. Compa ison o mos popula KBE-capable CAD sys ems, based on (Gó ski 2025)
Fea u e / Sys em
In en o
SolidWo ks
CATIA V5
Fusion 360
Pa ame ic
modelling
Yes ( ully
suppo ed)
Yes ( ully
suppo ed)
Yes (ad anced,
scalable)
Yes
Su ace
modelling
Mode a e
(hyb id- iendly)
Mode a e
Ad anced (GSD
module)
Mode a e
Mesh handling
Limi ed
Limi ed
Good (wi h DSE
module)
Basic
Rule-based
modelling
iLogic, VBA
sc ip ing
Equa ions,
mac os
Knowledgewa e
(na i e)
Basic sc ip s
(Py hon)
Ex e nal da a
linking
Excel, OLE, APIs
Excel, mac os
Ex e nal ables,
XML, APIs
CSV, API
Simula ion
in eg a ion
Basic FEA
Robus FEA sui e
Ad anced
mul iphysics
In eg a ed
FEA/CFD
Usabili y
Medium
(enginee ing
ocus)
High (in ui i e)
Complex (s eep
lea ning)
High (beginne -
iendly)
Licensing cos
High
High
Ve y high
Low/ ee ( o
s a ups)
Bes sui ed o
SMEs,
enginee ing
eams
Gene al design
i ms
La ge en e p ises,
R&D
S a ups,
educa ional use
Based on he con ex de eloped h oughou his chap e , whe e ana omically cus omized
p oduc s a e gene a ed using in elligen CAD models and embedded design logic, a ew
obse a ions can be made ega ding sys em selec ion:
1. Fo la ge-scale implemen a ions in hospi als, esea ch labs, o medical de ice
manu ac u e s dealing wi h high a iabili y and complexi y, CATIA V5 emains
unma ched in capabili y. I s Knowledgewa e module suppo s comple e KBE
wo k lows, while i s su acing ools handle o ganic geome y wi h p ecision. Howe e ,
i s complexi y and licensing cos es ic i s adop ion o ins i u ions wi h signi ican
echnical and inancial esou ces.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 51
2. Fo enginee ing- ocused en i onmen s ha equi e solid au oma ion unc ionali y and
a e al eady amilia wi h Au odesk ecosys ems, In en o o e s a good balance o
pe o mance and accessibili y. I s iLogic sc ip ing allows o modula and con igu able
models, well-sui ed o gene a ing p os he ic componen s, su gical guides, and
mechanical in e aces o implan s. I also in eg a es well wi h ex e nal sp eadshee s
and o ms, making i p ac ical o clinical design au oma ion pipelines.
3. Fo smalle eams, s a ups, o educa ional en i onmen s, Fusion 360 p esen s a
compelling op ion. I s uni ied pla o m co e s design, simula ion, and manu ac u ing,
wi h cloud collabo a ion ea u es ideal o dis ibu ed wo k. While i s sc ip ing and
au oma ion capabili ies a e limi ed compa ed o In en o o SolidWo ks, i o e s
su icien lexibili y o simple indi idualized p oduc s such as passi e o hoses o
su gical empla es. Addi ionally, i suppo s 3D p in ing wo k lows na i ely, educing
he ic ion be ween design and ab ica ion.
4. Fo gene al-pu pose design wi h s ong simula ion and documen a ion ea u es,
SolidWo ks is a iable al e na i e o In en o , pa icula ly i simula ion accu acy o
design e i ica ion a e c i ical. I may be p e e ed by eams al eady embedded in he
Dassaul Sys èmes ecosys em o amilia wi h SolidWo ks’ in e ace and plugin
ecosys em.
5. In en i onmen s whe e non-pa ame ic mesh edi ing is essen ial (e.g. maxillo acial
econs uc ion, umo modelling), none o he abo e sys ems is su icien alone. A
hyb id wo k low using specialized mesh p ocessing ools (e.g. Blende , Meshmixe )
combined wi h one o he pa ame ic CAD sys ems is o en he bes app oach. CATIA
has some mesh capabili ies (Digi ized Shape Edi o ), bu o smalle ope a ions, a
Fusion-Blende o In en o -Meshmixe pipeline is o en mo e ealis ic.
The choice o CAD sys em o KBE applica ions in biomedical design should be guided by:
• echnical equi emen s – le el o model complexi y, equency o case a ia ion, need
o simula ion o analysis,
• scope o au oma ion - whe he ull au oma ion is needed o pa ial is enough,
• scalabili y - olume o cases pe week/mon h and le els o a iabili y,
• use p o ile – clinicians and o he medical (non- ech) pe sonnel biomedical enginee s,
CAD echnicians, o hyb id oles.
• economic cons ain s - licensing cos s and models (pe manen , subsc ip ion, okens),
main enance, aining a ailabili y.
In mos academic o SME con ex s, he combina ion o Fusion 360 ( o gene al modelling
and au oma ion) and In en o ( o in elligen model de elopmen and in eg a ion in o design
sys ems) p o ides an op imal balance be ween unc ionali y and cos . Fo la ge -scale sys ems
whe e unding and expe ise a e a ailable, CATIA V5 (o Siemens NX) emains he gold
s anda d o implemen ing deeply in eg a ed KBE sys ems capable o handling ana omical
a iabili y a indus ial scale.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 52
3.2 Design Au oma ion Case S udies
Au oMedP in : A Sys em o O hopedic De ice Design
The Au oMedP in sys em, desc ibed in nume ous p e ious publica ions (Gó ski 2025,
Gó ski e al. 2024a, Gó ski e al. 2020) ep esen s a comple e and modula implemen a ion o
an au oma ed design and manu ac u ing amewo k o ana omically cus omized o hopedic
and p os he ic de ices. I was de eloped as a demons a o o a ully unc ional digi al
wo k low ha in eg a es 3D scanning, da a p ocessing, ule-based CAD model gene a ion, and
addi i e manu ac u ing p epa a ion - all wi hou equi ing ac i e pa icipa ion o an enginee
du ing he ou ine ope a ion. The sys em is cen e ed on he p inciple ha complex design
logic, expe decision pa hways, and model a iabili y can be cap u ed and embedded wi hin
in elligen CAD s uc u es, accessible h ough an in ui i e use in e ace and ope a ed by
medical o echnical s a . The desc ip ion in his chap e is based on he a o emen ioned
ea lie published wo ks, mos ly (Gó ski 2025, Gó ski e al. 2024a).
A he co e o Au oMedP in lies a se o in elligen , ule-d i en CAD models de eloped in
Au odesk In en o using he iLogic en i onmen . These models ep esen modula p oduc s
such as cosme ic and mechanical uppe limb p os heses, as well as w is -hand o hoses and
ankle- oo o hoses (AFOs). Each model is designed o egene a e dynamically based on
an h opome ic da a ex ac ed om 3D scans. This enables he apid c ea ion o indi idualized
de ice geome ies ha con o m o clinical and unc ional needs wi hou edesign om
sc a ch. The main ideas a e p esen ed in Fig. 3.2.
The p ocess begins wi h non-con ac an h opome ic measu emen , using a s uc u ed-
ligh 3D scanne moun ed on a mobile o ixed ig. A suppo ing s uc u e s abilizes he
pa ien ’s limb, while a so wa e module (Au oMedP in Ope a o Panel) coo dina es scan
acquisi ion, limb posi ioning, and da a cap u e. The scanned mesh is hen p ocessed h ough
a chain o mac o-enabled so wa e ools (no ably MeshLab), which pe o m cleanup,
o ien a ion co ec ion, and c oss-sec ional analysis. Poin cloud da a, diame e s, and o he
measu emen s a e ex ac ed and s o ed in a s uc u ed o ma o u he p ocessing.
A isual applica ion called he Limb Calib a o - de eloped in Uni y 3D - allows he use o
manually de ine ana omical e e ence planes (e.g., w is , end o limb), adjus ing c oss-sec ion
posi ioning as equi ed. This ensu es consis ency in measu emen in e p e a ion and model
alignmen . The measu ed alues a e w i en o ex e nal iles ( ypically Excel sp eadshee s),
which se e as inpu o he au oma ed design able. This able ans o ms he aw ana omical
da a in o a comple e con igu a ion da ase sui able o egene a ing he in elligen CAD
model.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 53
Fig. 3.2. Wo k o Au oMedP in sys em based on (Gó ski e al. 2024)
The ac ual model egene a ion is launched ia Excel mac os and ba ch sc ip s. The sys em
eads he con igu a ion ile, opens he pa ame ic CAD model, and applies all necessa y
adjus men s - including geome y scaling, ea u e ac i a ion/supp ession, and module
selec ion. All o his occu s wi hou use in e ac ion wi h he CAD in e ace i sel . Upon
comple ion, he CAD sys em expo s a manu ac u ing- eady ile (e.g., STL) o he ab ica ion
wo ks a ion.
The inal s age o he pipeline in ol es he apid manu ac u ing module, whe e he digi al
model is sliced using p ede ined se ings and sen o an ex usion-based 3D p in e . The
p in ing p ocess is semi-au oma ed, wi h pos -p ocessing and assembly pe o med by ained
echnical s a . The o e all p ocess – om scanning o p oduc ion- eady design – can be
comple ed in unde 15 minu es in ypical cases, making Au oMedP in a iable solu ion o
same-day de ice ab ica ion unde ce ain condi ions.
The sys em's modula a chi ec u e is a key enable o i s adap abili y. Fou majo
unc ional modules a e iden i ied: 1) 3D scanning and measu emen ; 2) au oma ic design; 3)
in e ac i e use in e ace; and 4) addi i e manu ac u ing. Each module can be deployed
indi idually o in combina ion depending on he ope a ional con ex . Fo example, in a clinic
lacking in-house ab ica ion capabili ies, design da a may be sen o a cen alized p oduc ion
si e using a dis ibu ed manu ac u ing model. Al e na i ely, scanning and con igu a ion may
be comple ed locally while design gene a ion is pe o med emo ely on a cloud-hos ed CAD
se e - allowing ins i u ions o sha e access o high-cos licenses.
In e ms o so wa e, Au oMedP in elies on a combina ion o comme cial and open
sou ce ools (e.g., Au odesk In en o , Excel, MeshLab, slice so wa e) and p op ie a y
in e aces buil using Uni y and Visual S udio. The in eg a ion is achie ed h ough ex e nal
sc ip ing, mac o au oma ion, and ba ch execu ion, which oge he c ea e a seamless pipeline.

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 54
F om a usabili y s andpoin , he sys em minimizes ope a o bu den. Pa ien s can in e ac
wi h he sys em ia a ouch in e ace o VR headse o cus omize aes he ic ea u es o he
p oduc , while echnical and medical s a assis wi h scanning and pos -p ocessing. No
specialized CAD expe ise is equi ed du ing he ope a ional phase; all design logic is
embedded in he sys em and upda ed as pa o a con inuous imp o emen p ocess based on
use eedback, clinical ou comes, and es -case e alua ions.
Au oMedP in p o ides a comp ehensi e case s udy in he de elopmen o a eal-wo ld,
ully unc ional au oma ed design sys em ailo ed o biomedical applica ions. I exempli ies
how s uc u ed knowledge ep esen a ion, modula so wa e de elopmen , and in elligen
modelling can be combined o deli e a scalable solu ion o indi idualized pa ien ca e.
Au oma ed Design o O hoses
The au oma ed design o o hoses using he Au oMedP in sys em s eamlines he
c ea ion o pe sonalized medical de ices by in eg a ing 3D scanning, digi al modeling, and
addi i e manu ac u ing. This enhances p ecision, educes cos s, and ensu es a be e i o
pa ien s. To ealize he design o o hoses in an au oma ed manne , he modula s uc u e o
Au oMedP in is u ilized, consis ing o (Gó ski e al. 2024a, Gó ski e al. 2024c):
• 3D scanning & design s a ion – ope a o -con olled scanne and so wa e o da a
acquisi ion,
• use in e ace s a ion – a ouchsc een, VR headse , and in e ac i e so wa e o
cus omiza ion,
• apid manu ac u ing d a ion – 3D p in ing and pos -p ocessing ools o p oduc ion.
A unc ional p o o ype was i e a i ely imp o ed, wi h u he modi ica ions needed o
mass p oduc ion.
A basic, ounda ional p oduc con ained in he Au oMedP in sys em is w is hand
o hosis, de eloped since he ini ial e sions o he sys em (Gó ski e al. 2020). The o hosis is
cus omized on he basis o a non-con ac measu emen o geome y o pa ien ’s hand and
o ea m (o mi o image o he o he limb, when he ac ual limb is damaged and e.g. w apped
in plas e cas ). The measu emen is done by op ical 3D scanning, usually a he wo kplace
de eloped as a pa o he Au oMedP in sys em, de eloped a Poznan Uni e si y o
Technology. A e measu emen , da a is p ocessed om aw scans o econs uc ed, smoo h
limb model (Fig. 3.3). Ou o his model, se s o poin s a e ex ac ed o eed he in elligen CAD
model.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 55
Figu e 3.3. Da a p ocessing o 3D scans o he w is hand o hosis model (Gó ski 2025,
Gó ski e al. 2020)
The p oduc was o iginally designed in he Au odesk In en o CAD sys em (Figu e 3.4), as
an in elligen model – i s design can be changed eely by supplying i wi h a ious da a om
3D scanning, leading o au oma ed e-design.
Figu e 3.4 Design o o hosis in Au odesk In en o (Au oMedP in sys em ma e ials)
The o hosis consis s o basic pa s (Fig. 3.5):
- bo om pa (in con ac wi h palm),
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 56
- op pa (in con ac wi h back o he hand),
- op ionally – he bo om and op pa could be ans e sally di ided i o hosis
Figu e 3.5. W is hand o hosis – pa s (Gó ski e al. 2024a)
The o hosis design able con ains poin s ex ac ed om he limb econs uc ion. I is an
Excel sp eadshee . The poin ex ac ion is ealized by Au oMedP in sys em algo i hms, using
Excel and MeshLab so wa e. In eg al elemen is he a o emen ioned Limb Calib a o
applica ion (Fig. 3.2). The p ocess o da a ex ac ion is au oma ic. As a esul , new Excel
sp eadsheed is gene a ed. Then he use (e.g. a s uden ) opens he In en o model and
upda es he design able. A e upda ing, check o e o s and possible imp o emen s, he
model mus be sa ed o ex e nal ile o u he use. I is usually done in wo ways:
- whole o hosis is sa ed in STP ile, o he da abase
- indi idual pa s (solid bodies) a e sa ed as OBJ o STL iles o 3D p in ing.
Case s udy example
As an example o a pa ien , a case o a 26-yea old man was selec ed, wi h an inju y o his
igh w is , caused by bi e o a dog esul ing in bone c ush. A ull p ocess was unde gone and
eco ded o him (3D scanning shown in Figu e 3.6, inished wi h ob aining a comple e
unc ional o hosis (Figu e 3.7). The scanning p ocess in ol ed cap u ing bo h limbs
sepa a ely. The le (heal hy) a m was scanned in a co ec ana omical posi ion using a Da id
SLS-3 scanne . The igh (a ec ed) a m was manually scanned wi h an EinScan P o 3D scanne ,
posi ioned o pa ien com o .
bo om pa
op pa
openwo k
basic assembly ea u es
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 57
Figu e 3.6 Pa ien wi h inju ed hand 3D scanning using Au oMedP in sys em a PUT
Using he Au oMedP in sys em, au oma ed mac os ex ac ed key da a, inco po a ing 3
mm o se o lining and 4 mm s anda d hickness o du abili y. A CAD-based w is -hand
o hosis model was c ea ed on he basis o ex ac ed pa ame e s. The design was digi ally
aligned wi h he scan o ensu e a p ecise i and a oid s uc u al con lic s.
Figu e 3.7. Au oma ically designed model o he o hosis
The whole design p ocess ook app oxima ely 30 minu es a e he da a was ex ac ed
om he o hosis. The longes ime is ac ually he au oma ic egene a ion o he model by
Au odesk In en o . Also, o ensu e lack o ex a i e a ions, he o hosis was manually es ed
agains he scan o he a ec ed hand o collisions, using MeshLab so wa e, which ook
addi ional 15 minu es be o e accep ing he design and pu ing i in o p oduc ion.
Au oma ed Design o P os heses
The au oma ic p os hesis design p ocedu e desc ibed in he ollowing sec ion pe ains o
a solu ion de eloped wi hin he Au oMedP in p ojec . I in ol es a modula mechanical
p os hesis speci ically designed o cycling, desc ibed in ea lie wo k (Gó ski e al. 2024b,
Gó ski 2025).
Pe sonaliza ion o hand p os heses is a pa icula ly c ucial aspec o o hopedic p oduc
manu ac u ing. This is due o he imp o ed i o he p os hesis o he pa ien ’s s ump,
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 64
Ano he c i ical componen o an au oma ed wo k low is he o ches a ion engine ha
links all he s ages oge he (Ben-Nun e al. 2020). This laye manages he low o da a om
CAD con e sion h ough slicing and in o he gene a ion o machine ins uc ions. I moni o s
inpu di ec o ies, igge s p ocessing sc ip s, manages logging and e o epo ing, and
in e aces wi h ex e nal sys ems such as p oduc li ecycle managemen (PLM) pla o ms. In
many cases, he o ches a ion logic is implemen ed using gene al-pu pose p og amming
en i onmen s such as Py hon o Node.js, o en coupled wi h wo k low au oma ion ools ha
allow o modula and scalable in eg a ion. This o ches a ion is essen ial o ensu e ha all
p ocessing s eps occu in he co ec sequence, ha e o s a e caugh ea ly, and ha all
ou pu s a e p ope ly s o ed, acked, and made a ailable o downs eam sys ems.
The inal ou pu o he wo k low is he nume ical con ol code ha ins uc s he addi i e
manu ac u ing machine how o build he pa . This code, which may ake he o m o G-code,
X3G, o o he p op ie a y o ma s depending on he machine, mus be ansmi ed o he
p in e in a eliable and aceable manne . Au oma ion a his s age ensu es ha he co ec
ile is associa ed wi h he co ec job, educing he isk o ope a o e o . In p oduc ion
en i onmen s, his o en in ol es di ec ne wo k ans e o machine, some imes h ough
secu e APIs o cloud-based in e aces (Baumann e al. 2017). Al e na i ely, sys ems may
p epa e USB d i es o o he physical media wi h he co ec ile s uc u e and naming
con en ions, eady o manual ans e whe e equi ed. In bo h cases, he goal is o minimize
o elimina e manual handling o digi al iles, he eby p ese ing he in eg i y o he p ocess.
Se e al indus ial sec o s ha e al eady adop ed end- o-end au oma ion in hei addi i e
manu ac u ing wo k lows. In he medical de ice indus y, o example, he p oduc ion o
pa ien -speci ic o hoses, implan s, and su gical guides has been e olu ionized by au oma ion
sys ems ha accep ana omical da a, p ocess i in o p in able geome y, and gene a e NC
code wi hou human in e en ion (Gó ski e al. 2022). These sys ems o en ope a e in
compliance wi h s ingen egula o y s anda ds and include obus alida ion s eps o ensu e
clinical sa e y. In ae ospace and au omo i e manu ac u ing, au oma ed wo k lows a e
employed o accele a e he p oduc ion o ix u es, jigs, and p o o ype componen s,
in eg a ing di ec ly wi h en e p ise da a sys ems o e ie e pa speci ica ions and
manu ac u ing equi emen s. In consume - acing applica ions, online cus omiza ion pla o ms
allow cus ome s o pe sonalize p oduc s ia web in e aces, wi h all subsequen p ocessing
s eps — om geome y egene a ion o slicing and job queuing — handled au oma ically in
he cloud.
Despi e hese successes, implemen ing au oma ion in addi i e manu ac u ing wo k lows
is no wi hou challenges. One o he p ima y obs acles is he lack o s anda diza ion ac oss
so wa e ools, machine in e aces, and da a o ma s (Xiao e al. 2018). While some ini ia i es,
such as he 3MF Conso ium, aim o es ablish uni ied o ma s ha p ese e bo h geome y

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 65
and me ada a, widesp ead adop ion emains limi ed. As a esul , many au oma ion pipelines
equi e cus omized adap e s o ansla o s o in e ace wi h speci ic slice s, CAD ools, o
p in e s. Ano he challenge is ensu ing obus ness in he ace o inpu a iabili y. Unlike
adi ional manu ac u ing, whe e inpu s a e highly s anda dized, addi i e wo k lows o en
deal wi h unique geome ies and cus ome -de ined pa ame e s. Au oma ed sys ems mus
he e o e include comp ehensi e e o handling, inpu alida ion, and excep ion managemen
ou ines o p e en unexpec ed ailu es. Addi ionally, managing complexi y in pa ame e
selec ion and manu ac u ing s a egy de ini ion can be di icul , pa icula ly when a emp ing
o scale au oma ion ac oss di e se p oduc amilies o ma e ials.
Success ul deploymen o au oma ion in addi i e manu ac u ing also depends on
in eg a ion wi h ups eam and downs eam p ocesses. A ully digi al h ead equi es ha
design inpu s, p oduc ion pa ame e s, quali y con ol da a, and logis ics in o ma ion low
seamlessly be ween sys ems. Au oma ion wo k lows mus be capable o in e acing wi h ERP
o PLM pla o ms o e ie e o upda e ele an in o ma ion, schedule jobs, and ack s a us.
Modula a chi ec u e is o en p e e ed, allowing indi idual componen s o he au oma ion
s ack o be upda ed, eplaced, o scaled independen ly. This no only acili a es sys em
main enance bu also suppo s inno a ion and cus omiza ion wi hou isking he s abili y o
he en i e pipeline. Howe e , a signi ican challenge in main aining such sys ems a ises om
he apid pace o de elopmen in addi i e manu ac u ing echnologies. New machines a e
equen ly in oduced, o en implemen ing di e en p o ocols o s anda ds o da a
ansmission, machine con ol, and manu ac u ing job managemen . As a esul , a sys em
once ully unc ional may equi e epea ed adap a ion o emain compa ible wi h e ol ing
ha dwa e ecosys ems. Fu he mo e, unce ain y ega ding he long- e m a ailabili y o
suppo o exis ing equipmen aises conce ns abou whe he wo n o obsole e machines can
be eplaced wi h equi alen s ha will unc ion eliably wi hin he es ablished au oma ion
amewo k. These ac o s in oduce a laye o complexi y ha demands o wa d-looking
planning, lexibili y in sys em design, and he capaci y o apidly in eg a e new s anda ds
wi hou dis up ing he con inui y o au oma ed wo k lows.
3.4 Scaling Up: Mass Cus omiza ion o Biomedical P oduc s
The inc easing demand o pe sonalized ca e, alongside ad ances in digi al echnologies,
has c ea ed e ile g ound o he de elopmen o scalable, cus omizable medical solu ions.
Among he mos ans o ma i e app oaches eme ging in his con ex is mass cus omiza ion, a
manu ac u ing s a egy ha me ges he e iciency o mass p oduc ion wi h he speci ici y o
indi idual ailo ing. While ini ially applied in consume sec o s like au omo i e and elec onics,
mass cus omiza ion is now making signi ican in oads in o he biomedical domain, pa icula ly
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 66
in he p oduc ion o ana omically cus omized p oduc s such as o hoses, p os heses, and
implan s (Gó ski 2025).
Mass cus omiza ion (MC) challenges he bina y o c a -based indi idualiza ion and high-
olume s anda diza ion. I allows o he c ea ion o highly indi idualized p oduc s wi h lead
imes and uni cos s app oaching hose o mass-p oduced goods. This is pa icula ly ele an
in biomedical applica ions, whe e ana omical a iabili y ac oss pa ien s makes cus omiza ion
no jus desi able bu clinically necessa y.
Wi hin he spec um o p oduc ion me hods— anging om piece p oduc ion o small-
se ies and mass p oduc ion—mass cus omiza ion occupies a unique posi ion. I le e ages he
scalabili y and logis ics o mass p oduc ion while enabling he gene a ion o case-speci ic
a ian s, made possible by au oma ion, digi al manu ac u ing, and in eg a ed design
wo k lows. Fo biomedical p oduc s, his means p oducing ailo ed solu ions a scale, wi h
consis ency and epea abili y, while e aining he clinical speci ici y equi ed o indi idual
ea men . The analysis below is based on p e ious wo k (Gó ski 2025, Gó ski e al. 2024a,
Gó ski e al. 2025, Zawadzki 2018).
The easibili y o mass cus omiza ion in medicine has been signi ican ly ad anced by he
con e gence o se e al key echnologies:
• 3D Scanning and imaging- mode n s uc u ed ligh and lase scanning sys ems enable
accu a e and as acquisi ion o ex e nal body geome ies. Fo in e nal s uc u es, CT
and MRI da a p o ide de ailed 3D models;
• Compu e -Aided Design (CAD) hey allow he ans o ma ion o ana omical da a in o
indi idualized p oduc geome ies. In elligen CAD models wi h embedded design logic
enable consis en adap a ion o a ying pa ien ana omies;
• Addi i e Manu ac u ing and 3D p in ing echnologies, especially hose adap ed o
medical-g ade ma e ials (e.g., biocompa ible polyme s, i anium alloys), allow on-
demand ab ica ion o p oduc s wi h complex geome ies, op imized o each pa ien ;
• Au oma ed Design Sys ems (ADS) – as demons a ed in he Au oMedP in amewo k,
ADS pla o ms in eg a e pa ien da a acquisi ion, ule-based CAD model egene a ion,
and manu ac u ing p epa a ion. This educes he need o manual CAD in e en ion
and sho ens he design cycle d ama ically;
• Knowledge-Based Enginee ing (KBE) – by o malizing expe knowledge and decision
logic, KBE helps au oma ing epe i i e design asks, managing con igu a ion
complexi y, and educing human e o . In biomedical con ex s, his ansla es in o
eliable, high-speed gene a ion o ana omically adap ed de ice models.
These enable s collec i ely suppo he ans o ma ion om adi ional one-o p oduc ion
o a scalable, semi-au oma ed cus omiza ion pipeline.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 67
Mass cus omiza ion is pa icula ly well-sui ed o a ange o biomedical p oduc s whe e
ana omical i and pa ien -speci ic unc ion a e essen ial:
• o hoses and p os he ic socke s: hese p oduc s a e inhe en ly pa ien -speci ic due o
a iabili y in esidual limb shape, mo emen ange, and pa ien condi ion. Mass
cus omiza ion enables digi al wo k lows whe e scanning, model adap a ion, and
ab ica ion occu in apid succession, p oducing clinically e ec i e and com o able
de ices.
• cus omized implan s: c anial pla es, spinal cages, and join eplacemen s bene i om
indi idualized design o imp o e in eg a ion and educe complica ion a es.
Au oma ed design wo k lows educe ime- o-su ge y and enable hospi als o
manu ac u e s o o e cus omized implan s as a se ice.
• su gical guides and p eope a i e models: cus om models used o planning o guiding
su ge y can be gene a ed om imaging da a, o e ing g ea e p ecision and educing
ope a i e ime. Wi h au oma ed design p ocesses, such ools can be p oduced apidly
and a a sus ainable cos .
• wea able ehabili a ion de ices: de ices such as w is -hand o hoses o dynamic
splin s can be cus omized o pa ien ana omy and he apy goals. Mass cus omiza ion
allows hese o be ab ica ed in small clinics using cloud-based ools and desk op
addi i e manu ac u ing.
T adi ionally, many cus omized medical de ices, especially o ho ic and p os he ic
p oduc s, ha e been manu ac u ed using manual echniques. These include hand-moulding
he moplas ics o lamina ing a ound plas e models de i ed om nega i e moulds. While
unc ional, hese p ocesses a e labo -in ensi e, di icul o scale, and highly dependen on
echnician skill.
The digi al shi in oduces epea abili y, documen a ion, and oppo uni ies o
au oma ion. Mode n wo k lows, such as hose implemen ed in Au oMedP in , begin wi h 3D
scanning, ollowed by digi al measu emen ex ac ion, in elligen CAD model gene a ion, and
addi i e manu ac u ing. Each s age eplaces a manual s ep, b inging aceabili y and educing
a iabili y.
S ill, scalabili y demands mo e han digi al ools, as i equi es au oma ion o decision-
making and in eg a ion ac oss sys ems. Only h ough KBE and obus design au oma ion can
la ge olumes o cus omized designs be handled e ec i ely wi hou inc easing he design
eam’s wo kload p opo ionally.
Despi e i s ad an ages, implemen ing mass cus omiza ion in medical de ice p oduc ion is
no wi hou challenges:
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 68
• c ea ing a sys em capable o gene a ing eliable, indi idualized designs equi es
ex ensi e domain knowledge and collabo a i e inpu om clinicians, enginee s, and
egula o y expe s.
• each pa ien p esen s a unique case, which in oduces high a iabili y in design inpu s.
Au oma ed sys ems mus be obus agains e o s, incomple e da a, o bo de line
ana omical cases.
• biomedical de ices mus mee s ingen sa e y and documen a ion s anda ds (mo e
on ha in chap e 7 o his e-book). Cus om p oduc s s ill equi e aceabili y,
consis en pe o mance, and app op ia e ce i ica ion—e en when made on-demand.
• ansi ioning o mass cus omiza ion o en ede ines p o essional oles. O ho ic
echnicians, o ins ance, mus adap om manual ab ica ion o ope a ing digi al
scanne s and e i ying CAD-gene a ed esul s.
• while addi i e manu ac u ing and au oma ion educe labo , ini ial se up cos s,
ha dwa e, so wa e, and de elopmen can be signi ican . Howe e , in many cases,
long- e m ROI jus i ies he in es men , especially o mid- o high- olume
cus omiza ion.
The ajec o y o mode n heal hca e inc easingly a o s pe sonalized medicine, whe e
ea men s, in e en ions, and de ices a e ailo ed o indi idual pa ien s. Mass cus omiza ion
o e s he echnological backbone o deli e on his p omise e icien ly. Wi h ma u ing digi al
in as uc u e, g ea e in eg a ion o AI, and enhanced imaging echniques, we can expec
inc easing up ake o hese sys ems ac oss hospi al ne wo ks and medical de ice
manu ac u e s.
C ucially, mass cus omiza ion mus emain ocused on alue c ea ion, imp o ing pa ien
ou comes, educing lead imes, and enabling clinicians o access be e ools wi h minimal
complexi y. In elligen au oma ion, especially when g ounded in clinically alida ed design
logic, allows heal hca e o o e cus omiza ion as s anda d p ac ice a he han an excep ion.
In conclusion, mass cus omiza ion ep esen s a scalable pa hway o deli e ing pa ien -
speci ic biomedical p oduc s in a con olled, cos -e ec i e, and epea able manne . As
enabling echnologies con inue o e ol e, and he gap be ween manual and digi al p ocesses
na ows, biomedical enginee ing s ands a he h eshold o a ans o ma ion whe e
pe sonaliza ion and scalabili y can coexis , b inging p ecision medicine in o he ealm o
ou ine ca e.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 69
4 Conclusions
4.1 Challenges and Fu u e T ends
The adop ion o Knowledge-Based Enginee ing (KBE) me hodologies in biomedical p oduc
design o e s emendous po en ial, ye i s implemen a ion in clinical and enginee ing se ings
emains complex. Se e al echnical, o ganiza ional, and egula o y ba ie s mus be add essed
o KBE sys ems o each hei ull ans o ma i e capaci y. A he same ime, apid
echnological de elopmen s, pa icula ly in a i icial in elligence, cloud compu ing, and
collabo a i e pla o ms, a e shaping he nex gene a ion o KBE-enabled biomedical design.
This sec ion ou lines key challenges and p esen s eme ging ends ha will de ine he
e olu ion o au oma ed, ana omically awa e design wo k lows.
One o he mos c i ical challenges in deploying KBE in biomedical con ex s lies in he
acquisi ion and o maliza ion o expe knowledge. Much o he ele an expe ise – whe he
clinical, biomechanical, o manu ac u ing- ela ed - emains aci , esiding in he ou ines and
in ui ion o expe ienced p o essionals. T ans o ming his implici knowledge in o o malized
design logic equi es s uc u ed in e iews, e e se enginee ing, long- e m obse a ion, and
in e disciplina y collabo a ion. Mo eo e , he knowledge mus be con ex ualized and
alida ed o ensu e i s accu acy and ele ance ac oss di e se cases.
E ec i e ep esen a ion o his knowledge is equally impo an . Decla a i e ules,
pa ame ic empla es, on ological maps, and decision ees mus be implemen ed wi hin
sys ems ha emain in e p e able, modi iable, and scalable. In medical applica ions, he isk
o embedding ou da ed o o e ly igid logic poses a sa e y conce n. The e o e, KBE
implemen a ions mus be coupled wi h con inuous eedback loops and obus alida ion
mechanisms, pa icula ly when used in high- isk scena ios such as implan design o su gical
planning.
Ano he challenge is in e ope abili y. While many CAD pla o ms now suppo pa ame ic
modelling and sc ip ing capabili ies, in eg a ion wi h addi i e manu ac u ing (AM) wo k lows,
pa ien da a sys ems, and simula ion ools is a om seamless. Di e ences in ile o ma s,
model ideli y, me ada a s anda ds, and au oma ion in e aces hinde he c ea ion o ully
au oma ed design pipelines. Cus om in e aces o middlewa e solu ions a e o en equi ed o
b idge gaps be ween da a sou ces (e.g., DICOM o mesh iles), CAD en i onmen s (e.g.,
SolidWo ks, Fusion 360, NX), and AM p epa a ion ools (e.g., slice s, pos -p ocessing sc ip s).
Wi hou s anda diza ion, hese b idges a e agile and di icul o scale.
Addi ionally, egula o y amewo ks o medical de ices emain o ien ed owa d s a ic,
epea able p oduc s, no dynamically gene a ed, pa ien -speci ic solu ions. E en hough KBE

Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 70
sys ems aim o inc ease consis ency and aceabili y, he indi idualized na u e o hei ou pu s
places hem in a g ey a ea wi h espec o ce i ica ion and liabili y. Fu u e egula o y models
mus e ol e o assess no only he p oduc bu also he design sys em i sel , alida ing he
logic, aceabili y, and eliabili y o au oma ed design pla o ms.
Despi e hese challenges, se e al echnological ends a e poised o ein o ce and expand
he capabili ies o KBE in biomedical enginee ing. Mos p ominen ly, he in eg a ion o
a i icial in elligence in o KBE wo k lows o e s new pa hways o da a-d i en op imiza ion,
p edic i e modelling, and eal- ime pe sonaliza ion. Machine lea ning models, ained on
ex ensi e simula ion o clinical da ase s, can se e as su oga e e alua o s in op imiza ion
loops, accele a ing he gene a ion o ana omically adap ed de ices. Addi ionally, AI can assis
in he ex ac ion o ana omical ea u es om scans, ecogni ion o pa hological pa e ns, o
iden i ica ion o op imal ma e ial con igu a ions—all asks ha enhance and complemen
ule-based KBE sys ems.
Ano he majo de elopmen is he mo e owa d cloud-based and collabo a i e pla o ms
o biomedical design. These pla o ms allow dis ibu ed access o in elligen design ools,
enabling clinicians, enginee s, and manu ac u e s o in e ac wi h pa ien da a and design logic
in eal ime. Such sys ems no only educe he la ency be ween da a acquisi ion and p oduc
gene a ion bu also suppo e sion con ol, audi ing, and mul i-use wo k lows— ea u es
essen ial o egula ed medical en i onmen s. In eg a ion wi h hospi al in o ma ion sys ems
(HIS), PACS/RIS sys ems, and addi i e manu ac u ing se ices allows o seamless ansi ions
be ween diagnosis, design, and p oduc ion.
Mo eo e , ad ances in web-based con igu a o s, pa ame ic geome y APIs, and digi al
wins a e making i possible o build KBE sys ems ha a e accessible no only o enginee s bu
also o clinical use s. Su geons, p os he is s, o ehabili a ion specialis s can in e ac wi h
simpli ied in e aces—inpu ing pa ien measu emen s, selec ing ea men scena ios, and
p e iewing de ice op ions—while he backend sys em dynamically applies expe logic o
gene a e op imized designs.
Looking o wa d, he con e gence o KBE wi h AI, cloud in as uc u e, and digi al heal h
ecosys ems poin s owa d a new pa adigm o biomedical p oduc de elopmen : one whe e
au oma ion does no educe human in ol emen bu a he augmen s i , cap u ing expe
easoning, ensu ing consis ency, and enabling apid, pa ien -speci ic ca e a scale.
4.2 Summa y and Recommenda ions
The g owing demand o ana omically cus omized biomedical de ices has p omp ed a
e hinking o adi ional design pa adigms in a o o au oma ed, knowledge-d i en
app oaches. This chap e has ou lined how Knowledge-Based Enginee ing (KBE), when
combined wi h mode n CAD sys ems and addi i e manu ac u ing, o e s a scalable and eliable
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 71
pa hway o deli e indi idualized solu ions in o ho ics, p os he ics, implan s, and su gical
planning.
Beginning wi h he a ionale and p inciples o ana omical cus omiza ion, he chap e
in oduced key modelling s a egies—pa ame ic and non-pa ame ic—and demons a ed
how hey in luence he design o pa ien -speci ic de ices. In elligen CAD models and KBE logic
s uc u es we e shown o o m he backbone o au oma ion, enabling apid egene a ion o
geome ies based on clinical and ana omical inpu s. Th ough eal-wo ld examples and case
s udies, i was illus a ed how design au oma ion sys ems can ans o m he enginee ing
p ocess om a ime-consuming, expe -d i en ask in o a s eamlined, ule-based pipeline.
The inco po a ion o ini e elemen analysis (FEA) was also explo ed, highligh ing how
simula ion and AI can be used o e alua e and op imize de ice pe o mance be o e
ab ica ion.
F om his ounda ion, se e al c i ical insigh s and ecommenda ions can be d awn, based
on ea lie wo k (Gó ski 2025, Gó ski e al. 2024a):
1. Fo maliza ion o domain knowledge is essen ial. Success ul implemen a ion o KBE in
biomedical design depends on he s uc u ed cap u e and encoding o clinical, ana omical,
and enginee ing expe ise. Ins i u ions de eloping such sys ems should in es in
in e disciplina y knowledge acquisi ion s a egies and ea knowledge i sel as a c i ical
enginee ing asse .
2. Robus modelling in as uc u e enables scalabili y. In elligen CAD models mus be
designed wi h pa ame e s abili y, egene a ion obus ness, and logic anspa ency in mind.
Reusabili y and modula i y a e key a ibu es o building scalable au oma ion wo k lows ha
can adap o e ol ing clinical needs.
3. Simula ion and AI ex end he alue o au oma ion. The in eg a ion o FEA and AI no
only s eng hens design alida ion bu also unlocks oppo uni ies o eal- ime
pe sonaliza ion, p edic i e adjus men s, and eedback-in o med op imiza ion. These
echnologies should be embedded in o ea ly s ages o sys em a chi ec u e a he han added
as a e hough s.
4. Clinical wo k lows mus be conside ed om he s a . To be adop ed in p ac ice,
au oma ed design sys ems mus align wi h exis ing clinical p ocesses, use compe encies, and
egula o y expec a ions. Web-based in e aces, guided inpu s, and au oma ic documen a ion
gene a ion a e c i ical enable s o clinical in eg a ion.
5. Con inuous alida ion and human o e sigh emain indispensable. While au oma ion
enhances e iciency and consis ency, he need o expe supe ision – pa icula ly in a e,
speci ic cases – emains. Hyb id wo k lows ha combine au oma ed p ocessing wi h expe
e iew s ike an op imal balance be ween speed and sa e y.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 72
Ul ima ely, he ansi ion owa d au oma ed biomedical design is no me ely a echnical
upg ade. I is a shi in how knowledge, echnology, and clinical ca e con e ge o se e
indi idual pa ien s mo e e ec i ely. By embedding medical logic in o digi al ools, KBE sys ems
help ensu e ha cus omiza ion becomes no a ba ie , bu a de aul s anda d in he nex
gene a ion o pa ien -cen e ed ca e.
Biomedical Inno a ions h ough Digi al
T ans o ma ion o Addi i e Technologies and
Knowledge Exchange - ERASMUS KA 220
BIOMEDIX
This p ojec has been unded wi h suppo om he Eu opean Commission. This publica ion e lec s he iews only o he au ho s, and he
Commission canno be held esponsible o any use which may be made o he in o ma ion con ained he ein.
Page | 73
Re e ences
Akkad K., Mehboob H., Alyamani R., & Ta lochan F. (2023). A Machine-Lea ning-Based App oach o
P edic ing Mechanical Pe o mance o Semi-Po ous Hip S ems. Jou nal o Func ional Bioma e ials,
14(3), 156.
Alcaide-Ma zal, J., Diego-Más, J. A., Asensio-Cues a, S., & Pique as-Fiszman, B. (2013). An explo a o y
s udy on he use o digi al sculp ing in concep ual p oduc design. Design S udies, 34(2), 264-284.
Al-Fakih E.A., Abu Osman N.A., Mahmad Adikan F.R. (2016). Techniques o In e ace S ess
Measu emen s wi hin P os he ic Socke s o T ans ibial Ampu ees: A Re iew o he Pas 50 Yea s o
Resea ch. Senso s, 16(7), 1119.
Bala amak ishnan T.M., Na a ajan S., & S ini asan, S. (2020). Roll-o e shape o a p os he ic oo : a
ini e elemen e alua ion and expe imen al alida ion. Med. Biol. Eng. Compu ., 58(10), 2259.
Baumann, F. W., Kopp, O., & Rolle , D. (2017). Abs ac API o 3D p in ing ha dwa e and so wa e
esou ces. The In e na ional Jou nal o Ad anced Manu ac u ing Technology, 92(1), 1519-1535.
Ben-Nun, T., Gamblin, T., Hollman, D. S., K ishnan, H., & Newbu n, C. J. (2020, No embe ). Wo k lows
a e he new applica ions: Challenges in pe o mance, po abili y, and p oduc i i y. In 2020 IEEE/ACM
In e na ional Wo kshop on Pe o mance, Po abili y and P oduc i i y in HPC (P3HPC) (pp. 57-69). IEEE.
Bonne X., Pille H., Fodé P., La as e F., & Skalli W. (2012). Fini e elemen modelling o an ene gy-
s o ing p os he ic oo du ing he s ance phase o ans ibial ampu ee gai . P oc. Ins . Mech. Eng. Pa
H, 226(1), 70.
Cas o-F anco A.D., Siquei os-He nández M., Ga cía-Angel V., Mendoza-Muñoz I., Va gas-Osuna L.E.,
& Magaña-Almague H.D. (2024). A Re iew o Na u al Fibe -Rein o ced Composi es o Lowe -Limb
P os he ic Designs. Polyme s, 16(9), 1293.
Chen D., Su P., O ikku i S., Va holomeos P., Tahmasebi K.N., & Ka amousadakis M. (2022). Analyzing
Dynamic Ope a ional Condi ions o Limb P os he ic Socke s wi h a Mecha onics-Twin F amewo k.
Applied Sciences, 12(3):986.
Cucos, M. M., Pis a, I. M., & Ripanu, M. I. (2018). P oduc enginee ing design enhancing by
pa ame e izing he 3D solid model. In MATEC Web o Con e ences (Vol. 178, p. 05011). EDP Sciences.
Da id Müzel S., Bonhin E.P., Guima ães N.M., & Guidi E.S. (2020). Applica ion o he Fini e Elemen
Me hod in he Analysis o Composi e Ma e ials: A Re iew. Polyme s, 12(4), 818.
Gó ski F., 2025, Compu e Aided Design o 3D P in able Ana omically Shaped Medical De ices:
Me hodologies and Applica ions, CRC P ess, Taylo & F ancis.
Gó ski, F., Denysenko, Y., Kuczko, W., & Żukowska, M. (2024b). Au oma ed Design and Vi ual Fi ing
o 3D P in ed Bicycle P os heses o Child en. In e na ional Con e ence Inno a ion in Enginee ing, 242–
253. Sp inge .
Gó ski, F., Denysenko, Y., Kuczko, W., Żukowska, M., Wichnia ek, R., Zawadzki, P., & Ryba czyk, J.
(2024a). Indi idualized 3D P in ed O hopaedic and P os he ic De ices Using Au oMedP in
Technology--Me hodologies and Examples. Ad ances in Science and Technology. Resea ch Jou nal,
18(6).