March 19, 2024

Abridge to Deliver Generative AI Microservices to Millions of Clinicians and Their Patients with NVIDIA

Research collaboration and investment from NVIDIA to help Abridge scale multilingual clinical conversation platform across the entire U.S. healthcare system.

PITTSBURGH, PA (March 19, 2024) – Abridge, the leader in generative AI for clinical documentation, today announced a collaboration with NVIDIA to utilize NVIDIA’s compute resources, foundation models, and expertise in efficiently deploying AI systems at scale to bolster Abridge’s mission to improve clinician and patient lives through technologies powered by generative AI. The company also announced an investment from NVentures, NVIDIA’s venture capital arm.

Every day in the U.S., over a million physicians, five million nurses, and many other medical professionals have conversations with their patients. However, most of this content is never captured in the medical record. The founding insight of Abridge was that AI systems could structure information from these “oceans of missing data,” automating much of the clerical work that has burdened clinicians and providing novel insights that could inform care. As Abridge’s solution rapidly gains widespread adoption throughout the U.S., realizing this promise will require both methodological breakthroughs and the intelligent deployment of computing power to support this research. Abridge is working with NVIDIA to advance these technologies and to serve them efficiently in production to clinicians across the U.S. and, eventually, internationally.

Going from raw audio in noisy environments to meticulous documentation requires many AI technologies to work together seamlessly. Language identification, transcription, alignment, and diarization must happen within seconds. Conversations must be structured according to the sorts of medical information contained in each utterance, and powerful language models must be applied to transform the relevant evidence into summaries. Another suite of models powers all of the actions required to verify the accuracy of each line in the note and to support efficient editing. Each of these models must be trained, often leveraging pre-trained foundation models, and deployed into production where performance demands are critical. Abridge has long been a power user and early adopter of NVIDIA’s technologies for effective training and efficient deployments. In particular, Abridge’s production deep learning stack is powered by the new NVIDIA NIM inference microservices, part of the NVIDIA AI Enterprise software platform and built on the robust foundation of NVIDIA TensorRT-LLM. NIM is engineered to facilitate seamless AI inferencing at scale, helping ensure that AI models can be deployed with confidence. These innovations have reduced latency from minutes to seconds, saving clinicians hours.

“We are thrilled to work with NVIDIA to elevate our research efforts. This collaboration will help us to redefine the possibilities for clinically specialized foundation models, advance medical speech and language technology, and innovate in the space of responsible AI,” said Zachary Lipton, Chief Technology and Scientific Officer of Abridge and the Raj Reddy Assistant Professor of Machine Learning at Carnegie Mellon University. “We’re also working with NVIDIA on core research efforts and plan to explore novel collaborative research in healthcare applications that could improve the lives of clinicians and patients every day.”

“Generative AI is transforming nearly every aspect of healthcare,” said  Mohamed “Sid” Siddeek, corporate vice president and head of NVentures. “Abridge is tapping NVIDIA technologies to help further its foundational research efforts, pioneer and scale methods to advance state-of-the-art AI systems, and responsibly deploy these in healthcare settings.”

Learn more about Abridge at NVIDIA GTC, a global AI conference, during Powell’s session on the Adoption of generative AI in healthcare, on Tuesday, March 19, at 11 AM PDT.


Jill O’Brien