Layer Well being, an organization spun out of MIT, says its first product, Distill, makes use of synthetic intelligence to shortly carry out any medical, administrative, or analysis process that requires chart assessment from unstructured knowledge. Use instances embrace registry submissions, high quality measurement, curation of real-world proof, medical doc enchancment, and income cycle administration.
The startup is backed by $4 million in funding from GV (Google Ventures), Basic Catalyst, and Inception Well being, and amongst its beta clients is Froedtert & the Medical School of Wisconsin well being community, which is utilizing Distill to help high quality enchancment efforts. The platform is being utilized by the group’s nurse abstraction staff throughout chart assessment, serving to them extra successfully discover and submit knowledge to medical registries.
Layer Well being says that Distill integrates into present merchandise and workflows, ingesting medical notes and analyzing them at scale. “Behind the scenes, Layer Well being’s machine studying (ML) algorithms leverage the facility of huge language fashions (LLMs) to ship correct outcomes with out the necessity for labeled knowledge, decreasing growth time from months to as little as a day. Distill additionally learns and adapts from buyer interactions, creating extremely environment friendly customer-specific fashions fine-tuned for particular use instances,” the corporate says.
One other person the corporate cites is xCures, a well being know-how firm that’s utilizing Distill to arrange and construction well being knowledge for extra exact most cancers remedy suggestions and environment friendly medical trial matching. Utilizing Layer’s language fashions educated and validated on xCures’ distinctive knowledge, the corporate can extra precisely extract intricate and nuanced particulars from affected person medical information.
Layer Well being’s CEO is David Sontag, Ph.D., an MIT professor with greater than 100 revealed papers in AI and machine studying.
