AI & Biomarkers: optimising patient selection

Context and Challenges

New personalized therapeutic approaches require the integration of multiple complex data sources: liquid biopsy, medical imaging, and molecular and immunological analyses. AI represents a promising solution for processing this information and facilitating clinical decision-making.

Project Objectives

• Develop and validate AI algorithms for advanced analysis of liquid biopsy data, including the identification of predictive and prognostic biomarkers.
• Integrate liquid biopsy biomarkers with clinical, radiological, and molecular data to enhance the selection of candidates for innovative immunotherapies.
• Propose a reliable and accessible predictive model, enabling rapid and personalized decision-making in lung cancer therapeutic management.

Methodology

• Use of advanced AI techniques (machine learning, clustering, parsimonious regression) to identify and validate optimal biomarker combinations.
• Normalization of liquid biopsy data via a federated learning infrastructure to ensure reproducibility and robustness of results across multiple centers.
• Simultaneous exploitation of medical imaging, biological, and clinical data to strengthen algorithm precision.

Expected Results

• Clinical identification and validation of new liquid biopsy biomarkers capable of predicting the efficacy of immunotherapies and other innovative treatments.
• Deployment of intelligent decision-making tools integrated into daily clinical practice to improve care and prognosis for patients with respiratory diseases.

Involved Institutions

• Laboratory of Clinical and Experimental Pathology – Nice University Hospital
• INRIA

Funding: Calls for projects – "Santé Innovation et Exceptionnel" edition "Albert CALMETTE", Alpes-Maritimes Department.