Neuromorphic Parallel Computing Hardware Based on Quantum Dots for 12-Lead Electrocardiogram Monitoring

• • Single-array 12-lead ECG monitoring achieves >98% training accuracy on the MIT-BIH Arrhythmia Database and Chinese Twelve-Lead ECG Challenge Database, eliminating the need for twelve separate circuits and reducing hardware cost by an order of magnitude while mitigating device non-uniformity that degrades performance as array scale increases. • • Five-class ECG signal classification (normal N and abnormal A, V, L, R) attains 96.2% recognition accuracy, validated by a 5×5 confusion matrix, demonstrating robust discrimination of arrhythmia types critical for clinical triage and reducing false negatives in portable monitoring. • • The 12-transistor STAC array directly processes one-dimensional ECG temporal sequences without signal format conversion, integrating feature extraction and feature fusion layers into a single circuit, which removes external computing and control circuits and shortens processing time compared to conventional two-layer transistor arrays. • • Myocardial infarction detection is validated by pronounced Q waves in leads V1, V2, II, III, and aVF when drain current is altered, confirming that weight tuning within the array can influence specific gravity in output results and enabling precise localization of ischemic events.