Skip to Main content Skip to Navigation

Contribution à la modélisation des durées de séjour du CHU de Grenoble

Cécile Delhumeau 1
TIMC-IMAG - Techniques de l'Ingénierie Médicale et de la Complexité - Informatique, Mathématiques et Applications, Grenoble - UMR 5525
Abstract : This thesis provides a methodology to identify diagnosis related groups (DRG) from Grenoble University Hospital which have a length of stay (LOS) longer than the national reference LOS, and also to identify outliers (patients with extreme length of stay) in these groups. Long LOS are critical because the hospital's rooms tend to be used too much, compared to what the financial score (value) of the corresponding DRG (pathology) would allow. It is therefore important to identify the groups that are responsible for high costs in terms of Indice Synthétique d'Activité (ISA) point. The ISA point is a unit used in the French national program settled to medicalize the information system in hospitals with the aim of optimizing hospital efficiency by measuring their activity and determining the corresponding budget to provide them with (french Prospective Payment System). We compared the differences between the length of stays distribution for the DRG in Grenoble and the LOS distribution for the corresponding DRG in the national database. For each DRG we built a profile by comparing the quartiles of the two distributions. We used statistical classification methods, such as Principal Components Analysis, Hierarchical Clustering, Discriminant Analysis and mixtures models, to identify the groups with highest costs. Empirically 9 different profile types seem to be distinguished. Hierarchical clustering methods identified a classification of DRG in 4 groups. One of them (16 DRG's) counted one third of patients who were highly responsible for costly ISA points, and for which an efficient economical action could be easily undertaken. The mixture models distinguished 3 groups of DRG, and validated the classification derived from the Hierarchical Clustering Method.
Complete list of metadata

Cited literature [70 references]  Display  Hide  Download
Contributor : Thèses Imag <>
Submitted on : Thursday, January 29, 2004 - 5:39:42 PM
Last modification on : Tuesday, June 1, 2021 - 3:44:12 PM
Long-term archiving on: : Wednesday, September 12, 2012 - 12:55:48 PM


  • HAL Id : tel-00004392, version 1


Cécile Delhumeau. Contribution à la modélisation des durées de séjour du CHU de Grenoble. Autre [q-bio.OT]. Université Joseph-Fourier - Grenoble I, 2002. Français. ⟨tel-00004392⟩



Record views


Files downloads