Fairness and Bias Assessment

Focus of Assessment

• Review of dataset representativeness across patient populations

• Assessment of model performance across demographic subgroups

• Identification of bias and equity risks

• Review of data collection and labelling practices

• Recommendations for bias mitigation and monitoring

Governance Value

• Reduced risk of unintended harm to specific populations

• Improved transparency around AI limitations and risks

• Stronger alignment with ethical and responsible AI principles

• Increased trust from clinicians, patients, and regulators

All assessments are delivered through a socio-technical lens, considering data, models, people, workflows, and governance together.