← All projects
Data science / AI & machine learning · 2026

cvrie

Can a model spot a stroke on a brain scan, or recognise an illness from how patients describe it? Two medical studies: a classifier that tells normal scans, bleeding and ischemic strokes apart on CT scans (89% accuracy), and an unsupervised model that groups a thousand patient testimonies into 12 pathologies.

Pythonscikit-learnpandasJupyter
Team
2 people
Context
Epitech project, 2nd year
Status
Done
Patient testimonies grouped into 12 topics (NMF, t-SNE)
1 / 4Patient testimonies grouped into 12 topics (NMF, t-SNE)

Results

89.3%validation accuracy, 3-class stroke CT classifier
0.865macro F1 score
12topic clusters from about 1,000 testimonies