Researchers at the University of Utah’s Department of Psychiatry and Huntsman Mental Health Institute have developed an innovative software toolkit, RiskPath, that uses explainable artificial intelligence (XAI) to predict chronic and progressive diseases years before symptoms arise. Their findings, published in Patterns, mark a major step forward in preventive health care.
What Is XAI and Why It Matters
Unlike traditional AI systems, XAI offers transparency by explaining complex decisions in ways humans can understand. RiskPath leverages this approach, offering more than just predictions—it provides insight into the “why” behind disease risk.
Achieving Unprecedented Accuracy
RiskPath analyzes patterns in longitudinal health data—collected over several years—with an impressive accuracy range of 85–99%. In contrast, most current prediction models correctly identify at-risk individuals only 50–75% of the time.
Understanding Risk Over Time
By using advanced time-series AI algorithms, RiskPath tracks how risk factors evolve throughout a person’s life. For instance, the study found that screen time and executive function become critical risk indicators for ADHD as children enter adolescence.
Simplifying Risk Assessment
Despite its ability to process hundreds of health variables, RiskPath often needs only 10 key indicators to make accurate predictions. This makes the tool practical for real-world clinical settings, where time and resources are limited.
Visualizing Risk for Timely Interventions
RiskPath also offers user-friendly visualizations that show when specific life stages contribute most to chronic diseases risk. These insights allow health providers to identify optimal windows for preventive action.
Broad Applications and Future Goals
As reported by medicalxpress, the team validated RiskPath using three large-scale patient cohorts, predicting eight conditions including depression, anxiety, ADHD, hypertension, and metabolic syndrome. They now aim to integrate the tool into clinical decision-making systems, expand it to include more diseases and diverse populations, and explore its use in mental health research.