Scholarlink

<p>Infectious disease spread prediction is crucial for timely public health interventions, resource allocation, and outbreak control. Traditional epidemiological models often fail to account for complex socio-economic and environmental interactions, limiting their predictive accuracy. This study presents BaSTRoN (Bayesian Spatio-Temporal Recurrent Network), a novel model that integrates Bayesian spatio-temporal interactions with deep learning to predict infectious disease spread in India. Leveraging 12 years of state-wise weekly disease incidence data alongside socio-economic and environmental indicators, BaSTRoN captures complex dependencies across space and time. Comparative analysis with existing models–including Poisson Lognormal (PLN), Bayesian space-time (B-ST), and Bayesian space-time interaction (B-ST-I)–demonstrates the superior performance of BaSTRoN, achieving an improvement of 9.8% in <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41870_2025_2695_Article_IEq1.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="19" /> </InlineMediaObject> <EquationSource Format="TEX">\(\text {R}^{2}\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mtext>R</mtext> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation> score for a six-week prediction horizon. Spatial patterns and disease hot-spots are visualized using Local Indicators of Spatial Association (LISA), while Granger causality, Structural Equation Modeling (SEM), and explainable AI techniques enhance interpretability. Through an ablation study, this research also offers valuable insights on the impact of individual socio-economic factors contributing to the disease spread.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

BaSTRoN: a Bayesian model for predicting infectious disease spread using socio-economic and environmental factors

  • 期刊论文

  • 发表日期：

    2025年8月11日

查看全文
  • Lakshmi Priya Swaminatha Rao,
  • Aiswarya Suresh,
  • Adithya Muthukumar

摘要

Infectious disease spread prediction is crucial for timely public health interventions, resource allocation, and outbreak control. Traditional epidemiological models often fail to account for complex socio-economic and environmental interactions, limiting their predictive accuracy. This study presents BaSTRoN (Bayesian Spatio-Temporal Recurrent Network), a novel model that integrates Bayesian spatio-temporal interactions with deep learning to predict infectious disease spread in India. Leveraging 12 years of state-wise weekly disease incidence data alongside socio-economic and environmental indicators, BaSTRoN captures complex dependencies across space and time. Comparative analysis with existing models–including Poisson Lognormal (PLN), Bayesian space-time (B-ST), and Bayesian space-time interaction (B-ST-I)–demonstrates the superior performance of BaSTRoN, achieving an improvement of 9.8% in \(\text {R}^{2}\) R 2 score for a six-week prediction horizon. Spatial patterns and disease hot-spots are visualized using Local Indicators of Spatial Association (LISA), while Granger causality, Structural Equation Modeling (SEM), and explainable AI techniques enhance interpretability. Through an ablation study, this research also offers valuable insights on the impact of individual socio-economic factors contributing to the disease spread.