MedTech Terms
    The authoritative reference
    All terms
    Software & AIAI / ML in DevicesSoftware LifecycleClinical Evidence

    Algorithmic Bias and Fairness

    Systematic differences in model performance across demographic or clinical subgroups.

    Reviewed by Christian Espinosa, Founder, Blue Goat CyberLast reviewed September 19, 2026

    Definition

    Algorithmic bias occurs when a model performs differently across subgroups (e.g., by sex, race, age, comorbidity) due to training data imbalance, label noise, or proxy variables. Fairness analyses quantify and mitigate these gaps. Regulators expect manufacturers to report performance by relevant subgroups and explain any gaps, rather than only overall accuracy. FDA's draft AI-enabled device guidance and the EU AI Act's data governance rules (Article 10) both address representativeness and bias. Mitigations include more diverse data collection, re-weighting, subgroup-specific thresholds, labeling of limitations and post-market monitoring of subgroup performance.
    What the regulation says
    Regulators emphasize fair and unbiased performance for AI/ML-based MedTech. The FDA

    What this means in practice

    FDA's GMLP principles call for representative datasets and subgroup performance analysis. EU AI Act high-risk requirements explicitly address bias monitoring across the lifecycle.

    Examples

    • A diagnostic AI trained predominantly on data from one ethnic group performs poorly when used on patients from other ethnic groups, leading to misdiagnoses.
    • A risk stratification algorithm for disease progression overestimates risk for one sex due to proxy variables correlated with socioeconomic status in the training data.
    • A wearable device's algorithm, designed to detect a medical condition, shows lower accuracy for individuals with darker skin tones because the training images were not diverse enough.
    Common pitfalls
    • •Failing to define fairness metrics relevant to the specific clinical context and patient population.
    • •Overlooking the potential for bias to be introduced or amplified at various stages of the AI/ML lifecycle, from data acquisition to deployment.
    • •Assuming that a globally accurate model is also fair across all demographic subgroups.

    Frequently asked questions

    Algorithmic bias most commonly stems from unrepresentative or imbalanced training datasets, where certain patient subgroups are underrepresented or poorly labeled.

    Cross-references

    Part of

    Related terms

    Grouped by theme
    Cited by

    Where this term appears across MedTech Terms.

    Sources

    3 sources

    Sources support the definition or relevant regulatory context, not necessarily every claim on this page. Each is graded against our source-tier hierarchy so you can see what rests on binding law versus commentary.

    Tier 2Regulator guidance and consensus· 3
    Link health: 3 verified· last checked 2026-06-20
    FDA/Health Canada/MHRA·1IMDRF·1MDCG·1
    1. 1
      Good Machine Learning Practice
      Tier 2 Verified
      FDA/Health Canada/MHRAfda.gov
    2. 2
      IMDRF - Software as a Medical Device
      Tier 2 Verified
      IMDRFimdrf.org
    3. 3
      MDCG Software Guidance
      Tier 2 Verified
      MDCGhealth.ec.europa.eu

    Inline markers like [1] jump to the matching reference above.