A reinforcement learning approach to establishing a Q&A symptom checker to evolve the performance of the visual diagnosis system for dermatological diseases

The partner is creating artificial intelligence which can help diagnose over 1,300 skin conditions with dermatologist-level accuracy. They are focused on building the functionality that is to be deployed through their app and web-interface that makes it possible to snap a photo, ask questions, and get an instant diagnosis. The partner is very focused on the project given its critical need as part of the overall system/solution. The main objective of the project is to enhance and finalize Triage’s diagnostic system to the point where it is ready to be launched. Their current technology can perform a visual examination of a patient’s skin, with the existing capability limited by the inability to incorporate other context (e.g. symptom data, patient history) from the user for higher accuracy.

Faculty Supervisor:

Amir-massoud Farahmand

Student:

Mohamed Akrout

Partner:

Triage

Discipline:

Computer science

Sector:

Medical devices

University:

University of Toronto