NYU Hospital Deploys AI to Predict Patient Readmissions
A new artificial intelligence (AI) computer program developed at New York University (NYU) is revolutionizing the healthcare industry. The program, known as “NYUTron,” is being used to predict the likelihood of patients being readmitted to the hospital shortly after discharge.
Lavender Jiang, lead researcher at NYU’s Center for Data Science, explained that NYUTron analyzes doctors’ unaltered notes, estimating a patient’s risk of death and length of hospital stay, along with other crucial factors for their care. The AI program is designed to predict four out of five patients who may require readmission, according to a report published in the journal Nature.
The program is a groundbreaking example of what developers call a “big language model,” which can understand the creative and personalized notes doctors frequently take. Jiang notes that this technology could provide healthcare providers with real-time alerts about potential issues, allowing them to address concerns before they escalate.
By automating routine tasks, NYUTron allows medical professionals more time to spend with their patients— ultimately improving the quality of patient care. Researchers at NYU trained the program using millions of clinical notes from the electronic health records of over 336,000 patients, resulting in a 4.1 billion-word language “cloud” that includes records such as radiology reports, patient progress notes, and discharge instructions.
While the AI program has shown impressive predictive capabilities, Dr. Eric Oermann, lead researcher and neurosurgeon, emphasized the model as a support tool for healthcare providers rather than a replacement for a doctor’s judgment tailored to an individual patient. Oermann believes future studies could further explore NYUTron’s ability to extract billing codes, predict infection risk, and identify potential medication orders, among other applications.
Researchers are confident that with continued development, large language models will make the development of “smart hospitals” a reality. This advancement in the healthcare system not only has the potential to save lives but also has broader implications for improving the efficiency and effectiveness of healthcare delivery.
Funding for the study was provided in part by the US National Institutes of Health, signaling the initiative’s potential to transform healthcare and improve patient outcomes across the United States. For more information on major language models, the Brookings Institute offers additional resources.