Cognitive Maps Exploration trough Kernel Density Estimation

Antoine Lhuillier (1) , Christophe Hurter (1, 2) , Hélène Amieva (3) , Emmanuel J Barbeau (4) , Christophe Jouffrais (2)

1 LII - ENAC - Equipe Informatique Interactive
2 IRIT-ELIPSE - Etude de L’Interaction Personne SystèmE
3 BPH - Bordeaux population health
4 CERCO - Centre de recherche cerveau et cognition

Currently approximately 860,000 people are affected by Alzheimer’s disease in France. This is why the study of Alzheimer’s disease has been identified as a major societal challenge. In the PAQUID cohort study, subjects performed a lexical evocation task by saying the maximum number of city names within 3 minutes. This task is directly related to the concept of cognitive map. The analysis of the list created by this task provides a unique opportunity to study the spatial mental representation of geographical space for elderly people before and after developing the dementia. We visualized graphs of this list by connecting cited city with a line. Since these graphs become cluttered with numerous lines, we applied the KDEEB bundling technique. We then compared graphs representing different periods before and after dementia. Our first results show that graph complexity is related to aging, as well as the clinical status of the subject. The ultimate goal of the project is to develop techniques and tools to study the cognitive maps of elderly subjects in the years preceding Alzheimer’s disease. The tools that will be developed shall rely on image based technique (e.g. bundling) and help to better detect and understand the evolution of Alzheimer’s disease.

Mots clés: Alzheimer ' s disease; PAQUID; Cognitive maps; Bundling; Large data visualization

Domaines: Informatique [cs]; Interface homme-machine [cs.HC]

Antoine Lhuillier, Christophe Hurter, Hélène Amieva, Emmanuel J Barbeau, Christophe Jouffrais. Cognitive Maps Exploration trough Kernel Density Estimation. IEEE VIS Workshop Visualizing Electronic Health Record Data (EHRVis 2014), University of Maryland; Vienna University of Technology, Nov 2014, Paris, France. ⟨hal-01522067⟩