At Audensiel, I developed an AI solution for stroke care, designing and optimizing deep learning models for stroke detection from medical images. The models achieved 98% accuracy.
I also developed multimodal predictive models using medical imaging and clinical data for post-stroke risk assessment and 3-month mRS score prediction, achieving 80% precision. I implemented approaches for automatic radiology report generation, evaluated with a METEOR score of 0.4.
During my PhD at University of Western Brittany, LATIM, I developed deep learning models to predict the evolution of complex gait indicators from time-series data and physiological signals. My work spans applied AI, predictive modeling, and healthcare software, including an earlier software development role at Dapsys.
