NVIDIA releases AI-based healthcare tools for hospitals, research organizations

Photo: Westend61/Getty Images

Chipmaker NVIDIA announced the launch of FLARE (Federated Learning Application Runtime Environment), an open-source software platform offering a common computing foundation designed to improve collaboration on AI model development in healthcare.

The Flare platform can integrate with existing AI initiatives, including the open-source MONAI framework for medical imaging, using a server-client technique, according to NVIDIA.

With this setup, learned model parameters from each participant are sent to a common server and aggregated into a global model.

NVIDIA has already led federated learning projects that help segment pancreatic tumors, classify breast density in mammograms to inform breast cancer risk and predict oxygen needs for COVID patients.

Flare supports additional architectures beyond server-client, including peer-to-peer and cyclic. Its goal is to make federated learning accessible to a wider range of applications. Flare will also be used to power federated learning solutions at the American College of Radiology (ACR), where research teams will apply AI to radiology images for breast cancer and COVID-19 applications.

Rhino Health, a partner and member of the NVIDIA Inception program, has also integrated Flare into its federated learning solution. In this case, the platform is aiding researchers at Massachusetts General Hospital to develop an AI model that can more accurately diagnose brain aneurysms.

WHY THIS MATTERS

AI is viewed as a potentially powerful tool for healthcare innovation, with applications ranging from facilitating cancer screenings to improving tumor identification and treatment planning. The challenge is in deploying and integrating a relatively new and complex technology into healthcare workflows.

The Netherlands Cancer Institute (NKI) research and treatment centers currently uses NVIDIA's AI Enterprise software suite to test AI workloads on higher-precision 3D cancer scans than are commonly used today.

The higher memory capacity afforded by AI Enterprise, allows researchers to use high-resolution images for training, which in turn helps clinicians better target the size and location of a tumor every time a patient receives treatment.

THE LARGER TREND

Artificial intelligence can also make physicians' lives easier, according to experts and former clinicians. However, it must take a human-centered approach, easily integrate into physician workflows and be friendly to use.

AI and machine learning are improving clinical processes now, through technology such as natural language processing that aids chart documentation over a patient's lifetime. Payers are continuing to shift their focus to AI and machine learning, with COVID-19 helping to drive changes not just in technology, but also in attitude, according to Shreesh Tiwari, principal at ZS, who made the comments during the HIMSS State of Healthcare event in June.

ON THE RECORD

"Open-sourcing NVIDIA Flare to accelerate federated learning research is especially important in the healthcare sector, where access to multi-institutional datasets is crucial, yet concerns around patient privacy can limit the ability to share data," Dr. Jayashree Kalapathy, associate professor of radiology at Harvard Medical School and leader of the MONAI community's federated learning working group, said in a statement.