My work focuses on developing explainable machine learning methods applied to complex medical data. The thesis resulted in the creation of an interpretable neural network and its theoretical connection to kernel logistic regression, while my postdoctoral research concentrates on leveraging longitudinal data from the Système National des Données de Santé (SNDS) to improve decision-making in oncology.

• Interpretable Neural Network and Kernel Logistic Regression
  Development of SATURNN, an explainable neural network based on additive splines, with a reformulation as kernel logistic regression (EKLR). This approach guarantees interpretability and convergence, addressing the needs of precision medicine. [More details]
• Representation Learning on SNDS Data
  My postdoctoral research focused on analyzing SNDS data, including evaluating representation learning methods. I also currently work on a simulated realistic breast cancer SNDS-like database and develop a Python package (pySNDS) to facilitate the use of SNDS data for research purposes. [More details]

• Presentation at Journée Scientifique Laënnec (15/05/2025, Marseille): IA4Elderly: Exploitation des données du SNDS pour la représentation des parcours de soins des personnes âgées atteintes d’un cancer. [Slides]
• Presentation at Atelier IACD – Extraction et Gestion des Connaissances (January 2025, Strasbourg): Representation Learning pour la codification des parcours thérapeutiques de patientes atteintes de cancer du sein à partir de données de remboursement : un benchmark pour des tâches de clustering. [Slides]
• Publication in 2024 IEEE 37th International Symposium on Computer-Based Medical Systems (CBMS): Encoding breast cancer patients’ medical pathways from reimbursement data using representation learning: a benchmark for clustering tasks. [Article]