Updated: Oct 20
Federated learning is fundamentally changing the way healthcare AI is developed and deployed. By eliminating the need to move or replicate data, it breaks down what has been one of the biggest challenges facing medical researchers and healthcare AI developers - access to large, diverse, disparate datasets. With federated learning, researchers can utilize datasets that were created and remain at different locations, without creating additional IT or operational complexity.
Rhino Health is announcing its industry-first implementation of the newly introduced NVIDIA FLARE in its ‘Federated Learning as a Platform’ solution, in collaboration with the NVIDIA team. NVIDIA FLARE, short for Federated Learning Application Runtime Environment, is an open-source extensible SDK that makes it possible for developers to customize their federated learning approach for domain-specific applications.
Rhino Health’s ‘Federated Learning as a Platform’ solution for the healthcare industry enables healthcare researchers to collaborate effectively and efficiently on AI model creation, deployment, and improvement without sharing patient data nor putting patient privacy at risk.
“NVIDIA FLARE enables collaborators to quickly adapt machine learning workflows to a distributed paradigm,” said Mona G. Flores, MD, head of medical AI for NVIDIA. “Rhino Health is making it easier to facilitate large-scale medical research collaborations that will ultimately improve the standard of care and patient outcomes.”
This combination of NVIDIA’s industry-leading technology and our medical research workflow and clinical data expertise makes the Rhino Health Platform a powerful tool in the development and deployment of healthcare AI. It’s already happening today.
We’ve recently teamed with the National Cancer Institute’s Early Detection Research Network’s (EDRN) pancreatic cancer working group on a joint collaboration to create AI models that accelerate diagnosis of Pancreatic Ductal Adenocarcinoma (PDAC) and improve patient outcomes.
Together with Assuta Medical Centers, we’re creating never-before-possible access to the diverse datasets required to ensure that AI models work accurately and more consistently for more patients.
We’re helping researchers at Massachusetts General Hospital develop an AI model that more accurately diagnoses brain aneurysms - making it fast and easy for them to set up research collaborations with several other institutions around the world in a matter of days.
All of this builds on the many learnings Dr. Flores and I took from the EXAM study we co-led last year, when we brought together 20 institutions around the world. The EXAM study was essentially the creation and validation of an AI algorithm that predicted the supplemental oxygen needs of people coming to the emergency department with symptoms of COVID-19. Published in Nature Medicine this fall, EXAM is the largest federated learning study to-date in healthcare and provided proof that this distributed approach can work.
I count myself lucky to have worked closely with fellow researchers from Mass General, Brigham and Women, and NVIDIA these past few years. In 2019, we began jointly exploring the benefits of federated learning. This included the ADOPS study, and then the aforementioned EXAM study. And today, at Rhino Health, we continue to closely collaborate with healthcare leaders at NVIDIA as we advance our federated learning platform - Dr. Mona Flores, Kimberly Powell, Dr. Holger Roth, Prerna Dogra, Renee Yao, Dr. Andy Feng, Brad Genereaux - to name a few. Today’s introduction of NVIDIA FLARE is an important milestone on the journey to realizing our shared vision for transforming how healthcare AI is developed and deployed leveraging federated learning.