NYU Langone Health Creates NYUTron

A new artificial intelligence (AI) program developed by researchers at New York University can accurately predict patients' risk of death, length of hospital stay, and other factors essential to care. The NYUTron program is trained using unaltered text from electronic health records (EHRs). In a study published online on June 7 in the journal Nature, the research team found that NYUTron could predict 80% of those readmitted to the hospital, a roughly 5% improvement over a standard, non-LLM computer model requiring reformatting medical data.

The researchers say that NYUTron's ability to learn from EHR text without needing specialized formatting could make it a valuable tool for physicians and other healthcare providers. NYUTron could help clinicians make better care decisions and improve patient outcomes by providing accurate predictions about patients' health status.

The study's lead author, Lavender Jiang, BSc, a doctoral student at NYU's Center for Data Science, said, "Our findings highlight the potential for using large language models to guide physicians about patient care."

NYUTron is still under development, but researchers believe that it has the potential to revolutionize the way healthcare is delivered. NYUTron could help clinicians make better care decisions and improve patient outcomes by providing accurate predictions about patients' health status.

David’s POV

The development of NYUTron is a promising step forward in artificial intelligence (AI). The program's ability to accurately predict patients' risk of readmission and other health outcomes can revolutionize how healthcare is delivered. By providing clinicians with this valuable information, NYUTron can help them make better care decisions and potentially improve patient outcomes.

While AI will not have the critical thinking needed to replace doctors, it can augment their work and create many more efficiencies.