2022/05/13 by Clémentine Courtès, Courtès, C., Emmanuel Franck +11 · 1 citation
Decision Sciences · Mathematics · Psychology · #49M99 93B45 (Primary) 93-10 #92D30 (Secondary) #COVID-19 epidemiological studies #FOS: Mathematics #Mental Health Research Topics #Optimization and Control (math.OC) #demographic modeling and climate adaptation
paper · pdf · doi:10.48550/arxiv.2205.06539
openalex publication_date 2022/05/13 · openalex created_date 2023/02/16 · openalex updated_date 2026/07/28
Modelling epidemics via classical population-based models suffers from\nshortcomings that so-called individual-based models are able to overcome, as\nthey are able to take heterogeneity features into account, such as\nsuper-spreaders, and describe the dynamics involved in small clusters. In\nreturn, such models often involve large graphs which are expensive to simulate\nand difficult to optimize, both in theory and in practice.\n By combining the reinforcement learning philosophy with reduced models, we\npropose a numerical approach to determine optimal health policies for a\nstochastic epidemiological graph-model taking into account super-spreaders.\nMore precisely, we introduce a deterministic reduced population-based model\ninvolving a neural network, and use it to derive optimal health policies\nthrough an optimal control approach. It is meant to faithfully mimic the local\ndynamics of the original, more complex, graph-model. Roughly speaking, this is\nachieved by sequentially training the network until an optimal control strategy\nfor the corresponding reduced model manages to equally well contain the\nepidemic when simulated on the graph-model.\n After describing the practical implementation of this approach, we will\ndiscuss the range of applicability of the reduced model and to what extent the\nestimated control strategies could provide useful qualitative information to\nhealth authorities.\n