Sojourn time estimation in partially observed piecewise deterministic Markov processes - application to myeloma modeling
Amélie Vernay  1, *@  , Benoîte De Saporta  1, *@  , Alice Cleynen  1, 2, *@  
1 : Institut Montpelliérain Alexander Grothendieck
Centre National de la Recherche Scientifique, Université de Montpellier
2 : John Curtin School of Medical Research [Canberra, Australia]
* : Auteur correspondant

We consider a problem of relapse time estimation in Piecewise Deterministic Markov Processes (PDMPs) whose Euclidean component is a proxy biomarker for the status of myeloma patients. One of the main difficulties of the problem lies in the fact that our process is only partially observed, which few works have considered until now. We provide a method based on iterative regression for estimating the parameters of a PDMP observed in discrete time and through noise. We assess the performances of our procedure through a simulation study and discuss the limitations of our approach.



  • Poster
Personnes connectées : 4 Vie privée
Chargement...