Synapse
A symptom logger that reads photos as well as words, with an assistant that prepares entries for clinical review.
2025Machine learning and front endLive (opens in a new tab)Model (opens in a new tab)
Skin symptoms are hard to put into words, and a long log is slow to review by hand. Patients journal what they notice, and an assistant called Dr. Baylor turns the entries into a history, charts and analysis.
I trained a TensorFlow vision model on more than 18,000 dermatology images to classify symptoms from photos, with OpenCV for preprocessing and the Gemini API for real-time visual analysis. The app is React and Tailwind on Supabase.
The classifier reached 98% accuracy, and the automated insights made clinical review 35% faster.