Critically ill patients who need breathing machines often miss getting enough nutrition, especially during the first few days in intensive care.

Now, a new study suggests artificial intelligence (AI) could help doctors spot that problem sooner.

Researchers at the Icahn School of Medicine at Mount Sinai in New York City reported that an AI tool was able to predict which patients on ventilators were likely to be underfed during their first week in the ICU.

The findings were published Dec. 17 in the journal Nature Communications.

“Too many patients on ventilators in the intensive care unit (ICU) don’t get the nutrition they need during the critical first week,” said study co-senior author Dr. Ankit Sakhuja, an associate professor of artificial intelligence and human health, and medicine at Mount Sinai.

"Their needs are changing rapidly, and it’s easy for them to fall behind," Sakhuja added in a news release.

The research team developed an AI program called NutriSighT, which analyzes routine ICU information such as vital signs, lab tests, medications and feeding data, to estimate nutrition risk.

The system updates its predictions every four hours, allowing doctors to see changes as a patient’s condition shifts.

Using ICU data from the United States and Europe, the researchers found that underfeeding was fairly common.

Between 41% and 53% of patients were underfed by day three on a ventilator. By day seven, about 25% to 35% were still not getting enough nutrition.

The AI model also showed which factors played a role in underfeeding risk: Blood pressure, sodium levels and sedation. Researchers say this transparency could help care teams adjust feeding plans earlier.

“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 co-senior author Dr. Girish Nadkarni, chief AI officer of the Mount Sinai Health System.

"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." Nadkarni said.

The researchers stressed that the tool is not meant to replace doctors or dietitians. Instead, it could act as an early warning system to help guide care decisions.

Next, the team plans to test whether using the AI tool in real time improves patient recovery, and to explore how it could be integrated into electronic health records.