Heart disease prediction using IoT based framework and improved deep learning approach: Medical application. (January 2023)

• Record Type: Journal Article
• Authors: Rajkumar, G.; Gayathri Devi, T.; Srinivasan, A.
• Is Part Of: Medical engineering & physics. Volume 111(2023)
• Publication Date: 2023-01
• Subjects: Heart disease prediction -- Median studentized residual approach -- Harris hawk optimization (HHO) -- Modified deep long short-term memory (MDLSTM) -- Improved spotted hyena optimization (ISHO) -- Preprocessing -- Classification; Biomedical engineering -- Periodicals; Biomedical Engineering -- Periodicals; Physics -- Periodicals; Génie biomédical -- Périodiques; Biomedical engineering; Electronic journals; Periodicals; 610.28
• DOI: 10.1016/j.medengphy.2022.103937
• Languages: English
• ISSNs: 1350-4533

Abstract

Highlights: This system proposes an enhanced deep learning-based framework for predicting the heart disease. The general publicly available Hungarian heart disease dataset is utilized for the implementation, which includes heart disease related data collected from patients through IOT sensor devices. The input dataset is preprocessed using Median Studentized Residual approach for resolving error data and missing values. Preprocessed data values are feature selected by Harris Hawk Optimization (HHO) approach. The selected features are then classified into normal and abnormal by Modified Deep Long Short-Term Memory (MDLSTM). The modification in LSTM output is altered using Improved Spotted Hyena Optimization (ISHO) algorithm. The results are implemented in the working platform of Phyton with the metrics such as specificity, sensitivity, F-Score, Kappa value, Accuracy, BER and Execution time. Abstract: Heart disease is the biggest cause of death in the globe. The method of predicting cardiac disease is exceedingly complex. It can only be done properly if the doctor has a lot of expertise and is well-versed in the condition. IoT-based illness prediction is a relatively recent technology for accurately classifying diseases based on sensor data. This system proposes an enhanced deep learning-based framework for predicting the heart disease. The general publicly available Hungarian heart disease dataset is utilized for the implementation, which includes heart disease related data collected from patients through IoT sensor devices. The input dataset is preprocessed using Median Studentized Residual approach for resolving error data and missing values. Preprocessed data values are feature selected by Harris Hawk Optimization (HHO) approach. The selected features are then classified into normal and abnormal by Modified Deep Long Short-Term Memory (MDLSTM). The modification in LSTM output is altered using Improved Spotted Hyena Optimization (ISHO) algorithm. The results are implemented in the working platform of Phyton with the metrics such as specificity, sensitivity, F-Score, Kappa value, Accuracy, BER and Execution time. The analyzed results shows that the implemented methodology is superior in the prediction of heart disease with an accuracy of 98.01% and reduced error rate of 91.11 compared with other existing techniques.