The deployment addresses the primary challenge that has traditionally prevented health systems from automating clinical chart review: the unstructured nature of complex clinical narratives. Traditional natural language processing tools often fail because they lack longitudinal context or produce unreferenced outputs that require clinicians to re-read the entire chart to verify accuracy.
Layer Health’s platform approaches the chart through a human-in-the-loop workflow that mirrors how experienced clinical abstractors operate:
The software reviews the complete longitudinal patient record across both structured EHR fields and unstructured notes, extracting required clinical registry elements while generating direct, source-linked evidence citations for every suggested answer. Instead of hunting through hundreds of pages across disparate encounters, abstractors are presented with pre-populated registry fields paired with highlighted source passages within the original clinical documentation. This allows clinical abstractors to validate or adjust extracted data points rapidly while keeping final clinical discretion in human hands.
Following internal validation against Fairview’s historical clinical data, the implementation will scale outward from stroke care into the health system’s highest-volume registries: the Commission on Cancer’s National Cancer Database (NCDB) and the American College of Cardiology’s NCDR registries. Across peer health systems, the platform has demonstrated greater than 70 percent time savings per abstracted chart while maintaining over 98 percent accuracy against audited registry benchmarks.
David Sontag, Chief Executive Officer and Co-Founder of Layer Health, framed the operational shift: “Fairview is approaching the work of clinical registries in a way that aligns with the most innovative quality teams in the country. Our platform brings clinically intelligent machine learning to a workflow that mirrors how abstractors already work, so these teams can spend less time searching for answers and more time moving quality improvement initiatives forward.
Source link
