HeartMate App-Based Remote Monitoring for Cardiovascular Risk Factor Management

Study design: This is a single-center, prospective, open-label, randomized controlled exploratory pilot trial with two parallel groups and 1:1 allocation. A total of 200 participants (100 per group) are planned, with approximately 170 evaluable participants expected after an anticipated 10-20% dropout.

Background and rationale: Traditional hospital-based cardiac rehabilitation has low participation because of travel, time, and cost barriers. Smartphone-based remote patient monitoring and digital cardiac rehabilitation may lower these barriers and enable continuous monitoring of cardiovascular risk factors. Smartphone-derived digital biomarkers, including remote photoplethysmography (rPPG) and heart rate variability (HRV), may provide non-invasive markers correlated with conventional cardiovascular risk indicators.

Randomization: After completion of baseline assessment, participants are randomly allocated 1:1 to the intervention or control group using a random number table with block randomization (block size of 4). The allocation sequence is generated by a person independent of the study, and allocation is concealed using sequentially numbered, password-protected locked files. Investigators performing screening and enrollment do not have access to the allocation sequence.

Intervention group: Participants use the HeartMate app-based remote cardiac rehabilitation program for 6 weeks. The program provides a daily heart check using rPPG, aerobic and resistance exercise, breathing training, and a cognitive-behavioral mind program. Trained staff provide message-based counseling and exercise and lifestyle guidance at least weekly; the Severance research team reviews monitoring logs at least weekly and handles all medical decisions and adverse events.

Control group: Participants receive usual outpatient follow-up and lifestyle education materials. The HeartMate app is installed for rPPG-based measurement only, with no exercise or mind program, so that digital-biomarker measurement is identical across groups and to minimize confounding from app installation.

Assessments: Clinical assessments (6-minute walk distance, grip strength, blood pressure, height, weight, electrocardiogram), blood tests (lipid profile, fasting glucose, renal and hepatic function, electrolytes; HbA1c for participants with diabetes; NT-proBNP for participants with heart failure), questionnaires (SF-12, PHQ-9, GAD-7, KOSS-24), and digital biomarkers (rPPG-derived heart rate, HRV, respiration, activity; reference contact PPG for accuracy verification) are collected at baseline and at 6 weeks. Standardized measurement conditions are applied for rPPG (fasting, seated stable posture, camera distance 30-50 cm, uniform lighting, repeated measurement, and signal-quality control).

Outcomes: The primary outcome is the between-group difference in the change in 6-minute walk distance from baseline to 6 weeks, analyzed primarily by independent-samples t-test (exploratory) with ANCOVA as a supportive analysis adjusting for baseline and covariates. Secondary outcomes include safety (adverse events and serious adverse events) and adherence (app usage days and target usage rate of at least 80%). Exploratory outcomes include changes in blood pressure, blood lipids, glucose and HbA1c, quality of life and psychological measures, and the agreement between rPPG-derived and reference contact-PPG digital biomarkers.

Statistical considerations: As an exploratory pilot, the sample size is intended to provide reliable preliminary estimates (effect size, standard deviation, dropout rate) for a future definitive randomized controlled trial rather than to confirm efficacy. Analyses follow the intention-to-treat principle, with missing data handled per the statistical analysis plan.

Data management: Data are identified only by a unique subject code and stored on an encrypted secure server (AES-256 encryption; FIPS 140-3 physical and logical security). rPPG facial video is not stored; only de-identified RGB time-series signals are retained. Records are kept for 3 years after study completion and then destroyed irrecoverably.