ESTRO37
Pos e
p esen ed a :
P ojec S32: decision suppo sys em o lung
cance pa ien s
J.L. López-Gue a1, Bea iz Pon es2, A. Mo eno3, C is ina Rubio2, F.J. Núñez3, Isabel Nepomuceno2, J. Mo eno3, J. Cacicedo4, J.M. P aena-Fe nandez5, Ge man An onio Escoba Rod iguez3, C.
Pa a3, Jose Riquelme2, M.J. O iz-Go dillo1.
1Uni e si y Hospi al Vi gen del Rocio, Radia ion Oncology, Se illa, Spain. 2Depa men o Compu e Language and Sys ems, Uni e sidad de Se illa, Se ille, Spain. 3Uni e si y Hospi al Vi gen del
Rocio, G oup o Technological Inno a ion, Se illa, Spain. 4C uces Uni e si y Hospi al, Radia ion Oncology, Bilbao, Spain. 5Uni e si y Hospi al Vi gen del Rocio, Me hodology Uni , Se illa, Spain.
Pu pose and Objec i e
•A decision suppo sys em (DSS) has been p oposed o p edic su i al
and apply knowledge om ou ine ca e da a a he han solely elying on
clinical guidelines in lung cance (LC) pa ien s.
Ma e ial and me hods
Resul s
•Ou DSS success ully handled a high numbe o he e ogeneous a iables, demons a ing po en ial o enhancing p edic ion o su i al.
•The DSS could assis physicians in o mula ing an e idence-based managemen ad ice in pa ien s wi h LC.
•This DSS migh be used in a clinic as an objec i e guide o indi idualize ea men , and discussions wi h pa ien s, acco ding o
p ognosis.
Conclusions
•To implemen he echnological a chi ec u e o his DSS, we in eg a ed a
se o open sou ce ools which allowed us o egis e in o ma ion du ing
daily clinical p ac ice h ough elec onic heal h eco ds and use his
in o ma ion o au oma ically execu e di e en Da a Mining analyses.
•I is based on he XGBoos and Gene alized Linea Models algo i hms
applying a 10- old c oss alida ion o explo e i s po en ial o p edic ing
su i al om a he e ogeneous da ase .
•P ospec i e mul icen e da a om 543 consecu i e LC pa ien s ha we e
seen in consul a ion in he adia ion oncology depa men s om Janua y
2013 o July 2017 we e a ailable o enable he de elopmen o he
p edic ion model.
•The e we e 229 (42%) ali e pa ien s and 314 dea hs a he ime o he
s udy. The da a se had mo e han 400 i ems bu only a p opo ion o
hem (including, among o he s, age, gende , his ology, pe o mance
s a us, s age, and ea men app oach) wi h disc imina o y abili y
acco ding he algo i hms we e used.
•Di e en ime´s pe iods (p e- ea men , ea men ) we e assessed o
p edic ion. Addi ionally, a subse o pa ien s wi h a minimum ollow-up o
18 mon hs o ali e pa ien s was also assessed.
•A ea unde he ecei e -ope a ing cha ac e is ics cu e (AUC) measu ed
pe o mance. The esul s we e compa ed wi h he AUC ob ained using
he basic i ems included in he guidelines (p e ea men da a [s age,
his ology] and ea men da a [ adio he apy, su ge y, and sys emic
he apy]).
•The compa ison o he di e en AUCs ob ained om 10 simula ions o
each ule was done using he one-way analysis o a iance (ANOVA),
ob aining hei con idence in e als a 95%. Mul iple compa isons, when
he hypo hesis o homoscedas ici y was no e i ied, we e pe o med by
he Games-Howell co ec ion.
Da a
P edic i e model o mo ali y
Using da a mining Using guidelines
Lung cance pa ien s
All pa ien s
(N=543)
Pa ien s wi h a minimum ollow
-
up o 18
mon hs o ali e pa ien s
(N=451)
All pa ien s
(N=543)
Pa ien s wi h a minimum ollow-
up o
18 mon hs o ali e pa ien s
(N=451)
N* AUC N* AUC N* AUC N* AUC
Using
p e- ea men da a 20 0.84 (0.77-0.90) 6 0.74 (0.69-0.79) 2
0.60 (0.56
-
0.64)
2 0.64 (0.58-0.71)
Using
only ea men da a 9 0.78 (0.72-0.84) 10 0.81 (0.78-0.84) 3
0.60 (0.56
-
0.65)
3 0.65 (0.58-0.72)
Using
all da a 35 0.88 (0.83-0.92) 34 0.80 (0.77-0.83) 5
0.63 (0.58
-
0.68)
5 0.67 (0.60-0.75)
Non-small cell lung cance
(N=405) (N=343) (N=405) (N=343)
Using
p e- ea men da a 20 0.79 (0.72-0.85) 20 0.70 (0.64-0.76) 2
0.57 (0.51
-
0.62)
2 0.58 (0.50-0.67)
Using
only ea men da a 24 0.77 (0.72-0.82) 25 0.78 (0.73-0.84) 3
0.63 (0.56
-
0.70)
3 0.66 (0.57-0.75)
Using
all da a 17 0.81 (0.80-0.83) 32 0.77 (0.71-0.85) 5
0.64 (0.60
-
0.71)
5 0.66 (0.59-0.74)
Small cell lung cance
(N=138) (N=108) (N=138) (N=108)
Using
p e- ea men da a 19 0.82 (0.74-0.91) 20 0.73 (0.59-0.87) 2
0.67 (0.52
-
0.81)
2 0.74 (0.61-0.87)
Using
only ea men da a 22 0.76 (0.67-0.84) 23 0.90 (0.83-0.97) 3
0.42 (0.34
-
0.50)
3 0.47 (0.38-0.56)
Using
all da a 33 0.92 (0.86-0.98) 34 0.96 (0.92-0.99) 5
0.61 (0.54
-
0.68)
5 0.67 (0.58-0.77)
*Numbe o a iables wi h disc imina o y abili y acco ding he algo i hms
*Numbe o a iables acco ding guidelines
Table. AUC( mean and 95% CI ) o p edic ing su i al using ei he da a mining
analyses o basic i ems included in he guidelines in lung cance pa ien s.
Compa ison o he AUCs and 95% CI o p edic ing su i al
in all lung cance pa ien s when using da a mining analyses
s he guidelines.
Compa ison o he AUCs and 95% CI o
p edic ing su i al in non-small cell lung
cance pa ien s when using da a mining
analyses s he guidelines.
Compa ison o he AUCs and 95% CI o
p edic ing su i al in small cell lung cance
pa ien s when using da a mining analyses
s he guidelines.
PO-0859
Jose Luis Lopez Gue a DOI: 10.3252/pso.eu.ESTRO37.2018
Clinical ack: Heal h se ices esea ch / heal h economics