AI Model Validated for Predicting Infection Risk in Cirrhosis
A new machine learning tool helps doctors rule out spontaneous bacterial peritonitis in patients with cirrhosis.
Updated on Sept. 22, 2026 in Asthma
Live Poll
Should medical professionals rely on machine learning models to non-invasively diagnose patient conditions?
Researchers validated a machine learning model designed to non-invasively exclude spontaneous bacterial peritonitis in patients with cirrhosis and ascites. This tool, analyzed using data from the Veterans Health Administration, aims to support clinical risk-stratification.
Why it matters
Spontaneous bacterial peritonitis is a serious complication for those with cirrhosis, and this model helps clinicians streamline care decisions. The study was conducted to ensure the model remained accurate following shifts in healthcare delivery after the COVID-19 pandemic.
In a validation cohort of 4,192 patients with cirrhosis, researchers confirmed the machine learning model achieved a 93.6% negative predictive value. A focused subgroup of 107 patients showed a 100% negative predictive value for spontaneous bacterial peritonitis at a 5% threshold.
The players
Veterans Health Administration
The largest integrated healthcare system in the United States, which provided the patient data used to validate this machine learning model.
The details
The model incorporates twenty specific clinical and laboratory values to assess the risk of spontaneous bacterial peritonitis in patients admitted with cirrhosis and ascites. By processing these biomarkers, the algorithm provides a probability score that helps clinicians decide whether invasive diagnostic procedures are necessary. The research was based on manual chart reviews of patients treated at two tertiary-care VA hospitals.
Timeline
Patients were admitted to the study cohort between 2020 and 2023.
Health Landscape
This development marks a shift toward integrating machine learning into the routine diagnostic workflow for complex liver conditions. It follows a pattern set by the Veterans Health Administration's clinical decision support initiatives to improve patient outcomes via data-driven tools.
If you or a loved one are managing cirrhosis, this research highlights the growing role of predictive tools in reducing unnecessary invasive testing. These results are worth discussing with your doctor to understand how your care team uses clinical data to monitor your infection risk.
The takeaway
Advanced diagnostic models are becoming more precise at flagging patients who are at low risk for common complications of cirrhosis. Patients should continue to report new symptoms of discomfort or fever promptly, as these tools support—rather than replace—clinical evaluation.
Further reading
For more information on managing respiratory and systemic complications in chronic disease, visit our Asthma section.
Source note: This article includes information reported by Ovid.
Live Poll
Should medical professionals rely on machine learning models to non-invasively diagnose patient conditions?
