The COVID-19 pandemic highlighted the fragility of our healthcare system, and in particular the risk of overload of intensive care units (ICU). ICUs were kept running solely by reorganising their staff and equipment, and in particular by cancelling postponable activities (such as elective surgery). Given the ageing population, the increasing frequency of co-morbidities and their associations, and the risk of pathogen emergence amplified by global change, the pressure on ICUs is expected to increase, both chronically and acutely. An important contribution to anticipating these needs, and to guiding hospital strategy over the long term, consists in modelling the activity of an intensive care unit subject to regular and crisis regimes having in sight the French healthcare population of the next few decades, based on the critical care activity for the last years described beforehand.

The first step will be to use data from the PMSI database (Programme de Médicalisation des Systèmes d'Information) to draw up a national overview of existing critical care activities over the past ten years, including pre-pandemic activity, the COVID-19 health crisis, and post-pandemic dynamics, while analysing inter-regional heterogeneity. In addition, the fraction of the general population likely to be admitted to critical care will be inferred using hidden Markov models. Secondly, the results of these analyses will be cross-referenced with demographic projections, epidemiological data on co-morbidities and scenarios of epidemic outbreaks of emerging and re-emerging pathogens, in order to produce a national projection of critical care needs over the next 30 years. Finally, in a third phase, a virtual ICU will be subjected by simulation to this health pressure, in regular and crisis activity, which will make it possible to identify numerically the weaknesses and potential organisational levers to ensure the future resilience of the national critical care system.

Keywords: Projected epidemiology; demographic forecasts; pandemic preparedness; hospital optimisation; intensive care modelling.