Stochastic simulation of clinical pathways from raw health databases

Martin Prodel (1) , Vincent Augusto (2) , Xiaolan Xie (2) , Baptiste Jouaneton (3) , Ludovic Lamarsalle (3)

1 I4S-ENSMSE - Ingénierie des systèmes de soins et des services de santé
2 CIS-ENSMSE - Centre Ingénierie et Santé
3 HEVA [Lyon]

This paper presents a method to automatically create stochastic simulation models of clinical pathways from raw databases. We introduce an automatic procedure to convert a process model, discovered with process mining, into an actionable simulation model. The concept of state charts is used and enriched to incorporate the distinctive features of healthcare processes into the model. The clinical pathway model is used to simulate new patients' sequence of events. The resulting model is validated by comparing key performances indicators with historical data. Finally, we use the model to perform an automatically setup sensitivity analysis. The whole process is automated and can be used with any input data.

Mots clés: Computational modeling; Databases; Data models; Analytical models; Stochastic processes; Data mining; Hospitals

Domaines: Informatique [cs]; Mathématiques [math]

Martin Prodel, Vincent Augusto, Xiaolan Xie, Baptiste Jouaneton, Ludovic Lamarsalle. Stochastic simulation of clinical pathways from raw health databases. 2017 13th IEEE Conference on Automation Science and Engineering (CASE 2017), Aug 2017, Xi'an, China. ⟨10.1109/COASE.2017.8256167⟩. ⟨hal-01860734⟩