Simulation et apprentissage Monte-Carlo de stratégies d'intervention en santé publique
|Advisor:||Gagné, Christian; Reinharz, Daniel|
|Abstract:||Decision makers in public health system, such as the one in the province of Quebec, have a growing need for assessment tools to support their decisions on the interventions to implement. This master’s thesis proposes a generic simulator optimized for public health issues, while being extensible to other areas. It details the software architecture and all the features that make it a tool of choice for decision makers. It also presents the optimization of existing intervention strategies using Monte Carlo reinforcement learning. This includes the proposal of a new algorithm for selecting actions when learning on populations of individuals evolving in parallel. We conclude with the application of this infrastructure to two public health issues : diabetic retinopathy, that has already been the subject of work by other researchers, and osteoporosis, a current application that has been validated by health care specialists.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||18 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.