Image Embedding

Machine learning models are proven to have superior performance on computer vision applications. At Optum, we developed a family of inventions ranging from representing a single medical claim into an image, to modelling patient history as a series of images (video). These inventions enabled utilizing computer vision ML models for healthcare applications such as FWAE prediction and disease risk prediction. We are currently working on extending these inventions to work on genomics data to produce polygenic risk scores using ML.