Pilot Validation of an AI-based Audiovisual Fatigue Assessment Tool (mAI Fatigue) in Chronic Liver Disease: A Multicentre Study

Fatigue affects more than half of patients with chronic liver disease and significantly impairs their quality of life, yet it remains often unrecognized because assessment relies almost entirely on subjective patient reports. The study tested whether audiovisual markers from facial and vocal expressions, captured using the mAI Fatigue tool, could serve as objective indicators of fatigue. A prospective multicentre study in India enrolled 111 adults (aged 18-65 years) – 55 healthy controls and 56 patients with chronic liver disease and moderate to severe fatigue. Over four weeks, participants completed ten assessments with validated tests, reaction time measurements, and audiovisual recordings. Patients with liver disease had significantly slower reaction times than controls (882 ms vs 776 ms). When audiovisual data were aggregated at the individual participant level, correlations strengthened and the predictive model achieved high accuracy. Results were strongest in older patients, women, those with severe fatigue, and certain liver disease types. AI-based audiovisual markers show promise as a useful objective complement to fatigue assessment.