Researchers at the Icahn School of Medicine at Mount Sinai in New York City developed an AI tool that predicts which ICU patients on ventilators are at risk of underfeeding.
The model, called NutriSightT, analyzes routine ICU data to estimate underfeeding risk during days three to seven of mechanical ventilation, according to a Dec. 22 news release from the health system. It updates predictions every four hours based on changing clinical data, including vital signs, lab results, medications and feeding patterns.
In the study, 41% to 53% of patients were underfed by day three, and 25% to 35% remained underfed by day seven. The tool is designed to help clinicians adjust nutrition plans earlier and avoid critical gaps in care.
Mount Sinai researchers emphasized that NutriSightT is not intended to replace clinical judgment but to serve as an early warning system. The research team is planning future multi-site trials, EHR integration and expansion toward personalized nutrition targets.
The study was published Dec. 17 in Nature Communications.
