Remote monitoring of physical activity captured by the MyHeart Counts app identifies pulmonary arterial hypertension

IntroductionThe diagnosis of Pulmonary arterial hypertension (PAH) is often delayed. Wearable devices and smartphone applications have potential to remotely monitor patients and as access to technology increases, potentially identify patients at risk of cardiovascular disease, and reduce diagnostic delays. MethodsWe utilised data from 109 UK participants with up to 8 years of real-world activity monitoring captured using the MyHeart Counts app. We analysed physical activity, heart rate data and questionnaire responses from patients with idiopathic PAH (IPAH, n=34), disease controls (n=14), and healthy individuals (n=61). Validation studies were performed in an age and sex matched US MyHeart Counts cohort (28 IPAH, 23 other cardiovascular disease controls, and 22 healthy individuals). ResultsThere were significant differences in activity levels between patients with PAH and controls. Activity metrics improved for patients with PAH following diagnosis. Correlations were observed between wearable-derived metrics and clinical variables such as 6MWT distance and NT-proBNP levels, supporting their potential to complement traditional risk assessment tools. Utilising pre-diagnostic data, we developed a classifier to identify patients with PAH from controls (ROC AUC = 0.87{+/-}0.07). The addition of patient responses to questionnaires, particularly on lifestyle enhanced model performance (ROC AUC = 0.94{+/-}0.07). The model was then tested in a US MyHeart Counts validation cohort. Following re-training (accounting for differences in control population) the model provided a performance of ROC AUC = 0.74{+/-}0.11. ConclusionThis study highlights the feasibility of using wearable and smartphone technology to enable early detection, longitudinal monitoring, and remote risk assessment in PAH.