Structure estimation of binary graphical models on stratified data: Application to the description of injury tables for victims of road accidents

摘要

Graphical models are used in many applications such as medical diagnostic,computer security, etc. More and more often, the estimation of such models hasto be performed on several predefined strata of the whole population. Forinstance, in epidemiology and clinical research, strata are often definedaccording to age, gender, treatment or disease type, etc. In this article, wepropose new approaches aimed at estimating binary graphical models on suchstrata. Our approaches are obtained by combining well-known methods whenestimating one single binary graphical model, with penalties encouragingstructured sparsity, and which have recently been shown appropriate whendealing with stratified data. Empirical comparions on synthetic data highlightthat our approaches generally outperform the competitors we considered. Anapplication is provided where we study associations among injuries suffered byvictims of road accidents according to road user type.