Author Name
Sravanthi Uppara, Venkatesh B, Chakrapani T, Ahmed Basha S, Suvarna K
Abstract
In this study, a Data-Shifting Neural Network (DSNN) applied to electrocardiogram (ECG) signals is used to design a VLSI circuit for abnormal heartbeat detection. In order to increase model generalization, the suggested DSNN generates numerous temporal variants of each pulse by using a data-shifting augmentation strategy to improve the training dataset. The algorithm achieves a high classification accuracy of almost 97.17% by merging these augmented signals with the original dataset, guaranteeing accurate detection of both normal and pathological heartbeats. The design prioritizes low power consumption and compactness, which makes it appropriate for wearing medical equipment. With careful consideration of transistor-level design, memory needs, timing limitations, and low-power optimization approaches, the DSNN is built and simulated at the circuit level using Cadence tools. To guarantee effective hardware implementation, important performance parameters such transistor count, supply vo