TriKSV-LG: a robust approach to disease prediction in healthcare systems using AI and Levy Gazelle optimization

Kavitha Dhanushkodi, Prema Vinayagasundaram, Vidhya Anbalagan, Surendran Subbaraj and Ravikumar Sethuraman

Computer Methods in Biomechanics and Biomedical Engineering, 2025, vol. 28, issue 11, 1783-1799

Abstract: A seamless connection between the Internet and people is provided by the Internet of Things (IoT). Furthermore, lives are enhanced using the integration of the cloud layer. In the healthcare domain, a reactive healthcare strategy is turned into a proactive one using predictive analysis. The challenges faced by existing techniques are inaccurate prediction and a time-consuming process. This paper introduces an Artificial Intelligence (AI) and IoT-based disease prediction method, the TriKernel Support Vector-based Levy Gazelle (TriKSV-LG) Algorithm, which aims to improve accuracy, and reduce the time of predicting diseases (kidney and heart) in healthcare systems. The IoT sensors collect information about patients’ health conditions, and the AI employs the information in disease prediction. TriKSV utilizes multiple kernel functions, including linear, polynomial, and radial basis functions, to classify features more effectively. By learning from different representations of the data, TriKSV better handles variations and complexities within the dataset, leading to more robust disease prediction models. The Levy Flight strategy with Gazelle optimization algorithm tunes the hyperparameters and balances the exploration and exploitation for optimal hyperparameter configurations in predicting chronic kidney disease (CKD) and heart disease (HD). Furthermore, TriKSV's incorporation of multiple kernel functions, combined with the Gazelle optimization strategy, helps mitigate overfitting by providing a more comprehensive search space for optimal hyperparameter selection. The proposed TriKSV-LG method is applied to two different datasets, namely the CKD dataset and the HD dataset, and evaluated using performance measures such as AUC-ROC, specificity, F1-score, recall, precision, and accuracy. The results demonstrate that the proposed TriKSV-LG method achieved an accuracy of 98.56% in predicting kidney disease using the CKD dataset and 98.11% accuracy in predicting HD using the HD dataset.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://hdl.handle.net/10.1080/10255842.2024.2339479 (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:taf:gcmbxx:v:28:y:2025:i:11:p:1783-1799

Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/gcmb20

DOI: 10.1080/10255842.2024.2339479

Access Statistics for this article

Computer Methods in Biomechanics and Biomedical Engineering is currently edited by Director of Biomaterials John Middleton

More articles in Computer Methods in Biomechanics and Biomedical Engineering from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().