Affiliations: Department of Computer Science and Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Chennai
Correspondence:
[*]
Corresponding author. S. Thumilvannan, Research Scholar, Department of Computer Science and Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Saveetha University, Chennai, Tamilnadu, India-602105. E-mail: [email protected].
Abstract: The survival of patients’ deaths owing to Heart Disease (HD) could be improved with the assistance of an enhanced approach for predicting the risk of diabetes and HD. Nevertheless, such schemes are developed rarely. Thus, this paper proposes a new Power Lognormal Distribution-Semi-Supervised Learning-centric Restricted Boltzmann Machine (PLD-SSL-RBM) diabetes and HD risk level prediction model for IoT data. The missing data are removed by partial Derivation of the Hamilton-Cluster Centered-K-means Clustering (DH-CC-KC) to efficiently train the classifier and then, the data are aggregated. Next, to reduce the dataset size, the features are reduced with Shell Sort-Principal Component Analysis (SS-PCA). Then, the fuzzy rule-based decisions are created with the T-test-centric Uniform Distribution-Elephant Herd Optimization Algorithm (T-test-UDEHOA) Correlated Features (CF) to classify the risk levels accurately. Lastly, the risk levels of HD and diabetes are predicted; in addition, by employing the Elliptic Curve Cryptography (ECC)7encryption technique, the data is securely stored on the medical database. The proposed risk prediction model’s performance is analyzed on the Framingham dataset. As per the experimental outcomes, when analogized to the prevailing methodologies, the proposed technique attained a higher accuracy of 99.55%.
Keywords: Internet of Things (IoT), heart disease and diabetes risk, Restricted Boltzmann Machine (RBM), correlated features, Elephant Herd Optimization Algorithm (EHOA), Correlated Feature (CF)
