Hospital management of an epidemic in intensive care: a predictive tool for saturation

Abstract The COVID-19 epidemic has highlighted the importance of hospital resource management in exceptional health contexts. The hospital capacity of intensive care units has emerged as a determining factor in overall hospital management. The objective of this study is to create a tool for predicting the saturation status of an intensive care unit. The predictive model was developed based on a SIR compartmental model. The model was trained using hospitalization data (admissions, discharges, and capacity) from the Raymond Poincaré Hospital, AP-HP. The model parameters (transmission and recovery rates) were estimated using the least squares method. A web application, OHASiS® (Outil Hospitalier d'Analyse de Situations de Saturations), was developed using Shiny and embeds the model. The model provides a reliable prediction over 7 days of the evolution of the number of admissions and discharges in COVID-19 intensive care units. Information on capacity makes it possible to predict a possible saturation of the service. A sensitivity analysis of the model revealed no significant difference between predicted and observed attendance data. The OHASiS® tool allows for the visualization of epidemiological data and helps with medical decision-making as the epidemic evolves. It would be desirable to integrate this application into future hospital management plans. This tool could also be used during simulation exercises. Validation of the predictive model in the context of seasonal epidemics still needs to be conducted. Key messages • A SIR compartmental model was used to develop a tool for predicting the saturation status of ICU. • The web application OHASiS® can help with medical decision-making and should be integrated into future hospital management plans, though further validation is needed for use during seasonal epidemics.