Classification automatique de séries chronologiques de patients souffrant de douleurs chroniques - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Automatic classification of time series of patients with chronic pain

Classification automatique de séries chronologiques de patients souffrant de douleurs chroniques

Résumé

The most popular unsupervised classification algorithms allow the identification of hard or probabilistic partitions. However, with complex data sets such as those in the health domain, these partitions have their limits. Indeed, they do not allow to model atypical or imprecise observations. This study aims to analyzing sequential data of patients with from chronic pain. A first step is to extract characteristics from these time series and then selecting the most important attributes. A second step, consists to use the evidential c-means (ECM) clustering algorithm on the extracted attributes. The ECM method generates a credal partition that has the ability to model many forms of uncertainty. This partition can then be transformed into a hard partition in order to study individuals according to this uncertainty criterion. The explicability of the hard partition identified through descriptive analysis and statistical tests allowed us to identify the profile of these chronic pain patients.
Fichier principal
Vignette du fichier
1002881.pdf (124.66 Ko) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-04063246 , version 1 (12-04-2023)

Identifiants

  • HAL Id : hal-04063246 , version 1

Citer

Armel Soubeiga, Jessem Ettaghouti, Violaine Antoine, Alice Corteval, Nicolas Kerckhove, et al.. Classification automatique de séries chronologiques de patients souffrant de douleurs chroniques. Extraction et Gestion des Connaissances (EGC) - 2023, Jan 2023, Lyon, France. pp.651-652. ⟨hal-04063246⟩
44 Consultations
15 Téléchargements

Partager

Gmail Facebook X LinkedIn More