Cliexa Reports 76% Reduction in Racial Bias for Opioid Detection AI

The evaluation of the cliexaAI Opioid Use Disorder Solution tracked performance across 1,571 patients, achieving an 82.3% overall accuracy rate. Beyond raw predictive power, the findings highlight a 59% reduction in age-based accuracy disparities. Most notably, high-risk detection for Black patients climbed from 58.3% to 75.0%, keeping gender-based performance variances below three percentage points. These results emerge as healthcare providers increasingly demand rigorous validation to meet the American Medical Association’s criteria for safe AI integration.

Mehmet Kazgan, founder and CEO of cliexa, emphasized that the platform is designed to make clinical decisions defensible rather than merely automated. The architecture relies on a dual-layer approach: a deterministic rules engine anchored in medical evidence combined with predictive models trained on clinical data. This structure is intended to function alongside existing electronic medical records, allowing clinicians to inspect the logic behind risk stratifications. Having moved from the 2022 Mayo Clinic Platform Accelerate cohort to full commercial qualification, cliexa is now preparing for broader deployment across the Mayo Clinic Care Network.