AI improves personalised ICU care and patient recovery
ICU data-driven predictions could transform patient care, providing real-time insights into nutrition needs and supporting more personalised strategies.
Researchers at the Icahn School of Medicine at Mount Sinai have developed an AI tool capable of predicting which critically ill ventilated patients may be underfed, potentially enabling earlier nutritional intervention in intensive care units.
NutriSighT, the AI model, analyses routine ICU data, including vital signs, lab results, medications, and feeding information. Predictions are updated every four hours, allowing clinicians to identify patients at risk of underfeeding during days three to seven of ventilation.
The study found that 41–53% of patients were underfed by day three, while 25–35% remained underfed by day seven.
The model is dynamic and interpretable, highlighting key factors such as blood pressure, sodium levels, and sedation that influence underfeeding risk. Researchers emphasise that NutriSighT supports personalised nutrition and guides clinical decisions without replacing medical judgement.
Future research will focus on prospective multi-site trials, integration with electronic health records, and expansion to broader, individualised nutrition targets. Investigators hope these advances will enhance patient outcomes and enable more tailored ICU care.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
