Characterizing the Performance Gap in Human Activity Recognition for Older Adults

2026-10-01

AI Summary

This study addresses the limited generalizability of wearable human activity recognition (HAR) models, whose performance gains on young-cohort benchmarks fail to transfer to older adults. Leveraging wrist-worn accelerometer data from the MyMove dataset, we systematically evaluate multiple deep learning architectures and cross-dataset transfer learning strategies, employing leave-one-out cross-validation to ensure result reliability. Our analysis reveals inherent age biases in existing benchmarks and demonstrates that merely scaling network architectures cannot bridge the intergenerational performance gap. The core contribution lies in showing that incorporating age-diverse representations via self-supervised pre-training on the UK Biobank significantly enhances activity recognition accuracy for older adults, effectively narrowing the performance disparity across age groups.

Abstract

Human activity recognition (HAR) from wrist-worn accelerometers is increasingly used for health and behavioral tracking. Yet, most wearable HAR models are developed and evaluated on datasets dominated by younger adults, leaving it unclear whether benchmark progress generalizes across age groups. In this work, we leverage MyMove, our carefully annotated, free-living older-adult HAR dataset (mean age 71), to evaluate deep-learning architectures and training regimes under both leave-one-subject-out and cross-dataset transfer. We find that improvements on younger-adult benchmarks fail to transfer equally to data collected from older adults, resulting in a persistent and often widening performance gap. However, richer representations, particularly frozen self-supervised features pretrained on the age-diverse UK Biobank dataset, substantially improve performance on data from older adults and consistently narrow the performance gap, at modest cost to younger-adult performance, though disparities remain. These findings suggest that benchmark gains and architectural scaling alone provide an incomplete picture of progress in wearable HAR, and broader advances may require representations that better capture population diversity, alongside personalized adaptation to individual movement patterns and routines.

Problem

• Human Activity Recognition
• Performance Gap
• Older Adults
• Wearable Sensors
• Domain Generalization

Innovation

• Human Activity Recognition
• Self-supervised Learning
• Performance Gap
• Representation Learning
• Older Adults