IA for Health

Project objectives

The RHU SPRINT will involve using Machine Learning methods for early prognosis, to help identify patients with progressive disease, and better tailor their treatment, using omic profiles of skin and blood and validate new biomarkers predictive of monoclonal antibody efficacy, exploiting cutting-edge technologies already developed by our members to analyze in depth the epithelial-mesenchymal transition and its role in disease progression, and accelerating market access for new drug candidates for cutaneous T-cell lymphoma.

Artificial intelligence will be used to:

– determine the optimal low-dimensional, explainable and causal biomarker(s) for “phenotyping” cutaneous T-cell lymphoma patients, i.e. the smallest number of biomarkers that best describe the disease.

– train, improve, validate and test domain-oriented biomarker signatures and associated patient stratification algorithms, taking into account cancer progression.

– estimate the prognostic value and overall predictive performance of cutaneous T-cell lymphoma phenotyping, paving the way for translation into clinical workflow.

Project milestones

Task 2.1 Domain-driven multidimensional biomarker extraction for characterization of CTCL patients (TheraPanacea)

Objective: extract a clinical and histopathological signature to predict progression in CTCL patients.

Task 2.2 Grouping of patient population/phenotyping for progression by group; characterization of progression by group (TheraPanacea)

Objective: determine meaningful characterization of the patient population into patient subgroups with consistent progression patterns for explainable phenotyping of CTCL patients.

Task 2.3 Patient-specific progression models and personalized risk prediction (TheraPanacea)

Objective: provide an explainable scoring model capable of assessing the patient’s current and future stage of disease progression for patient-level stratification.