Development of a classification system for assessing apathy’s degree in patients with behavioral variant of frontotemporal dementia

Apathy is a behavioral symptom present in many neurological and psychiatric pathologies and it’s defined as a quantitative reduction of voluntary, goal-directed behaviors. Today, apathy is assessed by questionnaires administered to the patient and/or caregiver and providing score, which compared to a threshold (norms), is able to determine the degree of apathy (clinical evaluation). However, this method of evaluating apathy is biased by the subjective nature of the patient or caregiver’s perspective. Most of time, the patient is unaware of his or her real disorder (anosognosia). The objective of this thesis entitled "DEVELOPMENT OF A CLASSIFICATION SYSTEM FOR ASSESSING APATHY'S DEGREE IN PATIENTS WITH BEHAVIOURAL VARIANT OF FRONTOTEMPORAL DEMENTIA" is to address the specific question concerning the limitations of the assessment of neuropsychiatric symptoms (NPS) using interviews and rating scales, carrying out an objective and quantitative assessment of apathy. This thesis contributes to the ECOCAPTURE research program, whose general objective is to improve the characterization and assessment of apathy using behavioral sensing under ecological environments. We developed a classification system applied on a dataset composed of 12 healthy controls (HC) and 14 patients with behavioral variant of frontotemporal dementia (bvFTD). The system follows these collected data: kinematic 3D-accelerometer data, video-based behavioral data, eye-tracking glasses data, neuropsychological and MRI data, in order to identify a behavioral signature of apathy. Data were obtained from bvFTD patients and HC enrolled in the ECOCAPTURE clinical study (https://clinicaltrials.gov/ct2/show/NCT03272230). In order to classify and predict the form and intensity of apathy in bvFTD subject, based on the whole dataset, we used firstly, a clustering approach and secondly, a regression approach based on MDMR (multivariate distance matrix regression). We obtained a classification of the dataset into three different clusters, one for the healthy volunteers (HC) and two different for the bvFTD patients (FTDa, FTDb). Moreover, we demonstrated the possibility to estimate a prediction of the apathy level based on the ECOCAPTURE dataset.