A Heart Attack Rescue Detection Algorithm and System

Abstract

Myocardial infarction (MI), also known as heart attack, stays the leading cause of mortality and morbidity. Statistical reports showed that there are about 300 million people who live alone. Once outbreaks of MI, it is time-critical for them to be rescued and emergency call must be made automatically. Along with the ageing population and inadequate amounts of medical personnel, a heart attack rescue detection algorithm and system have been proposed to reduce the death rate of myocardial infarction sufferers. An innovative heart attack rescue detection algorithm has been proposed using Support Vector Machine; performance evaluation with 10-fold cross-validation shows that it yields overall accuracy (OA), sensitivity (Se), and specificity (Sp) of 95.8%, 94.7% and 96.9% respectively. Compared to traditional threshold-based classifier, the proposed work improves OA, Se, and Sp by 10%. Likewise, compared to traditional hidden Markov and Gaussian mixture model, the improvement is 18%. The detection algorithm incorporates with Bluetooth Low Energy (BLE) based remote door-lock system. The system is hosted by an Android Smartphone (API Level 18 or Android 4.3 or above is required). BLE is used as a means for signal connection. The principle and characteristics of BLE are presented. The Android application is created using Android Studio. To increase the convenience and security, QR code technology is implemented together with the security function from BLE. Once MI of user is detected, the user’s phone will receive a warning message from the MI detector via Bluetooth. Then, an emergency call will be made immediately. At the same time, a security key will be sent to medical centre. When medical personnel have arrived, the security key will help to unlock the door in order to grasp the golden time of rescue.