A Deep Learning-Based Automated System for Cardiopulmonary Resuscitation and Defibrillation

Abstract

Cardiac arrest is one of the leading causes of death globally, with out-of-hospital cases often classified as severe. Therefore, a rapid and accurate response is crucial to improving survival rates. This study presents the design and implementation of a low-cost, portable, automated cardiopulmonary resuscitation (CPR) device integrated with an intelligent defibrillation system. It is composed of three main components: ECG acquisition using the AD8232 module, a mechanical compression mechanism driven by a slider-crank setup, and a defibrillator unit capable of delivering high-voltage shocks. A convolutional neural network (CNN) was employed to classify ECG signals into shockable and non-shockable rhythms. A dataset composed of three separate sets was utilized for training and testing, achieving an accuracy of 98.5%. Evaluations showed that the CPR device effectively controlled compression depth using LED indicators, accurately detected rhythms, and reliably delivered energy through test capacitor configurations. The device offers a compact, affordable, and efficient solution for improving out-of-hospital cardiac arrest response.