Description
Date depot: 1 janvier 1900
Titre: Automatic diagnosis and prognosis systems built on the statistical exploitation of longitudinal medical data sets
Directeur de thèse:
Didier DORMONT (ICM)
Directeur de thèse:
Stanley DURRLEMAN (ICM)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
{{Keywords:}}
statistical learning, medical data, longitudinal data sets, personalization, classification, prediction, disease biomarkers, e-health.
{{Project:}}
Longitudinal medical data consist in the repeated observations of patients at multiple points in time. This PhD project aims to develop a statistical framework to automatically learn typical spatiotemporal patterns in such data sets. Such a learning method should open up the possibility to make accurate prediction about the future state of the patients based on their previous observations. These predictive systems have important applications to derive automatic prognosis system, assess treatment efficacy, personalize treatments and optimize the overall healthcare systems.
For instance, being able to robustly predict clinical symptoms only few months before their onset is a crucial public health challenge, which will open up the possibility to test treatments at the stage when they have the highest chance of success. The identification of pathological subtypes will allow a better selection of patients in clinical trials, with important applications in the pharmaceutical sector. Eventually, such predictive systems might be used to make recommendation to the clinicians about exams and treatments to be undertaken in the future, therefore optimizing the healthcare system with important economical and societal consequences.
The predictive systems will be evaluated on various databases of patients with neurodegenerative diseases, notably the data set of the Alzheimer’s Disease NeuroImaging Initiative (ADNI) and the data sets with patients with fronto-temporal dementia or amyotrophic lateral sclerosis. We will also evaluate our system to monitor the patients’ care pathway involving multiple visits at the hospital.
Doctorant.e: Ansart Manon