Elwahsh Haitham, El-Shafeiy Engy, Alanazi Saad, Tawfeek Medhat A
Computer Science Department, Faculty of Computers and Information,, Kafrelsheikh University, Kafrelsheikh, Egypt.
Department of Computer Science, Faculty of Computers and Artificial Intelligence, University of Sadat City, Sadat City, Egypt.
PeerJ Comput Sci. 2021 Jul 28;7:e646. doi: 10.7717/peerj-cs.646. eCollection 2021.
Cardiovascular diseases (CVDs) are the most critical heart diseases. Accurate analytics for real-time heart disease is significant. This paper sought to develop a smart healthcare framework (SHDML) by using deep and machine learning techniques based on optimization stochastic gradient descent (SGD) to predict the presence of heart disease. The SHDML framework consists of two stage, the first stage of SHDML is able to monitor the heart beat rate condition of a patient. The SHDML framework to monitor patients in real-time has been developed using an ATmega32 Microcontroller to determine heartbeat rate per minute pulse rate sensors. The developed SHDML framework is able to broadcast the acquired sensor data to a Firebase Cloud database every 20 seconds. The smart application is infectious in regard to displaying the sensor data. The second stage of SHDML has been used in medical decision support systems to predict and diagnose heart diseases. Deep or machine learning techniques were ported to the smart application to analyze user data and predict CVDs in real-time. Two different methods of deep and machine learning techniques were checked for their performances. The deep and machine learning techniques were trained and tested using widely used open-access dataset. The proposed SHDML framework had very good performance with an accuracy of 0.99, sensitivity of 0.94, specificity of 0.85, and F1-score of 0.87.
心血管疾病（CVDs）是最严重的心脏病。对实时心脏病进行准确分析具有重要意义。本文旨在通过使用基于优化随机梯度下降（SGD）的深度学习和机器学习技术来开发一个智能医疗框架（SHDML），以预测心脏病的存在。SHDML框架由两个阶段组成，SHDML的第一阶段能够监测患者的心率状况。已使用ATmega32微控制器开发了用于实时监测患者的SHDML框架，以确定每分钟的心率脉搏率传感器。所开发的SHDML框架能够每20秒将采集到的传感器数据广播到Firebase云数据库。该智能应用程序在显示传感器数据方面具有感染力。SHDML的第二阶段已用于医疗决策支持系统，以预测和诊断心脏病。将深度学习或机器学习技术移植到智能应用程序中，以分析用户数据并实时预测心血管疾病。检查了两种不同的深度学习和机器学习技术方法的性能。使用广泛使用的开放获取数据集对深度学习和机器学习技术进行了训练和测试。所提出的SHDML框架具有非常好的性能，准确率为0.99，灵敏度为0.94，特异性为0.85，F1分数为0.87。
