VCU-led global team uses AI to predict death risk in hospitalized liver patients

RICHMOND, Va. (WRIC) -- A Virginia Commonwealth University gastroenterologist and a global team of researchers have developed an artificial intelligence model that significantly improves doctors’ ability to predict which hospitalized cirrhosis patients are at greatest risk of dying.

The model, developed by the CLEARED Consortium and led by Dr. Jasmohan Bajaj of the VCU School of Medicine, was published Wednesday in the journal Gastroenterology. The AI tool outperforms older prediction methods and is being made available worldwide to help guide care for patients suffering from advanced liver disease.

“This means a doctor can have more confidence about which patients need the most urgent care, which ones might need hospice discussions with family members, who could need transfer to better-equipped hospitals and which patients are likely to recover,” said Bajaj, who is also with the VCU Stravitz-Sanyal Institute for Liver Disease and Metabolic Health and the Richmond Veterans Administration Medical Center. “Medically and nonmedically, we can better approach the patient if we have a better handle on the patient’s condition.”

Cirrhosis, caused by long-term liver damage from alcohol use, hepatitis or fatty liver disease, is a severe condition that can lead to repeated hospitalizations and life-threatening complications like infections, kidney failure and hepatic encephalopathy. And once these patients are hospitalized, they are faced with a high chance of death.

To tackle the issue, the team studied more than 7,000 cirrhosis patients treated at 121 hospitals across six continents, including in the U.S., Asia and Africa. They analyzed data on the patients’ hospital stays, complications, treatments and survival outcomes.

The researchers found that a machine learning model called Random Forest delivered the most accurate predictions. It scored 0.815 in accuracy — significantly higher than the 0.773 score of the traditional logistic regression model.

Even when simplified to just the 15 most predictive factors, the Random Forest model remained more effective than older methods. Key risk indicators included admission for kidney failure, brain complications and severe infections — all of which strongly raise the risk of death during hospitalization.

“High-risk patients could be shifted to another hospital for better treatment,” Bajaj said. “And the last thing is, if it’s likely they’re on the path toward decline and possibly palliative care, those decisions can be made earlier, when the patient is still awake and alert and can participate in making them.

“On the other hand,” Bajaj continued, “if the patient is low-risk, doctors can feel more confident focusing on recovery and discharge planning.”

The model was also tested on nearly 29,000 U.S. military veterans with cirrhosis treated at VA hospitals. Despite the veterans being an older, mostly male population, the AI tool still dominated over traditional scoring systems.

Now, the CLEARED Consortium is sharing the model with hospitals in the U.S. and internationally. With only 15 data points required, the tool is designed for easy use, allowing physicians to quickly assess a patient’s risk and take the proper steps.

“Better prediction means better planning,” Bajaj said. “When we know who is most at risk, we can target treatment, talk to families early and focus our resources where they matter most.”