Electrophysiology and Arrhythmias
Session Title: AI Driven ECG and Rhythm Analysis, More Reliable Results With Less Effort
Abstract 15673: Evaluation of the Use of Wearable Device Signals for Patients With Heart Failure: A Subgroup Analysis in the Michigan Predictive Activity & Clinical Trajectories in Health (MIPACT) Study
Background: Wearable devices are increasingly used by patients and the data made available to their care teams. However, interpretation of these data, especially in diverse populations, remains challenging. Herein, we report on the results of a subgroup analysis of wearable device data among participants with heart failure (HF) within a large observational digital health study.
Methods: The MIPACT Study enrolled 6765 adult Michigan Medicine patients with an iPhone 6 or newer between August 2018 and December 2019. All were provided with an Apple Watch. The electronic medical record was reviewed to determine New York Heart Association (NYHA) functional class and HF phenotype. Wearable activity data (e.g., step count, exercise minutes) were compared across phenotypes and functional classes using a one-way ANOVA.
Results: 118 participants had an ICD-10 code for HF within 90 days of enrollment and were confirmed to have HF on chart review. Participants were 57.4 (SD 14.2) years of age; 68 (57.6 %) were male and 29 (24.6%) Black. 29 (26.9%) participants had NYHA class 2 and 10 (8.5%) NYHA class 3 symptoms; 35 (29.9%) had HF with preserved EF (HFpEF) and 85 (81%) nonischemic cardiomyopathy. Participants with NYHA class 1 HF symptoms had significantly higher mean daily exercise minutes and mean daily step counts when compared to patients with NYHA class 2 HF symptoms and with a composite of NYHA classes 2-4 (Figure 1). Participants with HF with mildly reduced EF (HFmrEF) had significantly higher exercise minutes and step counts when compared with participants with HFpEF and HF with reduced EF (HFrEF). Participants with HFrEF and HFpEF were similar with respect to exercise minutes (p = 0.82) and step counts (p = 0.91).
Conclusion: Wearable physical activity measures vary by HF phenotype and functional classification with similar profiles for patients with HFpEF and HFrEF. Further research is necessary to understand longitudinal changes in wearable device signals and prognosis.