Authors: Dhage Rutuja P., Professor Bangar Abhishek P., Professor Karad Vandana A.
Abstract: Silent myocardial infarction (SMI) remains among the highest-risk elements of global healthcare because its safety de-pends on early detection, yet it produces no recog-nizable symptoms that would prompt a patient to seek care. Conven-tional reactive protection based on clinic-bound electrocardiograms, blood troponin assays, and imaging cannot confirm cardiovascular damage before irreversible necrosis occurs. This paper presents a structured narrative review of verified primary sources pub-lished between 2015 and 2026, organized into five themes: photoplethysmography (PPG) wave morphology, deeplearning sleep apnea tracking, vocal biomarker feature extraction, attention-based machine learning data fusion, and edge IoT medical de-vice deployment. Convolutional neural networks, long short-term memory encod-ers, and multi-head attention mechanisms achieve high reported precision for cardiovas-cular risk indexing, physiologi-cal feature compression, and acoustic voice mapping. Lightweight feature-based classifiers further demonstrate that these pipelines can run on low-power wearable and mobile hardware suitable for continuous monitoring. However, the evidence is currently dominated by single-cohort studies and optimistic intra-subject partitioning bias, public multi-sensor cardiac datasets are scarce, and missed-detection rates, algorithmic latency, and prospective clinical safety vali-dation are rarely reported. The paper therefore contributes a thematic synthesis, a compara-tive study matrix, a proposed conceptual seven-layer framework, and a prioritized biomedical research agenda.
