An IoT-enabled heart disease prediction framework using hybrid and multi-dilated convolution aided adaptive residual...

In the digital era, the medical industry creates a large amount of information related to patients. Manual processing of this produced information by a physician is very difficult. Therefore, the Internet of Things (IoT)-enabled heart disease detection is currently gaining high attention from various technical fields, particularly for personalized medical care. Still, in several cases, efficient detection of heart disease and 24-h consultation with an expert is not possible because of various reasons. Additionally, there are a lot of heart-related deaths, and the death count is rising every day. Prediction and detection of heart disease need high perfection and precision since a minor error could result in a serious condition or the death of an individual. So, an IoT-based network is developed in this paper to tackle these issues. The introduced approach is implemented in two phases. At the beginning phase, the required signal is collected and it is converted into spectrogram images with the help of a Short-Time Fourier Transform. In the second phase, sensor data are collected using IoT devices. This collected sensor data and the spectrogram images are given to the Hybrid and Multi-dilated Convolution based Adaptive Residual Attention Network (HMDCARAN) for predicting heart disease. The suggested HMDCARAN's parameters are tuned by the Modified Crayfish Optimization Algorithm. The outcome of the implemented network is compared with the traditional approaches to verify its effectiveness. Here, the developed framework achieved an accuracy of 96.52%, precision of 98.29%, and sensitivity of 97.18, which is enhanced than the other frameworks. Thus, the outcome proved that the designed network can identify heart disease in the initial stage and overcome the risk factors caused at the advanced stages of the heart disorder.