Google AI, Daniel Tse, medical image MD
Improving healthcare with AI
Google mission
Make products more useful
Help others innovate
Solve for humanity's big challenge
Google translate
Smart compose in gmail ( to finish a sentence by AI)
Google photos (to search by photo)
Tensorflow is a tool for AI.
Her min textile (Taiwanese company) is a good example.
Energy consumption
Environmental conservation
Healthcare
Why use ML for healthcare?
Lots of data to look thru.
Expertise is limited. (ex. radiologists 1:12,500 people. But there are fewer.)
Diabetic
Regular screening of eyes is key to prevent blindness.
How DR is diagnosed? Retinal fundus images.
India has a shortage of 127,000 of eye doctors. 45% of patients suffer of lost vision.
Adapt deep neutral network to read fundus images. 880K diagnoses are done.
What is next?
Clinical trials and regulatory
Partnership with hardware: Nikon, optos, V
Completely new novel scientific discoveries
Can we predict the cardiovascular risk by eye images?
Breast cancer prediction:
10Gpx per image to look into the breast
1 in 12 breast cancers are misdiagnosed
Pathologist + algorithm
Google AR microscope
Detecting skin cancer with AI (Stanford)
沒有留言:
張貼留言