Longitudinal cardio-respiratory fitness prediction through wearables in free-living environments

This study investigates the use of wearable technology to estimate cardiorespiratory fitness (CRF), a strong predictor of cardiovascular disease risk and mortality. Traditional methods of measuring CRF, such as VO₂max testing, require specialized equipment, limiting accessibility. The study developed machine learning models that analyze wearable sensor data to estimate fitness levels in free-living conditions. Using data from over 11,000 participants, including a longitudinal cohort of 2,675 individuals, the models demonstrated strong accuracy (r = 0.82). A control group was used, and external validation was performed with the UK Biobank Validation Study (N=181). The method showed statistical significance but tended to underpredict declines in fitness over time. Limitations include reliance on submaximal VO₂max testing and potential biases in lower-fitness populations.

To improve brain health, regular physical activity tailored to cardiovascular fitness levels is beneficial. While wearables can estimate fitness, incorporating aerobic exercises like brisk walking or cycling remains essential. Monitoring trends in CRF over time, rather than single assessments, offers better insights into health. As this study highlights prediction challenges in lower-fitness groups, individualized training plans should consider baseline fitness rather than relying solely on wearable-generated estimates.