adiaion Theapy Teamen Decision Making fo Posae Cance Paiens Based on PSA Dynamics Maiel S. Laiei Main L. Pueman Sco Tyldesley Seen Sheche
Ouline Backgound Infomaion Model Descipion Nex Seps Moiaion Tadeoffs Cuen poocol Ou goal The model
Moiaion High isk posae cance paiens migh be offeed neoadjuan homone heapy o induce posae cance cell deah and umo egession pio o hei adioheapy eamen Paiens ae monioed peiodically Maximal umo egession pobably occus when PSA eaches is nadi leel M. S Gleae S. La Bianca S. L. Goldebeg 000 No known when such leel will be achieed
Tadeoffs Toleance esponsieness Why decide o ea? - Aoid pogession - isk of cells becoming esisan - Toxiciy of homone heapy Why wai? - Maximum educion of umo size unde homone heapy - Moe infomaion
Cuen Poocol High-inemediae o high isk paien? Offe neoadjuan homone heapy pio o adiaion heapy Sa adiaion heapy if: 8 monhs of homone heapy hae been eceied o Nadi PSA 0.05 is eached afe 4 monhs o PSA leels sa o ise o PSA < ng/ml is no achieed afe 4 monhs
Ou Goal Impoe modeling of PSA kineics and esimaion of fuue PSA kineics Poide a fomal decision making ool o deemine when a paien should begin adiaion heapy eamen
Modeling PSA ln PSA α d ln PSA 0 d ~ N0 V PSA S s s imee Expeced minimum nadi PSA nadi Time
Timeline Obsee PSA Updae cue paamees Don Sa Sa T? eceie eamen Obsee PSA Updae cue paamees ζ P ime of nadi ζ Sae Updaing Kalman Fileing
Sa T Don Sa T Decision Tee Pob nadi is wihin ζ days Sa T New PSA Don Sa T
006 4 0 0 PSA 0 0. PSA s ime 8 6 4 0 30 60 90 0 50 80 0 40 0 30 60 90 0 50 80 0 40 Time fom NAH sa days minimum
Model Model Discee ime finie hoizon MDP Acion: Sae: ; Sa adiaion heapy Don sa adiaion heapy Paamee means PSA : Obseed PSA Paamee coaiances α 0 ~ ln V N PSA Y α N α α ; ~
Model Maximize Pobabiliy of Teaing wihin ζ of he Nadi Sa T P nadi ζ ; max P Don Sa T
Sa T - Disibuion of Time of Nadi nadi -/ ~ N ; whee 3 P nadi ζ ; max P 3 Disibuion of ime of nadi 60% elaie fequency 50% 40% 30% 0% 0% 0% 30 90 50 0 70 0 Time of nadi 390 450 50 570 630 690 >70
; max nadi P P ζ Sa T Sa T - Disibuion of Time of Nadi Disibuion of Time of Nadi du e F u nadi 3 4 π du e u 3 3 4 4 ζ ζ ζ π ζ P nadi Hinkley969 nadi -/ whee As P >0 N 3 3 ; ~ α 3 3 3 3
; max nadi P P ζ Don Don Sa T Sa T Sae Updaing Sae Updaing Y P Y P V N Y 4 3 3 3 ~ α F Q F F Y Q F ' ' ] [ Sae equaions: Updaing equaions: α 3 3 3 3 F V Q 4 3 3 3 Gien Y
Model Model Maximize Pobabiliy of Teaing Maximize Pobabiliy of Teaing wihin wihin ζ of he Nadi of he Nadi ; max nadi P P ζ α 3 3 3 3 F V Q 4 3 3 3 ' ~ Q F N Y du e u 3 3 4 4 ζ ζ ζ π Sa T [ ] Y Y dp F Q F F Y Q F ' ' Don Sa T
Nex Seps Appoximae soluion - Bayesian inegaions - Simulaion Sucue of policies Conside ohe possible objecies: - Minimize cos - Minimize disance of nadi obseed o nadi expeced - Maximize suial Addiional quesions: - When o ake nex eading? - When o change homone eamen? - Is 8 monhs a good endpoin? Model Validaion Design clinical sudy
Thank you Maiel Laiei Maiel.Laiei@saude.ubc.ca www.chcm.ubc.ca/ciheam