Lightweight 1D-CNN for Real-Time Single-Lead ECG Anomaly Detection
Objective & Abstract
Constructing an ultra-compact convolutional classifier for continuous beat classification on PhysioNet MIT-BIH datasets designed for wearable microcontroller deployment.
Investigated Methodologies
- 1D Temporal Convolutions
- Waveform R-peak segmentation
- Inter-patient evaluation
Datasets & Benchmarking Corpora
PhysioNet MIT-BIH Arrhythmia Database
All training and evaluation datasets undergo ethics audit and licensing compliance per AIAC Research Council standards.