Amalgam Rx, Inc. has taken home the "Overall Large Language Model of the Year" title at the ninth annual AI Breakthrough Awards, with judges recognizing Chiron, the company's foundation model for predictive patient intelligence, as the most impactful large language model in the global AI industry this year.
A Record-Breaking Field
Now in its ninth year, the AI Breakthrough Awards is the longest-running recognition program dedicated exclusively to the AI industry. This year's edition drew more than 5,000 nominations from AI companies across over 20 countries, the most competitive field in the program's history, with past and present winners including NVIDIA, Snowflake, Intuit, Zapier, Dell Technologies, and Qualcomm.
The win builds on Amalgam's track record in the category, following the company's 2024 win for Best Large Language Model Application.
How Chiron Works
Built on Amalgam's Medical-Grade AI™, Chiron is an autoregressive transformer trained on millions of longitudinal, de-identified patient records to predict the next medical event across therapeutic areas. Rather than building separate models for each disease, Chiron treats the patient journey as a single, unified sequence of tokenized clinical events, including diagnoses, medications, referrals, comorbidities, and demographics.
That unified approach allows one model to surface early warning signals for multiple conditions at once. In sleep apnea alone, Chiron delivered nearly a 2x improvement in diagnostic efficiency over traditional tools, identifying the same number of high-risk patients while cutting unnecessary diagnostic follow-ups by almost half.
Chiron is designed to be embedded directly into EHR workflows across major health systems, delivering predictions at the point of care and continuously refining itself through real-world feedback. Amalgam's partners use it to identify patients for therapies and clinical trials, while its adherence-risk predictions help enable earlier intervention before treatment failure occurs.
"Healthcare AI has been fragmented for too long; one model per condition, one data pipeline per disease, and critical longitudinal patterns falling through the cracks. We built our model to change that. By learning across the entire patient journey, it gives providers a comprehensive view of risk they've never had before and opens a wider window for intervention."— Ryan Sysko, Chief Executive Officer, Amalgam Rx
Built for Production, Not Just the Lab
"What distinguishes Chiron is not just the architecture, but the combination of foundation-model design, large-scale longitudinal data, and deep clinical integration that makes it production-ready," said Bharath Sudharsan, Chief Data Scientist and Head of AI at Amalgam. Training on millions of real patient records across therapeutic areas lets the model capture complex temporal relationships that rules-based systems and shallow machine learning cannot see.
Azizi A. Seixas, Ph.D., Interim Chair of the Department of Informatics and Health Data Science at the University of Miami Miller School of Medicine and Founder/Director of The Media and Innovation Lab, added that the focus now is on strategic implementation, embedding predictive intelligence directly into the EHR workflow to accelerate the treatment journey and close the gap between industry innovation and clinical impact.