AI support system for authoring clinical documentation

Year 2021
Project team David Karger & David Sontag with Steven Horng, Monica Agrawal, and Luke Murray
Time-consuming records Contribute to clinical errors and physician burnout
Machine learning Combined with human-computer interaction to synthesize knowledge
Seamless interface Reimagines documentation as part of clinical reasoning

Quality medical records

Electronic Health Record (EHR) systems hold potential to aid clinical diagnosis, operations and research. However, their data is often unstructured, and the systems are time-consuming and difficult to work with. These problems contribute to clinical errors and physician burnout. This team is developing MedKnowts, a system combining machine learning and human-computer interaction to reduce the effort needed both to synthesize knowledge for medical decisions and to create quality structured records. MedKnowts provides a seamless interface that reimagines documentation as part of clinical reasoning rather than a compliance requirement. The result will be faster documentation, a lower search burden, and higher quality notes for billing, compliance and communication.

Layer Health

Technology from this project spun out into a startup company, Layer Health

Luke Murray presents “ML-Driven Reimagination of Clinical Note-Taking” at IdeaStream 2022.

Large language models help decipher clinical notes