IFAC Pape sOnLine 58-24 (2024) 332–337
ScienceDi ec
A ailable online a www.sciencedi ec .com
2405-8963 Copy igh © 2024 The Au ho s. This is an open access a icle unde he CC BY-NC-ND license
.
Pee e iew unde esponsibili y o In e na ional Fede a ion o Au oma ic Con ol.
10.1016/j.i acol.2024.11.059
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
10.1016/j.i acol.2024.11.059 2405-8963
Copy igh ©
2024 The Au ho s. This is an open access a icle unde he CC BY-NC-ND license
(
h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
)
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail:
{
ana.ca olina.ca doso,josep.m. on
}
@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
Ana Ca olina Ca doso de Sousa e al. / IFAC Pape sOnLine 58-24 (2024) 332–337 333
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
Copy igh ©
2024 The Au ho s. This is an open access a icle unde he CC BY-NC-ND license
(
h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
)
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
P edic i e F amewo k o Elec ical
S imula ion Cycling in Spinal Co d Inju y ⋆
Ana Ca olina Ca doso de Sousa ∗,∗∗
Josep M. Fon -Llagunes ∗,∗∗
∗Depa men o Mechanical Enginee ing, Uni e si a Poli `ecnica de
Ca alunya, Ba celona, Spain
∗∗ Ins i u de Rece ca San Joan de D´eu, Esplugues de Llob ega , Spain
(e-mail: {ana.ca olina.ca doso,josep.m. on }@upc.edu).
Abs ac :
Enhancing he e icacy o spinal co d inju y (SCI) ehabili a ion is c ucial o a pa ien ’s op imal
eco e y. While unc ional elec ical s imula ion (FES) cycling s ands as a s anda d he apy,
achie ing no able imp o emen s p o es challenging due o he inhe en complexi ies embedded
in he dynamics o he mo emen . Indeed, o e coming he ime-consuming pa ame e selec ion
p ocess becomes impe a i e, p omp ing he de elopmen o p edic i e models h ough op imal
con ol simula ion. The cu en challenge lies in he demand o a bluep in ha conside s he
unique pa icula i ies o SCI FES cycling. In esponse, ou inno a i e app oach in oduces a
no el amewo k and showcases i s applica ion in sol ing p edic i e models. Le e aging open-
sou ce ools, including OpenSim and Blende , we buil he FES cycling model. Subsequen ly, we
ou lined wo p oblems o mula ions wi hin OpenSim Moco: (P1) mo ing om poin A o poin B
wi h di e en c ank esis ances, and (P2) acking a ge speeds. Ou s udy e eals he success ul
con e gence o hese simula ions, demons a ing he in eg a ed amewo k’s obus ness and
e iciency. Indeed, he p esen ed solu ion add esses he need o mul iple simula ions, he eby
mi iga ing he leng hy cons ain s o p io me hods and pa ing he way o p ac ical and ime-
e ec i e in eg a ion o digi al wins in clinical applica ions.
Keywo ds: Rehabili a ion enginee ing including ehabili a ion obo ics; Biomedical sys em
modelling; Simula ion and isualiza ion.
1. INTRODUCTION
Func ional elec ical s imula ion (FES) cycling is a alu-
able he apeu ic app oach o indi iduals wi h spinal co d
inju y (SCI), p omo ing lowe -body muscle heal h and
o e all imp o emen in hei quali y o li e ( an de Schee
e al., 2021). FES uses elec ical cu en s o ac i a e and
s imula e ne es o imp o e o es o e unc ions in indi-
iduals wi h neu ological condi ions o pa alysis. Despi e
i s bene i s, cus omising his ea men in clinics is o en
ime-consuming and ine icien , gi en ha con olling he
s imula ion signal equi es an in-dep h comp ehension o
he dynamic in e ac ion be ween he musculoskele al sys-
em and he e gome e (Pizzola o e al., 2021). Cu en ly,
clinicians ely on gene ic pa ame e s and a ial-and-e o
me hodology.
Biomechanical simula ions can help us es ima e ha d- o-
measu e quan i ies and be e pe sonalise he s imula ion
o each pa ien . Indeed, p e ious wo ks o e ed a com-
p ehensi e unde s anding o he in e ac ions be ween he
musculoskele al and mechanical componen s in cycling. In
a p io amewo k (de Sousa e al., 2016), we success-
ully in eg a ed speci ic cha ac e is ics o FES cycling o
⋆The Bea iu de Pin´os Fellowship P og amme suppo ed his
wo k (call BP2021), execu ed by he Ag`encia de Ges i´o d’Aju s
Uni e si a is i de Rece ca (AGAUR).
indi iduals wi h SCI. Howe e , he limi a ion o solely
pe o ming o wa d dynamics analysis equi ed mul iple
simula ions o explo e possibili ies be o e iden i ying he
mos sui able one. Fo ins ance, in (de Sousa e al., 2021),
we ex ensi ely explo ed he impac o passi e o hoses
on cycling speed, conduc ing o e 600 simula ions wi h
a ying o hoses pa ame e s o selec he mos easible
ou come. De eloping less ime-consuming app oaches is
s ill one o he challenges o achie ing clinically use ul
musculoskele al simula ions (F egly, 2021).
P edic i e simula ions can employ biomechanical models
o ehabili a ion ea men design (Feb e -Na ´ıa e al.,
2023). Using his me hod, he compu a ional model gene -
a es an objec i e mo emen o ecas h ough nume ical op-
imisa ion o iden i y he bes p o ile o each pa ien (e.g.,
iming and in ensi y o he s imula ion). Recen ly, Pa k
e al. (2022a) and Clancy e al. (2023) de eloped com-
pu a ionally e icien p edic i e simula ion amewo ks o
in es iga e, o example, how a cyclis could al e he pedal
o ce di ec ion (Pa k e al., 2022b). Despi e hei me i s,
he e a e conside a ions in i s applicabili y o FES cycling
as hese amewo ks ocused on heal hy pa icipan s and
sea ed pedalling e gome e s.
This pape add esses hese gaps by in oducing a no el
amewo k ha cap u es he speci ic cha ac e is ics o
FES cycling o indi iduals wi h SCI and signi ican ly
s eamlines he modelling p ocess. The p ima y esea ch
ques ion d i ing ou in es iga ion is: How can a p edic i e
model, as de ailed in Sec ion 2, o ecas and op imise he
cycling mo emen unde p ede ined kinema ics? The sub-
sequen sec ions (Sec ion 3 and 4) p esen he p elimina y
esul s and implica ions o ou model, o e ing insigh s in o
he ans o ma i e po en ial o p edic i e app oaches o
enhancing he he apeu ic impac o FES cycling in he
con ex o SCI ehabili a ion.
2. METHODS
Fi s , we used OpenSim (Delp e al., 2007) o build a
cycling model ailo ed explici ly o FES cycling o SCI
ehabili a ion (Sec. 2.1). OpenSim, an open-sou ce so -
wa e, was chosen o i s obus kinema ics and dynamics
ools, enabling he de elopmen and simula ion o in ica e
models. Subsequen ly, we o mula ed and sol ed op imal
con ol p oblems o he de eloped model using Open-
Sim Moco (Dembia e al., 2020) (Sec. 2.2). OpenSim Moco
is a dedica ed oolki designed o op imise mo ion and
con ol musculoskele al models wi hin OpenSim.
2.1 Modelling cycling
We sc ip he cycling model in Py hon using he OpenSim
API ools (Fig. 1) 1. Table 1 p esen s ele an pa ame e s
o modelling.
Human model We loaded he musculoskele al human
model called Gai 10do 18musc (Fig. 1(a)). We selec ed i
because o i s simplici y, ye i s ill ep esen s ou applica-
ion, which acili a es as e de elopmen . I ’s impo an o
no e ha his model is p ima ily in ended o educa ional
pu poses. Fo una ely, he model can be easily eplaced
wi h mo e ad anced ea u es in u u e i e a ions.
C ank bodies Because he c ank spindle, a ms, and
pedals a e no p e-exis ing in he OpenSim da abase,
we designed hese componen s using he ee Blende
so wa e, simila o (de Sousa e al., 2016). The momen s o
ine ia we e calcula ed conside ing ha he c ank spindle
is modelled as a la disc and he c ank a ms as cuboids,
assuming a uni o m mass dis ibu ion (Fig. 1(b)).
C ank join The c ank spindle is linked o he model’s
g ound h ough a pin join , allowing o a ion exclusi ely
in he sagi al plane wi hou ansla ional mo ion. Each
pedal is welded o he c ank a ms a dis ances o xcand yc
ela i e o he c ank cen e (Fig. 1(c)).
Fee ancho s To secu e he pedals o he human model,
we implemen ed cons an dis ance cons ain s (Cons an -
Dis anceCons ain ) be ween he cen e o he calcaneus
bone and he cen e o he pedal body (Fig. 1(d)). As
a esul , he ee ans e o ce o he pedal h ough he
cons ain o ces necessa y o uphold his ixed dis ance.
Locked join s We locked all deg ees o eedom no usu-
ally in ol ed in FES cycling: pel is ( il , lis , o a ion,
ansla ion), lumba (ex ension, bending, o a ion), and
ankle. Res ic ing his mo ion mi o s he biomechanical
1Cu en ly being de eloped and main ained a :
h ps://sim k.o g/ s/?g oup_id=1553
cons ain s o en obse ed in indi iduals wi h SCIs, allow-
ing us o simula e hei bio-mechanical limi a ions.
Coo dina es We se he ini ial pel is and lumba posi-
ion o mimic he pos u e o an indi idual on a ecumben
bicycle, commonly obse ed in SCI ehabili a ion, as illus-
a ed in (Bo e al., 2017) (Fig. 1(e)). Keeping he ankle
ixed and ea ing he oo and ibia oge he as a single
segmen o ms a ou -ba sys em, whe e he c ank angle
dic a es he angles o he hips and knees.
Ac ua o s We modi ied he human model by emo ing
all muscle o ces, as dealing wi h nume ous deg ees o
eedom and ac ua o s can in oduce complexi y. We op ed
o ideal ac ua o s on speci ic join s o simpli y he model
and acili a e he solu ion p ocess. In his model, we
inco po a ed ou coo dina e ac ua o s - o he le and
igh hips and knees - each capable o exe ing a maximum
o que o 100Nm.
C ank esis ance We in oduced esis ance o he c ank
by employing he Exp essionBasedCoo dina eFo ce. This
model’s exp ession co esponds o a cons an o que wi h
a a iable alue anging om 0 o 20Nm.
Table 1. Biomechanical model and c ank pa-
ame e s.
Human model
Ankles angles (αa) 90.0°
Pel is il angle (αh) 40.0°
Pel is x-dis ance (xh) 0.65m
Pel is y-dis ance (yh) 0.05m
C ank spindle
Mass (ms) 0.3kg
Radius ( ) 0.1m
Thickness (h) 0.02m
C ank a ms
Mass (mp) 0.3kg
Leng h (a) 0.175m
Wid h (b) 0.04m
Thickness (c) 0.01m
Pedal
Dis ance pedal (d) 0.012m
Pedal x-dis ance (xc) 0.08m
Pedal y-dis ance (yc) 0.15m
2.2 P edic i e cycling simula ion
OpenSim Moco o e s a se o cos and cons ain modules
o add ess speci ic aspec s o op imal con ol p oblems. In
ou app oach, we le e age he capabili ies o he CasADi li-
b a y (Ande sson e al., 2019) o ans o m he con inuous
op imal con ol p oblem in o a ini e-dimensional nonlin-
ea p og am. Fu he mo e, we used he well-es ablished
g adien -based nonlinea IPOPT sol e (W¨ach e and
Biegle , 2006) o ind op imal solu ions.
Mo eo e , we con igu ed a MocoS udy o cycling using
he model buil in Sec ion 2.1 in Py hon. Ou ini ial ocus
was de e mining he join ac ua ion equi ed o mo e he
c ank om an ini ial o a inal posi ion. Subsequen ly,
we ex ended ou in es iga ion o sol e o he op imal
join ac ua ion necessa y o main ain a cons an speed.
Table 2 p esen s all pa ame e s conside ed in his s udy
o P oblems 1 and 2.
334 Ana Ca olina Ca doso de Sousa e al. / IFAC Pape sOnLine 58-24 (2024) 332–337
(a) human model (b) c ank bodies
(c) pedal posi ion
(d) ee ancho s
𝑑𝑑
𝑥𝑥𝑐𝑐
𝑦𝑦𝑐𝑐
(e) human posi ion
𝑟𝑟
ℎ
𝑎𝑎
𝑏𝑏
𝑐𝑐
c ank spindle
c ank a ms and
pedals
𝑦𝑦ℎ
𝑥𝑥ℎ
𝛼𝛼𝑎𝑎
𝛼𝛼ℎ
Fig. 1. Componen s o modelling FES cycling. (a) The ep esen a ion o he human model used in he
s udy (Gai 10do 18musc) showcases i s musculoskele al s uc u e. (b) De ail o he c ank spindle, c ank a ms, and
pedals. Va iables ela ed o he la disc and cuboids a e highligh ed o demons a e hei design. (c) Rep esen a ion
o he poin o he pedal a achmen ela ed o he c ank cen e. (d) Visualisa ion o a oo posi ioned on he pedal,
emphasising he dis ance be ween hem. (e) The human is posi ioned in a pos u e esembling an indi idual on a
ecumben bicycle.
P oblem 1 (P1) The goal is o minimise he ime (Moco-
FinalTimeGoal) and e o (MocoCon olGoal) equi ed o
ansi ion om an ini ial o a inal c ank posi ion, ensu ing
inal imes a e below 2.0s. Simula ions co e 21 scena ios
wi h esis ances anging om RP1= 0 o 20Nm.
P oblem 2 (P2) The aim is o ack he c ank speed s a e
o 1.5s (MocoS a eT ackingGoal) wi h a c ank esis ance
o RP1= 10Nm. Addi ionally, we en o ced he maximum
speed as 1.2˙
θP2 o a oid o e shoo s. Eigh cons an c ank
speed p o iles a e gene a ed, a ying om ˙
θP2= 240.0 o
520.0°/s (i.e. 40 o 86 pm).
Ha dwa e speci ica ions The simula ions we e execu ed
on a wo ks a ion ea u ing an AMD Ryzen 7 5700U
p ocesso wi h Radeon G aphics, ope a ing a a clock
speed o 1801Mhz. This p ocesso has 8 co es and 16 logical
p ocesso s, p o iding su icien compu a ional powe o
complex biomechanical simula ions. The sys em an on
he Mic oso Windows 11 P o ope a ing sys em, and i s
pe o mance was u he augmen ed by 32.0GB o ins alled
physical memo y (RAM).
3. RESULTS
3.1 P oblem 1
The simula ions success ully con e ged o solu ions o all
c ank esis ances wi hin 2.0s (Fig. 2(a), o ange colou s
mean lowe esis ances while blue colou s mean highe
esis ances). We implemen ed a wa m-s a app oach o
enhance con e gence, u ilising solu ions om op imising
lowe esis ance le els as ini ial guesses o subsequen
p oblems in he di ec colloca ion algo i hm. Speci ically,
he solu ion ob ained o RP1= 0Nm was employed o
ini ialize he op imisa ion o RP1= 1Nm, and so o h.
Table 2. Simula ion pa ame e s.
(a) P oblem 1
Maximum end ime 2.0s
Ini ial c ank posi ion 0.0°
Final c ank posi ion 360.0°
Each join o que 100.0Nm
Maximum speed 360.0°/s
Resis ance (RP1) 0.0-20.0Nm
Mesh in e als 20
Con e gence ole ance 1 ×10−2
Cons ain ole ance 1 ×10−4
(b) P oblem 2
End ime 1.5s
Ini ial c ank posi ion 0.0°
Each join o que 100.0Nm
Ta ge speed ( ˙
θP2) 240.0 - 520.0°/s
Resis ance 10.0Nm
Mesh in e als 30
Con e gence ole ance 1 ×10−2
Cons ain ole ance 1 ×10−4
Ana Ca olina Ca doso de Sousa e al. / IFAC Pape sOnLine 58-24 (2024) 332–337 335
(a) human model (b) c ank bodies
(c) pedal posi ion
(d) ee ancho s
𝑑𝑑
𝑥𝑥𝑐𝑐
𝑦𝑦𝑐𝑐
(e) human posi ion
𝑟𝑟
ℎ
𝑎𝑎
𝑏𝑏
𝑐𝑐
c ank spindle
c ank a ms and
pedals
𝑦𝑦ℎ
𝑥𝑥ℎ
𝛼𝛼𝑎𝑎
𝛼𝛼ℎ
Fig. 1. Componen s o modelling FES cycling. (a) The ep esen a ion o he human model used in he
s udy (Gai 10do 18musc) showcases i s musculoskele al s uc u e. (b) De ail o he c ank spindle, c ank a ms, and
pedals. Va iables ela ed o he la disc and cuboids a e highligh ed o demons a e hei design. (c) Rep esen a ion
o he poin o he pedal a achmen ela ed o he c ank cen e. (d) Visualisa ion o a oo posi ioned on he pedal,
emphasising he dis ance be ween hem. (e) The human is posi ioned in a pos u e esembling an indi idual on a
ecumben bicycle.
P oblem 1 (P1) The goal is o minimise he ime (Moco-
FinalTimeGoal) and e o (MocoCon olGoal) equi ed o
ansi ion om an ini ial o a inal c ank posi ion, ensu ing
inal imes a e below 2.0s. Simula ions co e 21 scena ios
wi h esis ances anging om RP1= 0 o 20Nm.
P oblem 2 (P2) The aim is o ack he c ank speed s a e
o 1.5s (MocoS a eT ackingGoal) wi h a c ank esis ance
o RP1= 10Nm. Addi ionally, we en o ced he maximum
speed as 1.2˙
θP2 o a oid o e shoo s. Eigh cons an c ank
speed p o iles a e gene a ed, a ying om ˙
θP2= 240.0 o
520.0°/s (i.e. 40 o 86 pm).
Ha dwa e speci ica ions The simula ions we e execu ed
on a wo ks a ion ea u ing an AMD Ryzen 7 5700U
p ocesso wi h Radeon G aphics, ope a ing a a clock
speed o 1801Mhz. This p ocesso has 8 co es and 16 logical
p ocesso s, p o iding su icien compu a ional powe o
complex biomechanical simula ions. The sys em an on
he Mic oso Windows 11 P o ope a ing sys em, and i s
pe o mance was u he augmen ed by 32.0GB o ins alled
physical memo y (RAM).
3. RESULTS
3.1 P oblem 1
The simula ions success ully con e ged o solu ions o all
c ank esis ances wi hin 2.0s (Fig. 2(a), o ange colou s
mean lowe esis ances while blue colou s mean highe
esis ances). We implemen ed a wa m-s a app oach o
enhance con e gence, u ilising solu ions om op imising
lowe esis ance le els as ini ial guesses o subsequen
p oblems in he di ec colloca ion algo i hm. Speci ically,
he solu ion ob ained o RP1= 0Nm was employed o
ini ialize he op imisa ion o RP1= 1Nm, and so o h.
Table 2. Simula ion pa ame e s.
(a) P oblem 1
Maximum end ime 2.0s
Ini ial c ank posi ion 0.0°
Final c ank posi ion 360.0°
Each join o que 100.0Nm
Maximum speed 360.0°/s
Resis ance (RP1) 0.0-20.0Nm
Mesh in e als 20
Con e gence ole ance 1 ×10−2
Cons ain ole ance 1 ×10−4
(b) P oblem 2
End ime 1.5s
Ini ial c ank posi ion 0.0°
Each join o que 100.0Nm
Ta ge speed ( ˙
θP2) 240.0 - 520.0°/s
Resis ance 10.0Nm
Mesh in e als 30
Con e gence ole ance 1 ×10−2
Cons ain ole ance 1 ×10−4
1.0 3.0 5.0 7.0 9.0 11.0 13.0 15.0 17.0 19.0
C ank esis ance [Nm]
1.020
1.022
1.024
1.026
1.028
Final ime [s]
(a) Final imes o R = 0 o 20 Nm
0.0 0.2 0.4 0.6 0.8 1.0
Time [s]
0
100
200
300
400
C ank speed [°/s]
(b) C ank speed o R = 10 Nm
Maximum speed
C ank speed
0.0 0.2 0.4 0.6 0.8 1.0
Time [s]
−100
−50
0
50
100
Momen [Nm]
(c) Le hip
0.0 0.2 0.4 0.6 0.8 1.0
Time [s]
−100
−50
0
50
100
Momen [Nm]
(d) Righ hip
0.0 0.2 0.4 0.6 0.8 1.0
Time [s]
−100
−50
0
50
100
Momen [Nm]
(e) Le knee
0.0 0.2 0.4 0.6 0.8 1.0
Time [s]
−100
−50
0
50
100
Momen [Nm]
( ) Righ knee
Fig. 2. Solu ions o P oblem 1, o ange colou s mean lowe esis ances while blue colou s mean highe esis ances. (a)
Du a ion o each solu ion o ansi ion om he ini ial o he inal c ank posi ion ac oss all esis ances. (b) C ank
speed s a e ajec o y om he solu ion wi h a esis ance o 10Nm. (c) Le hip momen . (d) Righ hip momen .
(e) Le knee momen . ( ) Righ knee momen .
The objec i e o minimising he inal ime in ou s udy
led o solu ions whe e he speed was maximised. This
ade-o is essen ial in achie ing e icien ansi ions be-
ween c ank posi ions, as e idenced in he example o
RP1= 10Nm (Fig. 2(b)). The esul s consis en ly show
ha he inal ime sligh ly su passed 1.0s. The e o e, he
speed was main ained close o he maximum 360.0°/s.
Mo eo e , he join momen s o highe esis ances a e
no ably highe han hose o lowe esis ances (Fig. 2(c- ),
highe esis ances a e illus a ed in blue colou s).
3.2 P oblem 2 (P2)
The simula ions success ully con e ged o solu ions o all
c ank speed p o iles (Fig. 3(a)). A wa m-s a s a egy was
implemen ed simila o P1. In his case, we ini ia ed he
p ocess by u ilising he solu ion ob ained o RP1= 10Nm
(see Fig. 2(b)) as an ini ial guess o acking he speed
p o ile a ˙
θP2= 360.0°/s. Subsequen ly, he esul s om
360°/s we e used o ind he solu ion o 400°/s, 400°/s
o 440°/s, and so on. The p ocess was hen e e sed,
using he solu ion om 360°/s o ind 320°/s, 320°/s o
ind 280°/s, and so o h. Fu he mo e, we calcula ed he
no malised oo mean squa ed e o (RMSE) (Fig. 3(b))
o measu e he a e age magni ude o he e o s be ween
he simula ions and he expec ed alues.
4. DISCUSSION
Ou p oposed amewo k success ully con e ged o he
ou lined p oblems (P1 and P2), showcasing i s capabili y
o add ess complex cycling- ela ed mo emen scena ios.
In eg a ing OpenSim Moco, CasADi, and IPOPT p o ed
a obus combina ion, e ec i ely sol ing in ica e p ob-
lems. Mo eo e , ou app oach o e comes he limi a ions o
p e ious me hodologies, such as he ex ensi e simula ion
equi emen s no ed in ou ea lie wo k (de Sousa e al.,
2021), by using an op imal con ol me hod.
In P1, he inal ime ends o inc ease wi h highe c ank
esis ance (Fig. 2(a)), indica ing a sligh di icul y o
he model in a aining he inal posi ion. This end is
u he e iden in he esul s depic ing highe momen s
associa ed wi h inc eased esis ances (Fig. 2(c- )). This
co ela ion aligns wi h he expec ed beha iou , as ele a ed
esis ances necessi a e heigh ened momen e o s and/o
ime o success ully execu e he desi ed c ank posi ion
ansi ion. In P2, he speed ajec o y gene ally aligns
wi h each e e ence speed (Fig. 3(a)) wi h a simila RMSE
h ough all speeds (Fig. 3(b)).
Gi en he inhe en edundancy in he p oblems we ad-
d ess, ou amewo k con on s scena ios whe e mo e han
one solu ion may mee he con e gence c i e ia. The e o e,
336 Ana Ca olina Ca doso de Sousa e al. / IFAC Pape sOnLine 58-24 (2024) 332–337
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time [s]
0
100
200
300
400
500
Speed [°/s]
(a) Speed ajec o y o each speed e e ence
240 280 320 360 400 440 480 520
Re e ence speed [º/s]
0.000
0.005
0.010
0.015
0.020
0.025
No malised RMSE [º/s]
(b) No malised oo mean squa ed e o
Fig. 3. Solu ions o P oblem 2. (a) C ank speed ajec o ies. (b) No malised RMSE o each speed ajec o y.
we acknowledge ha he solu ions p esen ed in ou e-
sul s migh no necessa ily ep esen he absolu e op imal
ou comes. This inhe en a iabili y in possible solu ions
is exempli ied, o ins ance, by he obse a ion ha he
model can a ain he inal posi ion mo e apidly wi h a
pedal o ce o 9Nm compa ed o 8 o 7Nm (Fig. 2(a)), o
e en ha some speeds p esen ed highe RMSE (Fig. 3).
The op imisa ion p edic ion depends on he ini ial guess,
wi h con e gence o en aking hou s when no p o ided.
Implemen ing a wa m-s a s a egy has p o en e ec i e,
enabling he disco e y o mos solu ions in less han 30min.
This s eamlined app oach signi ican ly educes compu a-
ional ime, enhancing he e iciency o ou op imisa ion
p ocess and acili a ing quicke explo a ion o op imal
solu ions in he con ex o ou simula ions.
Enhancing modelling is impe a i e o cap u e cycling dy-
namics accu a ely. The model we used (Gai 10do 18musc)
may gene a e excessi e passi e o ce du ing signi ican hip
and knee lexion, esul ing in inaccu a e p edic ions. Fo -
una ely, we can adap ou pipeline o accommoda e mo e
ealis ic models such as (Ca elli e al., 2019). Simila ly,
we can include o he bicycle ypes (e.g., (Bap is a e al.,
2022)) o unde s and he cycling dynamics be e . Ul i-
ma ely, inco po a ing a eal- ime digi al ep esen a ion o
he biomechanical sys em can enhance i s applicabili y in
clinics (Lloyd e al., 2023; Quinn e al., 2023). Addi ionally,
u u e i e a ions could explo e a ia ions in FES cycling
condi ions, conside ing al e ed muscle unc ion ypical o
SCI. The e o e, clinicians can adjus s imula ion pa ame-
e s o a ge speci ic muscle g oups based on p edic i e
simula ion esul s om each pa ien .
In he con ex o ee ancho s, i is wo h no ing ha
we p e iously explo ed a con ac o ce model (de Sousa
e al., 2016). This app oach was no conside ed he e due o
ex ended compu a ion imes. Mo eo e , we also conside ed
welding he ee o pedals. S ill, limi a ions, pa icula ly in
3D models, led us o ou cu en solu ion (Fig. 1), s iking
a p ac ical balance be ween compu a ional e iciency and
physical accu acy. Howe e , he al e na i e app oaches
emain a p omising di ec ion o modelling e inemen .
We also highligh ha he ou comes o P1 and P2 do no
necessa ily exhibi symme y be ween bo h legs (Fig. 2(c-
)), despi e he expec a ion o simila join o ques in
co esponding cycling posi ions. Speci ically, e o s will be
di ec ed owa ds en o cing symme y in he solu ions o
OpenSim Moco (MocoPe iodici yGoal). This upda e aims
o e ine he simula ions and ensu e a mo e consis en
ep esen a ion o join ac ions in synch onised cycling
posi ions.
Mo eo e , u u e esea ch should op imise muscle-d i en
models and compa e he simula ed muscle exci a ions wi h
eal-wo ld da a (e.g., (C ossley e al., 2024)) o alida ion.
The compa a i e analysis will con ibu e o he alida ion
o ou model and p o ide aluable insigh s o imp o -
ing he p edic i e capabili ies o muscle-d i en models.
Th ough his app oach, we can emula e he in icacies o
muscle ac i a ion pa e ns du ing FES cycling. No e ha
o ob ain accep able elec omyog aphy signals du ing FES,
we need me hods such as (Hambly e al., 2023, 2024).
Las ly, he pipeline p esen ed he e can become compu a-
ionally demanding and p esen scalabili y issues. The e-
o e, u u e wo k should conside op imising simula ion
algo i hms, scaling he compu a ional in as uc u e, and
in eg a ing eal-wo ld da a alida ion echniques. By p i-
o i ising hese e o s, we can e ine he model’s accu acy
and e iciency, ul ima ely achie ing u ili y in clinical ap-
plica ions.
Ana Ca olina Ca doso de Sousa e al. / IFAC Pape sOnLine 58-24 (2024) 332–337 337
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time [s]
0
100
200
300
400
500
Speed [°/s]
(a) Speed ajec o y o each speed e e ence
240 280 320 360 400 440 480 520
Re e ence speed [º/s]
0.000
0.005
0.010
0.015
0.020
0.025
No malised RMSE [º/s]
(b) No malised oo mean squa ed e o
Fig. 3. Solu ions o P oblem 2. (a) C ank speed ajec o ies. (b) No malised RMSE o each speed ajec o y.
we acknowledge ha he solu ions p esen ed in ou e-
sul s migh no necessa ily ep esen he absolu e op imal
ou comes. This inhe en a iabili y in possible solu ions
is exempli ied, o ins ance, by he obse a ion ha he
model can a ain he inal posi ion mo e apidly wi h a
pedal o ce o 9Nm compa ed o 8 o 7Nm (Fig. 2(a)), o
e en ha some speeds p esen ed highe RMSE (Fig. 3).
The op imisa ion p edic ion depends on he ini ial guess,
wi h con e gence o en aking hou s when no p o ided.
Implemen ing a wa m-s a s a egy has p o en e ec i e,
enabling he disco e y o mos solu ions in less han 30min.
This s eamlined app oach signi ican ly educes compu a-
ional ime, enhancing he e iciency o ou op imisa ion
p ocess and acili a ing quicke explo a ion o op imal
solu ions in he con ex o ou simula ions.
Enhancing modelling is impe a i e o cap u e cycling dy-
namics accu a ely. The model we used (Gai 10do 18musc)
may gene a e excessi e passi e o ce du ing signi ican hip
and knee lexion, esul ing in inaccu a e p edic ions. Fo -
una ely, we can adap ou pipeline o accommoda e mo e
ealis ic models such as (Ca elli e al., 2019). Simila ly,
we can include o he bicycle ypes (e.g., (Bap is a e al.,
2022)) o unde s and he cycling dynamics be e . Ul i-
ma ely, inco po a ing a eal- ime digi al ep esen a ion o
he biomechanical sys em can enhance i s applicabili y in
clinics (Lloyd e al., 2023; Quinn e al., 2023). Addi ionally,
u u e i e a ions could explo e a ia ions in FES cycling
condi ions, conside ing al e ed muscle unc ion ypical o
SCI. The e o e, clinicians can adjus s imula ion pa ame-
e s o a ge speci ic muscle g oups based on p edic i e
simula ion esul s om each pa ien .
In he con ex o ee ancho s, i is wo h no ing ha
we p e iously explo ed a con ac o ce model (de Sousa
e al., 2016). This app oach was no conside ed he e due o
ex ended compu a ion imes. Mo eo e , we also conside ed
welding he ee o pedals. S ill, limi a ions, pa icula ly in
3D models, led us o ou cu en solu ion (Fig. 1), s iking
a p ac ical balance be ween compu a ional e iciency and
physical accu acy. Howe e , he al e na i e app oaches
emain a p omising di ec ion o modelling e inemen .
We also highligh ha he ou comes o P1 and P2 do no
necessa ily exhibi symme y be ween bo h legs (Fig. 2(c-
)), despi e he expec a ion o simila join o ques in
co esponding cycling posi ions. Speci ically, e o s will be
di ec ed owa ds en o cing symme y in he solu ions o
OpenSim Moco (MocoPe iodici yGoal). This upda e aims
o e ine he simula ions and ensu e a mo e consis en
ep esen a ion o join ac ions in synch onised cycling
posi ions.
Mo eo e , u u e esea ch should op imise muscle-d i en
models and compa e he simula ed muscle exci a ions wi h
eal-wo ld da a (e.g., (C ossley e al., 2024)) o alida ion.
The compa a i e analysis will con ibu e o he alida ion
o ou model and p o ide aluable insigh s o imp o -
ing he p edic i e capabili ies o muscle-d i en models.
Th ough his app oach, we can emula e he in icacies o
muscle ac i a ion pa e ns du ing FES cycling. No e ha
o ob ain accep able elec omyog aphy signals du ing FES,
we need me hods such as (Hambly e al., 2023, 2024).
Las ly, he pipeline p esen ed he e can become compu a-
ionally demanding and p esen scalabili y issues. The e-
o e, u u e wo k should conside op imising simula ion
algo i hms, scaling he compu a ional in as uc u e, and
in eg a ing eal-wo ld da a alida ion echniques. By p i-
o i ising hese e o s, we can e ine he model’s accu acy
and e iciency, ul ima ely achie ing u ili y in clinical ap-
plica ions.
5. CONCLUSION
Ou esea ch has con ibu ed o SCI ehabili a ion by
in oducing a amewo k o mo emen p edic ion ha
in eg a es speci ic cha ac e is ics o FES cycling. Le e ag-
ing compu a ional ools, we gained aluable insigh s in o
op imal con ol s a egies o join ac ua ion du ing cy-
cling. The demons a ed con e gence o ou model unde -
sco es i s po en ial in p edic ing and op imising. Mo eo e ,
he insigh s gained p o ide a ounda ion o ine- uning
s imula ion pa ame e s in a pa ien -speci ic manne . This
accomplishmen will empowe clinicians and esea che s o
explo e nuanced aspec s o pe sonalised ea men s a e-
gies, b idging he gap be ween heo e ical biomechanics
and p ac ical clinical applica ions.
ACKNOWLEDGEMENTS
This esea ch was ca ied ou wi hin he Biomechanical
Enginee ing Lab (BIOMEC), a di ision o he Resea ch
Cen e o Biomedical Enginee ing (CREB) wi hin Uni-
e si a Poli `ecnica de Ca alunya (UPC) and he Ins i u
de Rece ca San Joan de D´eu. The s udy ecei ed suppo
om he Bea iu de Pin´os Fellowship P og amme, Call
BP2021.
REFERENCES
Ande sson, J. A. E., Gillis, J., Ho n, G., Rawlings, J. B.,
Diehl, M., Ma . 2019. CasADi: a so wa e amewo k o
nonlinea op imiza ion and op imal con ol. Ma hema -
ical P og amming Compu a ion 11 (1), 1–36.
Bap is a, R., Mo ei a, M., Pinhei o, L., Pe ei a, T., Ca -
mona, G., F ei e, J., Bas os, J., Bo, A. P. L., May 2022.
Use -cen e ed design and spa ially-dis ibu ed sequen-
ial elec ical s imula ion in cycling o indi iduals wi h
pa aplegia. Jou nal o Neu oEnginee ing and Rehabili-
a ion 19 (1), 45.
Bo, A. P., da Fonseca, L. O., Guima aes, J. A., Fachin-
Ma ins, E., Pa edes, M. E., B indei o, G. A., De Sousa,
A. C. C., Do ado, M. C., Ramos, F. M., 2017. Cy-
cling wi h Spinal Co d Inju y: A No el Sys em o Cy-
cling Using Elec ical S imula ion o Indi iduals wi h
Pa aplegia, and P epa a ion o Cyba hlon 2016. IEEE
Robo ics & Au oma ion Magazine 24 (4), 58–65.
Ca elli, D., Wesseling, M., Jonke s, I., Lamon agne, M.,
2019. A musculoskele al model cus omized o squa ing
ask. Compu e Me hods in Biomechanics and Biomed-
ical Enginee ing 22 (1), 21–24.
Clancy, C., Ga i, A., Ong, C., Maly, M., Delp, S., Dec.
2023. Muscle-d i en simula ions and expe imen al da a
o cycling. Scien i ic Repo s 13.
C ossley, C. B., Diamond, L. E., Saxby, D. J., de Sousa,
A., Lloyd, D. G., Che Fo nusek, Pizzola o, C., May
2024. Join con ac o ces du ing semi- ecumben sea ed
cycling. Jou nal o Biomechanics 168, 112094.
de Sousa, A., Ramos, F., Na aez Do ado, M., da Fonseca,
L., B´o, A., 2016. A Compa a i e S udy on Con ol
S a egies o FES Cycling Using a De ailed Muscu-
loskele al Model. IFAC-Pape sOnLine 49 (32), 204–209.
de Sousa, A., Sousa, F., Bap is a, R., B´o, A., 2021. Pas-
si e Knee O hoses Assis ance in Func ional Elec ical
S imula ion Cycling in an Indi idual Wi h Spinal Co d
Inju y. IEEE T ansac ions on Neu al Sys ems and Re-
habili a ion Enginee ing 29, 690–698.
Delp, S., Ande son, F., A nold, A., Loan, P., Habib, A.,
John, C. T., Guendelman, E., Thelen, D. G., 2007.
OpenSim: Open-Sou ce So wa e o C ea e and Analyze
Dynamic Simula ions o Mo emen . IEEE T ansac ions
on Biomedical Enginee ing 54 (11), 1940–1950.
Dembia, C., Bianco, N., Falisse, A., Hicks, J., Delp, S.,
2020. OpenSim Moco: Musculoskele al op imal con ol.
PLOS Compu a ional Biology 16, e1008493.
Feb e -Na ´ıa, M., Nas , A., Eza i, M., B own, P., Fon -
Llagunes, J., McPhee, J., 2023. P edic i e mul ibody
dynamic simula ion o human neu omusculoskele al sys-
ems: a e iew. Mul ibody Sys em Dynamics.
F egly, B. J., Jan. 2021. A Concep ual Bluep in o
Making Neu omusculoskele al Models Clinically Use ul.
Applied Sciences 11 (5), 2037.
Hambly, M. J., De Sousa, A. C. C., Lloyd, D. G., Pizzo-
la o, C., Sep. 2023. EMG-In o med Neu omusculoskele-
al Modelling Es ima es Muscle Fo ces and Join Mo-
men s Du ing Elec ical S imula ion. In: IEEE .. In-
e na ional Con e ence on Rehabili a ion Robo ics. Vol.
2023. pp. 1–6.
Hambly, M. J., de Sousa, A. C. C., Pizzola o, C., Jan. 2024.
Compa ison o il e ing me hods o eal- ime ex ac ion
o he oli ional EMG componen in elec ically s imu-
la ed muscles. Biomedical Signal P ocessing and Con ol
87, 105471.
Lloyd, D. G., Saxby, D. J., Pizzola o, C., Wo sey, M.,
Diamond, L. E., Palipana, D., Bou ne, M., de Sousa,
A. C., Mannan, M. M. N., Nasse i, A., Pe e oshchiko a,
N., Maha aj, J., C ossley, C., Mulholland, K., Collings,
T., Xia, Z., Co nish, B., De ap akash, D., Len on, G.,
Ba e , R. S., Ap . 2023. Main aining soldie mus-
culoskele al heal h using pe sonalised digi al humans,
wea ables and/o compu e ision. Jou nal o Science
and Medicine in Spo .
Pa k, S., Caldwell, G., Umbe ge , B., 2022a. A di ec
colloca ion amewo k o op imal con ol simula ion o
pedaling using OpenSim. PLOS ONE 17 (2), e0264346.
Pa k, S., Umbe ge , B., Caldwell, G., 2022b. A muscle
con ol s a egy o al e pedal o ce di ec ion unde
mul iple cons ain s: A simula ion s udy. Jou nal o
Biomechanics 138, 111114.
Pizzola o, C., Gunduz, M., Palipana, D., Wu, J., G an ,
G., Hall, S., Dennison, R., Za on e, R., Lloyd, D., Teng,
Y., 2021. Non-in asi e app oaches o unc ional eco e y
a e spinal co d inju y: The apeu ic a ge s and mul-
imodal de ice in e en ions. Expe imen al Neu ology
339, 113612.
Quinn, A., Saxby, D. J., Yang, F., C C de Sousa, A.,
Pizzola o, C., Ma . 2023. A digi al win amewo k o
obus con ol o obo ic-biological sys ems. Jou nal o
Biomechanics 152, 111557.
an de Schee , J., Goosey-Tol ey, V., Valen ino, S.,
Da is, G., Ho, C., 2021. Func ional elec ical s imula ion
cycling exe cise a e spinal co d inju y: a sys ema ic
e iew o heal h and i ness- ela ed ou comes. Jou nal
o Neu oEnginee ing and Rehabili a ion 18 (1), 99.
W¨ach e , A., Biegle , L. T., Ma . 2006. On he implemen-
a ion o an in e io -poin il e line-sea ch algo i hm
o la ge-scale nonlinea p og amming. Ma hema ical
P og amming 106 (1), 25–57.
