Scalable Vlsi Architecture For Deep Learning - Based Arrhythmia Identification

Effective real-time detection of arrhythmia, a serious cardiac disease, is necessary to allow for prompt medical intervention.In order to diagnose arrhythmia from ECG signals, this research presents a unique, power-efficient VLSI design that integrates a deep neural network (DNN).The design consumes just 11.098 W at 25 kHz and makes use of 28-nm bulk CMOS technology.Its classification accuracy in class-oriented evaluation is 92.37%, while in subject-oriented evaluation, it is 81.60%.The system's applicability for portable diagnostic systems is demonstrated by its optimisation to concentrate on low-frequency atrial activity in order to increase detection accuracy.