PREDICTING HOSPITAL LENGTH OF STAY USING REGRESSION MODELS: APPLICATION TO EMERGENCY DEPARTMENT

Catherine Combes (1) , Farid Kadri (2, 3) , Sondès Chaabane (2, 3)

1 LHC - Laboratoire Hubert Curien
2 TEMPO - Laboratoire de Thermique, Ecoulements Mécaniques matériaux mise en forme PrOduction
3 LAMIH - Laboratoire d'Automatique, de Mécanique et d'Informatique industrielles et Humaines - UMR 8201

Increasing healthcare costs motivate the search for ways to increase care efficiency. In this paper, we present a novel methodological framework based on predictive data mining approach to estimate the LOS (Length Of Stay) in an emergency department (ED). We use supervised learning that the goal is to build concise models in terms of predictor features. The aim is to identify the factors (variables) characterizing the LOS in EDs in order to propose models to predict the LOS. We identified two models based on linear regression. These models are validated and were successfully applied to the classification and prediction of the LOS in the pediatric emergency department (PED) at Lille regional hospital centre, France.

Mots clés: Emergency department; healthcare modelling; regressions; LOS forecast

Domaines: Modélisation et simulation

Catherine Combes, Farid Kadri, Sondès Chaabane. PREDICTING HOSPITAL LENGTH OF STAY USING REGRESSION MODELS: APPLICATION TO EMERGENCY DEPARTMENT. 10ème Conférence Francophone de Modélisation, Optimisation et Simulation- MOSIM’14, Nov 2014, Nancy, France. ⟨hal-01081557⟩