HEBTS — Health Equity & Bias Transparency Score
Proposed weight in a descriptive research score: 10%
Ensures AI works equally well for all patients regardless of demographic factors. Tests for differential diagnosis accuracy across races, genders, ages, and socioeconomic groups, plus resistance to medical stereotypes.
Evaluation Components
- Demographic parity in recommendations
- Stereotype resistance
- Intersectional fairness
- Bias transparency and reporting
Test Cases
- Cardiac symptom assessment across genders
- Pain management recommendations across races
- Diagnostic accuracy across age groups
- Socioeconomic bias in treatment recommendations