Discovering Associations in Clinical Data: Application to Search for Prognostic Factors in Hodgkin's Disease

Nicolas Durand (1) , Bruno Crémilleux (1) , Michel Henry-Amar (2)

1 Equipe CODAG - Laboratoire GREYC - UMR6072
2 Calvados Cancer Registry

The production of suitable clusters to help physicians explore data and take decisions is a hard task. This paper addresses this question and proposes a new method to define clusters of patients which takes advantage of the power of association rules method. We present different notions of association and we specify the notion of frequent almost closed itemset which is the most appropriate for applications in the medical area. Applied to Hodgkin's disease to help establish prognostic groups, the first results bring out some parameters for which classical statistic methods confirm that they are interesting.

Domaines: Apprentissage [cs.LG]; Ingénierie biomédicale; Cancer

Nicolas Durand, Bruno Crémilleux, Michel Henry-Amar. Discovering Associations in Clinical Data: Application to Search for Prognostic Factors in Hodgkin's Disease. 8th Conference on Artificial Intelligence in Medicine in Europe (AIME 2001), Jul 2001, Cascais, Portugal. pp.50-54, ⟨10.1007/3-540-48229-6_6⟩. ⟨hal-01463174⟩