Mar 02, 2022, 08:01 ET
DALLAS, March 2, 2022 /PRNewswire/ -- Dallas-based healthcare AI company, Pieces Inc, has announced the launch of a GPT-3 enabled version of their clinical AI software, Pieces Predict. GPT-3, a natural language model developed by the AI research and development company, OpenAI, is being deployed first within the Pieces reLOS application, an administrative use case focused on resource prioritization and optimizing hospital length of stay.
The software helps hospitals identify and prioritize clinical, operational and social barriers that would otherwise prolong a patient's hospitalization. As part of its explanatory mode, Pieces reLOS generates a note summary for multidisciplinary huddles that briefly explains the circumstance of the patient's hospitalization, including current clinical issues and administrative tasks needed for discharge. GPT-3 helps render this AI-generated summary into a context that is faster and easier to interpret by hospital staff.
"The GPT-3 capability enables Pieces to correspond with nurses, doctors, social workers and care managers on a whole different level, in a 'human-like' way, improving the speed and ease of communication," says Ruben Amarasingham, MD, CEO of Pieces, Inc. "The field of real-time healthcare predictive analytics is moving away from numbers alone as a primary mode of communication to one that can now include increasingly supple and subtle prose. We are excited about the possibilities ahead."
Pieces has built security, privacy and human-in-the-loop oversight features specifically around the GPT-3 component of the platform.
"As an industry, we need to approach natural language models in medicine with a certain level of caution and circumspection," says Yukun Chen, PhD, VP of AI at Pieces. "For us this means focusing on administrative use cases and a fairly substantive level of technology safeguards, human oversight and supervision. Eventually, we see this work playing a greater role in deeper clinical use cases, conversational AI and more sophisticated patient-related interactions."
U.S. hospital stays cost the broader health system at least $377.5 billion per year. In today's value-based care environment, hospitals are under increasing pressure to avoid patient harm and maintain quality while also lowering costs and staff burnout. Reducing hospital length of stay is an important indicator of a hospital's success in achieving these goals. Hospital sites interested in early adoption or learning more can sign up here.
About Pieces Predict
Pieces Predict is a cloud-based clinical decision support system that works across healthcare settings to improve outcomes for patients. Pieces insights are made available through the electronic health record or through a Predict web interface. The web-interface provides immediate access to Predict predictions, clinical insights, patient summaries and effectiveness, value and ROI reporting.
Pieces, Inc. is a healthcare artificial intelligence and technology company that connects health systems and the community to address clinical and social determinants of health through community networks and intelligent software and services. Our solution interprets patient information in real-time and connects health systems and community-based organizations to support healthier outcomes, both inside and outside of hospital walls. Using cloud-based artificial intelligence with clinically-based natural language processing (NLP) and physician-supervised machine learning, our tools help streamline clinician workflows and improve patient outcomes. Combined, our solutions, Pieces Predict and Pieces Connect, create a comprehensive and unique solution for connected community health.
Todd Stein for Pieces
Todd Stein Communications
SOURCE Pieces, Inc.
Share this article