1Research Scholar, Department of Electronics and Telecommunication Engineering, Shivaji University, Kolhapur, India. Email: mullaraisa23@gmail.com
2Professor & HOD, Department of Electronics and Telecommunication Engineering, Bharati Vidyapeth's college of Engineering, Kolhapur, India. Email: kamalakar.desai@bharatividyapeeth.edu
Background: The healthcare systems of the various countries have encountered many challenges caused by the COVID-19 epidemic, mainly in managing high-risk populations, such as diabetic patients at greater risk of complications. We propose a lightweight and scalable machine learning system for COVID-19 prediction among the high risk DM patients in this work.
Methodology: To enhance the predictive performance and maintain computational efficiency, the introduced approach integrates a feature fusion method, when merging clinical, demographic and comorbidity data. The architecture is designed to be deployed in low-resource settings by means of using small models enabling real-time processing and decision-making. An emphasis on model scalability and low computational cost with a novel effort in pruning method and a multi-modal feature fusion to capture the complex interactions among the COVID-19 risk factors and diabetes related health features are novel contributions.
Results: The proposed model significantly improves upon baseline models in predicting COVID-19 related outcomes including infection risk, hospitalization risk, and mortality, as demonstrated by extensive evaluations over real-world data.
Conclusion: The results are illustrative of how the model can help advance early-detection and targeted intervention approaches for physicians and public health authorities, the researchers said. This paper opens avenues for further investigation in feature-driven predictive healthcare systems, and highlights the importance of scalable, interpretable models for handling challenges related to pandemic-induced healthcare problems, which directly affect the well-being of the vulnerable population.
Keywords: COVID-19, Machine Learning (ML), Lightweight Models, Feature Fusion Approach.
How to cite this article: Mulla RM, Desai KR. Scalable Lightweight Models for COVID-19 Prediction in High-Risk Diabetic Patients. Int J Drug Deliv Technol. 2026;16(4s): 410-425. DOI: 10.25258/ijddt.16.4s.51
Source of support: Nil.
Conflict of interest: None