Skip to Main content Skip to Navigation
Theses

Modèles statistiques pour les systèmes d'aide à la décision basés sur la réutilisation des données massives en santé : application à la surveillance syndromique en santé publique

Abstract : Over the past few years, the Big Data concept has been widely developed. In order to analyse and explore all this data, it was necessary to develop new methods and technologies. Today, Big Data also exists in the health sector. Hospitals in particular are involved in data production through the adoption of electronic health records. The objective of this thesis was to develop statistical methods reusing these data in order to participate in syndromic surveillance and to provide decision-making support. This study has 4 major axes. First, we showed that hospital Big Data were highly correlated with signals from traditional surveillance networks. Secondly, we showed that hospital data allowed to obtain more accurate estimates in real time than web data, and SVM and Elastic Net models had similar performances. Then, we applied methods developed in United States reusing hospital data, web data (Google and Twitter) and climatic data to predict influenza incidence rates for all French regions up to 2 weeks. Finally, methods developed were applied to the 3-week forecast for cases of gastroenteritis at the national, regional and hospital levels.
Document type :
Theses
Complete list of metadata

Cited literature [203 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02516995
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, March 24, 2020 - 3:27:12 PM
Last modification on : Wednesday, September 14, 2022 - 10:20:04 AM
Long-term archiving on: : Thursday, June 25, 2020 - 2:26:10 PM

File

POIRIER_Canelle.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02516995, version 2

Citation

Canelle Poirier. Modèles statistiques pour les systèmes d'aide à la décision basés sur la réutilisation des données massives en santé : application à la surveillance syndromique en santé publique. Médecine humaine et pathologie. Université Rennes 1, 2019. Français. ⟨NNT : 2019REN1B019⟩. ⟨tel-02516995v2⟩

Share

Metrics

Record views

161

Files downloads

8