An artificial intelligence system reviewed medical records for a Cleveland Clinic study of more than 7,800 heart valve replacement patients, completing work that would have taken clinicians years, NVIDIA said on Nov. 13.
NVIDIA, whose computing technology powers the system, said the tool supports clinicians but still requires them to confirm its findings and make decisions. The company published the account in its Nov. 13 blog post, “How AI Is Helping Care Teams Get Patients the Right Care, Sooner.”
The company said Cleveland Clinic used Synapsis AI, developed by New Jersey-based clinical AI company Dyania Health, to support research covering patients who underwent aortic valve replacement between 2005 and 2021. The system reviewed their medical records, including heart ultrasound scans added in the years after surgery.
NVIDIA said the platform used that information to build a model predicting valve deterioration and the risk of needing another intervention three, five and 10 years after replacement. The company gave no numerical measure of the model’s accuracy or the time the automated review took.
Dr. Lars Svensson is chief of Cleveland Clinic’s Heart, Vascular & Thoracic Institute. “Using technology to bring together and analyze information that can help us predict the duration of implanted heart valves is a promising step towards enhanced outcomes and more personalized care for patients,” he said.
NVIDIA said Synapsis AI searches notes, reports and laboratory results, connecting findings that may be years apart to answer clinical questions. Its results identify patients who may need attention and link each finding to the underlying medical record, giving care teams evidence to review before they act.
The company said hospitals hold decades of patient histories, much of them in written notes, surgical reports, imaging interpretations and discharge summaries. Reading a patient’s full history can therefore require reviewing years of charts, work the company said the system can take off clinical teams’ desks.
NVIDIA said Synapsis AI processes records on a hospital’s own servers, behind its network security controls, so patient information stays onsite. Dyania uses NVIDIA software to build and train its clinical models, then uses other company tools to make them run faster on hospital computing equipment.
The company said the approach could also help identify patients eligible for clinical trials, find gaps in cardiology and cancer care, and fill clinical registries, databases that collect information about patients and treatments. Those uses would still leave clinicians responsible for confirming the evidence and deciding what care patients receive.
