Long-Term Remote Single-Lead Electrocardiogram Monitoring in Heart Failure Patients Using Machine Learning Models
Aim. To evaluate the effectiveness of a developed remote monitoring system for patients with chronic heart failure (CHF) based on single-channel electrocardiogram (ECG) analysis using machine learning models. Design. A multicenter, randomized, comparative study with consecutive patient enrollment. Materials and methods. The study included 251 patients hospitalized for decompensated CHF at the Cardiology Clinic of the Sechenov First Moscow State Medical University (Sechenov University) and the City Clinical Hospital No. 1 named after N.I. Pirogov. All patients underwent standard clinical, laboratory, and echocardiographic examinations. Before discharge, patients were randomized into two groups. In Group 1 (daily remote monitoring, n = 110), patients were monitored remotely using a personal recorder for complaints, body weight, blood pressure, and single-channel ECG parameters (including analysis of systolic and diastolic myocardial dysfunction). Patients in Group 2 (n = 141) were followed up as outpatients as part of standard practice; their condition was assessed after 4–6 months via telephone survey. The average follow-up period was 4.8 months. Results. In the remote monitoring group, 118 remote and 41 in-person consultations were conducted; therapy adjustments were made 122 times in 78 patients. Patients in this group showed a statistically significant reduction in the number of hospitalizations due to CHF decompensation (p = 0.025), atrial fibrillation paroxysms (p = 0.04), and blood pressure destabilization (p = 0.002), as well as in the incidence of stroke (p = 0.005) and hypertensive crises (p = 0.04). A positive trend toward a decrease in cardiovascular mortality was noted (p = 0.12). Conclusion. Remote monitoring of patients with CHF using artificial intelligence algorithms has proven its high clinical efficacy, significantly reducing the incidence of complications and readmissions for cardiovascular pathology. Keywords: heart failure, remote monitoring, machine learning, single-channel electrocardiogram, cardiovascular mortality, hospitalization, myocardial dysfunction