Researchers at the Icahn School of Medicine at Mount Sinai have developed an artificial intelligence (AI) tool designed to help identify which critically ill patients on ventilators in intensive care units (ICUs) are at risk of underfeeding. The study, published on December 17 in Nature Communications, suggests that early prediction could enable clinicians to adjust nutrition plans and improve patient care.
According to the research team, the first week for patients on ventilators is crucial because their nutritional needs can change quickly. "Too many patients on ventilators in the intensive care unit (ICU) don't get the nutrition they need during the critical first week," said Ankit Sakhuja, MBBS, MS, Associate Professor of Artificial Intelligence and Human Health, and Medicine (Data-Driven and Digital Medicine), who served as co-senior corresponding author. "Their needs are changing rapidly, and it's easy for them to fall behind. We wanted to explore a simple, timely way to identify who is most at risk of being underfed so that clinicians could intervene earlier, adjust care, and make sure each patient receives the right support when it matters most."
The AI tool developed by the team is called NutriSightT. It analyzes standard ICU data—including vital signs, lab results, medications, and feeding information—to predict which patients may be underfed between days three and seven of ventilation. The model was trained and validated using large deidentified ICU datasets from both Europe and the United States. It updates its predictions every four hours as patient conditions change.
NutriSightT is intended as a support system rather than a replacement for clinical decision-making. The researchers emphasize its role as an early-warning system that could guide timely interventions related to nutrition.
Future plans include conducting prospective multi-site trials to test whether acting on these AI-generated predictions leads to better outcomes for patients. Other next steps involve integrating NutriSightT into electronic health records systems and expanding its use toward more personalized nutrition targets.
"The significance of our study's findings is that, for the first time, it may be possible to identify which patients are at risk of underfeeding early in their ICU stay and tailor care to their individual needs," said Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health; Director of the Hasso Plattner Institute for Digital Health; Irene and Dr. Arthur M. Fishberg Professor of Medicine at Icahn School of Medicine at Mount Sinai; and Chief AI Officer of Mount Sinai Health System. "It represents an important step towards giving clinicians better information to make decisions about nutrition. Ultimately, the goal is to provide the right amount of nutrition to the right patient at the right time, which could help improve recovery and outcomes in critically ill patients and lay the groundwork for more personalized care strategies."
The paper is titled "NutriSighT: Interpretable Transformer Model for Dynamic Prediction of Underfeeding Enteral Nutrition in Mechanically Ventilated Patients." Authors include Mateen Jangda, Jayshil Patel, Akhil Vaid, Jaskirat Gill, Paul McCarthy, Jacob Desman, Rohit Gupta, Dhruv Patel, Nidhi Kavi, Shruti Bakare, Eyal Klang, Robert Freeman, Anthony Manasia, John Oropello, Lili Chan, Mayte Suarez-Farinas, Alexander W. Charney, Roopa Kohli-Sethh Girish N. Nadkarni,and Ankit Sakhuja.
The research received funding support from a National Institutes of Health (NIH) grant K08DK131286.e.
