Context-Dependent Age-Group performance hierarchies limit fairness interventions in PPG-based heart rate prediction

Background

Fairness-aware machine learning increasingly targets demographic performance disparities in clinical prediction, yet whether standard bias mitigation strategies genuinely improve equity in physiological signal analysis remains unclear. Age-based disparities in photoplethysmography (PPG)-based heart rate prediction present a particular challenge, as age-related performance differences may reflect context-dependent physiological structure rather than correctable artifacts.

Methods

We evaluated three fairness interventions — inverse-frequency weighting (IF), Group Distributionally Robust Optimization (GroupDRO), and adversarial debiasing (ADV) — applied via fine-tuning of a PPG foundation model across three clinical datasets spanning intensive care unit, laboratory, and consumer wearable contexts. Outcomes were assessed using a 2 × 2 framework classifying each intervention–dataset combination by the joint direction of change in mean absolute error (MAE) and fairness gap (FG) across age groups, yielding four outcome types: genuine improvement (G), leveling down (L), selective benefit (S), and both worse (W).

Results

Across nine intra-domain conditions, no intervention simultaneously improved both MAE and FG (0/9 genuine improvement). The dominant pattern was leveling down (5/9): FG decreased but was accompanied by MAE degradation, indicating that apparent fairness gains were achieved at the cost of overall predictive performance. Age-group difficulty ordering varied across clinical contexts at baseline and was not preserved under intervention.

In 18 cross-domain transfer conditions, genuine improvement was rare (4/18) and observed exclusively in non-MIMIC source configurations; models fine-tuned on MIMIC-sourced data yielded no genuine improvements (0/6). Embedding-level representation changes following fine-tuning did not reliably predict fairness outcomes.

Conclusions

Age-based fairness interventions in PPG heart rate prediction indicate a leveling-down pattern rather than genuine equity improvement, suggesting that age-related performance gaps reflect context-dependent physiological structure not fully addressable through standard bias mitigation. Cross-domain transfer further amplifies this instability. These findings suggest that fairness evaluation frameworks for age-stratified physiological prediction should account for context-dependent performance structure rather than treating observed gaps as correctable bias.

Author summary

When machine learning models predict health outcomes less accurately for some demographic groups than others, a common response is to apply “fairness” corrections during model training. These methods assume that performance gaps arise from fixable problems in the data or algorithm. We asked whether this assumption holds for

age-based disparities in heart rate prediction from wrist-worn and clinical sensors. Across three clinical settings — consumer wearables, laboratory recordings, and intensive care — we found that the same age group could be the easiest or hardest to predict depending on the setting, reflecting genuine physiological differences rather than a fixable flaw. When we applied three standard fairness methods, none simultaneously improved accuracy and fairness; instead, most appeared to improve fairness only by making predictions worse for everyone. Models corrected in one setting also performed poorly when moved to another. These results suggest that not all demographic performance gaps are correctable, and that reporting fairness improvements without checking whether overall quality declined can be misleading.