Particle.news
Download on the App Store

Mount Sinai's NutriSighT Predicts Underfeeding Risk in Ventilated ICU Patients

Prospective trials plus EHR integration will test whether real-time use improves nutrition delivery.

Overview

  • Researchers reported in Nature Communications on December 17 that NutriSighT uses routine ICU data to identify which patients are likely to be underfed during the first week of mechanical ventilation.
  • The model updates risk estimates every four hours based on vitals, labs, medications, and feeding records, offering clinicians timely signals for intervention.
  • Outputs are interpretable, highlighting contributors such as blood pressure, sodium levels, and sedation to help tailor feeding plans.
  • Across U.S. and European ICU datasets, underfeeding was frequent, affecting 41–53% of patients by day three and 25–35% by day seven.
  • Investigators stress the tool is decision support rather than a replacement for clinicians, with multi-site prospective testing and EHR workflow integration planned next.