Systematic Review

Covid-19 Outbreak Prediction: A Systematic Review of Mechanistic, Statistical, and Artificial Intelligence–Based Model

Abstract

With the emergence of the COVID-19 pandemic, reliable methods for predicting future disease trends must be developed to improve planning and public health interventions at national and global scales. An accurate and robust method of epidemic forecasting is a necessary foundation for evidence-based public health response and pandemic preparedness efforts. The emergence of the novel coronavirus outbreak sparked an unprecedented rapid development of various kinds of predictive frameworks, including mechanistic models, classical statistics-based approaches, machine learning methods, as well as hybrid or ensemble strategies, each offering unique strengths and weaknesses in different epidemiological situations. This study conducts a comprehensive and integrated evaluation of the mentioned predictive models, with a focus on their underlying assumptions, forecasting performance, and practical significance in real-life scenarios. Although important for generating future scenarios, mechanistic compartmental models such as SIR and SEIR and their variations are increasingly inaccurate when transmission coefficients change due to mutations, differences among populations, and sudden shifts in human behaviour. Despite providing a good shortterm prediction, statistical models struggle with the lack of stationarity in data and potential surveillance biases. Hybrid and ensemble approaches proved to be the most robust methods in times of high epidemiological uncertainty, reliably minimising biases associated with specific models and improving overall robustness. It was found that accuracy in forecasting can only be ensured when data quality, model complexity, forecast period, and structural changes in the epidemics’ evolution process are taken into account. The present review combines all these findings to point out some key limitations in all existing models and highlight possible areas for further research based on advanced data pipelines, interpretable AI and multi-level prediction environments.

Authors affiliation:

Anupreksha Goswami (ORCID): Department of Biotechnology, Techno India University, West Bengal, EM 4, Salt Lake, Sector V, Kolkata-700091, India.
Rojina Khatun (ORCID): Department of Biotechnology, Techno India University, West Bengal, EM 4, Salt Lake, Sector V, Kolkata-700091, India.
Malavika Bhattacharya (ORCID): Department of Biotechnology, Techno India University, West Bengal, EM 4, Salt Lake, Sector V, Kolkata-700091, India.

How To Cite: A. Goswami, R. Khatun, and M. Bhattacharya. Covid-19 Outbreak Prediction: A Systematic Review of Mechanistic, Statistical, and Artificial Intelligence–Based Model. Journal of Applied Health Sciences and Medicine, 6(10):25–34, 2026. https://doi.org/10.58614/jahsm6104