Selfie-Care Bringing AI to Skin Disease Analysis

What’s happening with Selfie-Care?

BEVERLY HILLS, CA ( 2020/12/29

Selfie-Care is a mobile app that allows the user to snap a photo of a skin disease, have it analyzed by Selfie-Care AI on the cloud, and receive a referral to the right type of local specialist skin doctor for the type of disease the AI has matched.

Psoriasis has many different types and according to Novartis the typical psoriasis patient waits two year from first diagnosis by a doctor to finding the right specialist. This delay frustrates patients and doctors. Skin specialists often say, I wish you’d come to me sooner.

My team won the Novartis Biome prize during the Mobile World hackathon. A quick proof-of-concept demo using generic Microsoft AI SaaS and a PowerPoint presentation. Not an MVP, a quick throw-away demo in the Agile spirit of build-one-to-throw-away.

Selfie-Care Progress

Currently building app, cloud AI server and website. The app in C++ Qt. For the cloud server AI platform, an idea is to use GPT-2/Tensorflow. I’m already a GPT-2 fork maintainer. However, Python (GPT-2) may not scale well on servers and Tensorflow is so complex it is hard to maintain.

We had a meeting to brainstorm Selfie-Care AI ideas. Now researching open source C/C++ platforms that are alternatives to Tensorflow, such as mxnet and opencv.

Interest in Selfie-Care includes NSF, the Mayo Clinic and many others who want to see our new technology in action.