Turning wearable heart rate data into meaningful physical activity metrics for cardiovascular care: a systematic review

Background/Introduction Heart rate monitoring by wearable devices could offer a physiological, personalised and continuous way to assess physical activity duration and intensity. However, methods that translate heart rate data into clinically meaningful physical activity metrics remain diverse and poorly standardised. Purpose To describe and classify how wearable heart rate data are used to quantify physical activity behaviour and to estimate physiological outcomes in adults with cardiovascular disease or related risk factors. Methods Following PRISMA guidelines, a systematic search was performed in PubMed, Web of Science, and CENTRAL for studies published between 2014 and April 2024. Eligible studies included adults with cardiovascular disease or related risk factors wearing wrist or chest heart rate monitors to estimate physical activity metrics or related physiological outcomes. Two reviewers screened records and extracted data. Findings were synthesised narratively. Methodological quality and risk of bias were assessed using the Crowe Critical Appraisal Tool. Results From 828 records, 20 studies met inclusion criteria (Figure 1). Twelve studies used wrist worn devices, seven used chest straps, and one did not specify device type. We identified four heart rate based estimation approaches (Figure 2): (1) heart rate zone analysis (14/20), quantifying time in moderate to vigorous zones to assess guideline or training adherence, but using heterogeneous definitions for maximum and resting heart rate; (2) physiological modelling (4/20), estimating outcomes such as energy expenditure indicators (physical activity level) or cardiorespiratory fitness (VO2max), often with limited reporting of model parameters; (3) change detection (1/20), using time series and machine learning techniques to detect shifts in activity behaviour over time; and (4) a derived personalised scoring system (1/20), translating intensity and duration into a single feedback score. Crowe scores ranged from 50% to 87% (mean 71%). Across approaches, external validation and reproducible reporting were frequently limited. Conclusions Heart rate based physical activity estimation holds the promise of physiologically meaningful and personalised physical activity monitoring in cardiovascular disease care. Modelling approaches and personalised scoring systems that link physical activity behaviour to cardiovascular outcomes may provide much needed clinical tools for physical activity management in patients. In a rapidly evolving wearable technology landscape, future research should prioritise standardised definitions, algorithm transparency, and prospective clinical validation.