Open Access

ARTICLE

Bayesian Stochastic INLA Application to the SIR-SI Model for Investigating Dengue Transmission Dynamics

Mukhsar^1,*, Andi Tenriawaru², Gusti Ngurah Adhi Wibawa¹, Bahriddin Abapihi¹, Sitti Wirdhana Ahmad³, I Putu Sudayasa⁴

1 Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Halu Oleo Kendari, Kendari, 93232, Indonesia
2 Department of Computer Sciences, Faculty of Mathematics and Natural Sciences, Universitas Halu Oleo Kendari, Kendari, 93232, Indonesia
3 Department of Biology, Faculty of Mathematics and Natural Sciences, Universitas Halu Oleo Kendari, Kendari, 93232, Indonesia
4 Department of Medicine, Faculty of Medicine, Universitas Halu Oleo Kendari, Kendari, 93232, Indonesia

* Corresponding Author: Mukhsar. Email:

Intelligent Automation & Soft Computing 2025, 40, 177-193. https://doi.org/10.32604/iasc.2025.058884

Received 23 September 2024; Accepted 12 December 2024; Issue published 24 February 2025

Abstract

Despite extensive prevention efforts and research, dengue hemorrhagic fever (DHF) remains a major public health challenge, particularly in tropical regions, with significant social, economic, and health consequences. Statistical models are crucial in studying infectious DHF by providing a structured framework to analyze transmission dynamics between humans (hosts) and mosquitoes (vectors). Depending on the disease characteristics, different stochastic compartmental models can be employed. This research applies Bayesian Integrated Nested Laplace Approximation (INLA) to the SIR-SI model for DHF data. The method delivers accurate parameter estimates, improved computational efficiency, and effective integration with early warning systems. The model compared to existing work using Markov Chain Monte Carlo (MCMC) using monthly DHF data from 10 districts in Kendari-Indonesia from 2020–2023. While MCMC requires 10,000 iterations with an 80,000 burn-in, INLA achieves parameter convergence with just 10,000 iterations. The parameter estimation results show that INLA provides a better fit, with the lowest deviance = 105.23, compared to MCMC. Risk analysis using INLA highlights dengue case dynamics from January to May each year. Kadia and Wua-Wua districts consistently show high case numbers, emphasizing the need for targeted interventions in Kendari City. Early surveillance and control efforts are essential to curb mosquito breeding in these areas starting in January. In contrast, the Puuwatu, Kambu, and Kendari Barat districts are sporadic outbreaks, often linked to cases originating in Kadia and Wua-Wua districts.

---

Keywords

---