Temporal models of care sequences for the exploration of medico-administrative data

Johanne Bakalara (1) , Thomas Guyet (2) , Olivier Dameron (3) , Emmanuel Oger (4) , André Happe (1)

1 REPERES - Recherche en Pharmaco-épidémiologie et Recours aux Soins
2 LACODAM - Large Scale Collaborative Data Mining
3 Dyliss - Dynamics, Logics and Inference for biological Systems and Sequences
4 Centre Hospitalier Universitaire de Rennes [CHU Rennes] = Rennes University Hospital [Ponchaillou]

Pharmacoepidemiology with medico-administrative databases enables to study impact of health products in real-life setting. These studies require to manipulate the raw data, the care trajectories, in order to identify pieces of data that may witness the medical information that is looked for. The manipulation can be seen as a querying process in which a query is a description of a medical pattern (e.g. occurrence of illness) with the available raw features from care trajectories (e.g. occurrence of medical procedures, drug deliveries, etc.). The more expressive is the querying process, the more accurate is the medical pattern search. The temporal dimension of care trajectories is a potential information that may improve the description of medical patterns. The objective of this work is to propose a formal framework that would design a wellfounded tool for querying care trajectories with temporal medical patterns. In this preliminary work, we present the problematic and we introduce a use case which illustrates the comparison of several querying formalisms.

Mots clés: Description logic; Ontologies; Pharmacoepidemiology; Temporal logics; SNDS

Domaines: Intelligence artificielle [cs.AI]; Base de données [cs.DB]; Logique en informatique [cs.LO]; Logique [math.LO]

Johanne Bakalara, Thomas Guyet, Olivier Dameron, Emmanuel Oger, André Happe. Temporal models of care sequences for the exploration of medico-administrative data. 2019 - Workshop IA&Santé, PFIA, Jul 2019, Toulouse, France. pp.1-8. ⟨hal-02265743⟩