NVIDIA and Dyania Health have partnered to deploy AI-powered clinical analysis within healthcare networks, allowing institutions to derive insights from full patient histories. Dyania’s Synapsis AI platform examines complete electronic health records—including years of progress notes, imaging, and lab data—to provide evidence-based answers to clinical queries. By leveraging NVIDIA’s accelerated computing, the system processes these records on-site, ensuring patient data remains securely within the hospital’s infrastructure without external transfers.
Together, the partners intend to equip hospitals with the ability to leverage the complete histories of patients they have treated, transforming that experience into identifying care needs sooner, making treatment choices more informed, and providing superior care for the next patient. Hospitals possess decades of clinical knowledge in their records, but much of it is hard to analyze on a large scale because of the narrative text and manual review necessary.
Turning Clinical Experience into Better Care
Synapsis AI performs clinical reasoning by working through a chart longitudinally, across every document, resolving what a patient actually had, when, and with what evidence. This requires substantial computing capacity, which NVIDIA provides through its accelerated computing platform and software. Dyania brings the clinical reasoning engine and implementation experience, while NVIDIA brings the computing power and software to run securely and efficiently on premises.
Synapsis AI is built on the NVIDIA AI Enterprise software stack and runs on NVIDIA TensorRT and TensorRT-LLM on the hospital’s own NVIDIA accelerated computing infrastructure. This keeps inference inside the hospital’s private network, ensuring patient records remain secure. Under the collaboration, the companies will work to bring Synapsis AI on NVIDIA infrastructure to health systems in the U.S. and internationally.
The partnership will also include publishing joint case studies and integrating Synapsis AI with NVIDIA’s healthcare software suite. Areas for testing include NVIDIA Confidential Computing for both training and inference, NVIDIA Dynamo for distributed inference serving, and scaling across many NVIDIA Blackwell Ultra GPUs to achieve near-real-time clinical insights at the point of care.
Predicting Heart Valve Longevity
The initial project from the collaboration is a Cleveland Clinic study involving 7,812 patients who underwent aortic valve replacement between 2005 and 2021. The study addressed the main worry of how long the replacement valve will last and whether another procedure is required. Synapsis AI identified the patient group and answered 88 clinical questions for each individual, covering the valve and its traits, co-existing conditions, lab trends, medications, follow-up ultrasound measurements, and subsequent clinical events.
Running on NVIDIA accelerated computing inside Cleveland Clinic‘s environment, Synapsis AI produced a detailed clinical dataset in far less time than a manual review would take. A predictive machine-learning model was trained to predict valve failure and reintervention risk timelines for every patient. The model found that the type of valve implanted was the strongest determinant of how quickly it deteriorated, while age and other medical conditions shaped whether and when a second procedure was performed.
Lars G. Svensson, MD, PhD, Chief of the Heart, Vascular & Thoracic Institute at Cleveland Clinic, observed that using technology to analyze data and forecast the lifespan of implanted heart valves is a positive step toward improved results and more personalized care. The heart valve study demonstrates the wider potential of the partnership, which could assist health systems in learning from their own history to determine which patients would benefit from specific therapies and why outcomes vary between patients.
The analysis also examined how follow-up information, such as heart ultrasounds following surgery, can substantially raise predictive accuracy around future interventions. The model estimates an individual patient’s risk at three, five, and 10 years and revises the outlook with each new ultrasound, providing a tool to help inform care and requiring external validation from clinicians. Synapsis AI agents will continuously train predictive models of this kind, improving recommendations for the next recommended action in a patient’s care as precision medicine advances, with Cleveland Clinic being one of the first to benefit from this technology.
NVIDIA’s AI Enterprise software stack supports the training of Dyania’s proprietary clinical models with the NVIDIA NeMo framework. This framework enables model training and reinforcement learning, allowing Synapsis AI to improve its clinical reasoning capabilities.
The collaboration between Dyania Health and NVIDIA will also involve integrating Synapsis AI with NVIDIA’s healthcare software stack, including NVIDIA Confidential Computing and NVIDIA Dynamo. These technologies will enable secure and efficient training and inference of clinical models, allowing health systems to learn from their patient populations more effectively.
Expanding the Collaboration
The companies are evaluating the use of Multi-Instance GPU for partitioning GPUs across concurrent clinical workloads, enabling health systems to process large amounts of patient data in parallel. This will allow Synapsis AI to provide near-real-time clinical insights at the point of care, enabling healthcare professionals to make more informed treatment decisions.
