Title : Smartphone-based heart failure monitoring using AI-driven acoustic sensors in low-resource settings: A feasibility approach
Abstract:
Background: Heart Failure (HF) remains a leading cause of cardiovascular death and hospital readmission worldwide, disproportionately affecting patients in Low and Middle-Income Countries (LMICs). In these regions, routine follow-up and echocardiographic monitoring are often inaccessible, leading to late presentations and avoidable complications. Recent advances in Artificial Intelligence (AI) and mobile Health (mHealth) offer an unprecedented opportunity to bridge this diagnostic gap through low-cost, scalable technologies.
Objective: This study explores the real-world feasibility and diagnostic performance of a smartphone-based, AI-driven acoustic monitoring system for early detection of decompensated heart failure in urban low-resource settings, using Karachi, Pakistan as a pilot site.
Methods: A prospective feasibility study was conducted from June 2024 to March 2025. Fifty ambulatory HF patients (NYHA II–III) were provided with a mobile app connected to an AI-enhanced digital stethoscope. The device captured heart and lung sounds, which were analyzed in real-time by a deep learning algorithm trained to detect pathologic features such as third heart sounds (S3), pulmonary crackles, and abnormal heart rate patterns. Alerts were automatically flagged to clinicians via a remote dashboard. Diagnostic performance was compared to NT-proBNP levels and standard clinical assessment. User satisfaction and acceptability were also assessed through structured interviews.
Results: The AI system achieved a sensitivity of 89.2% and specificity of 84.7% in predicting impending HF decompensation. In 18 cases, alerts triggered by the AI tool led to early clinical intervention, potentially preventing hospitalizations. Patients and caregivers expressed high confidence in the tool, with 92% rating it as easy to use and 86% willing to continue long-term use. The total cost per patient was under USD 20/month, highlighting its affordability for public health scaling.
Conclusion: This study demonstrates that AI-integrated acoustic cardiopulmonary monitoring via smartphones is not only feasible but highly effective in detecting early signs of heart failure decompensation in resource-limited environments. By empowering patients and enabling timely clinical decisions, this low-cost innovation has the potential to transform HF management, reduce hospital burden, and save lives—especially in regions where every hospital visit is a hardship. With further validation, this approach could redefine how we manage chronic cardiac diseases across the developing world.
