Cite this article：
Qazi, A., Simsekler, M.C.E. & Al-Mhdawi, M.K.S. Prioritizing Indicators for Rapid Response in Global Health Security: A Bayesian Network Approach. Int J Disaster Risk Sci (2024). https://doi.org/10.1007/s13753-024-00570-w
全球卫生安全快速响应中的指标优先级研究：一种贝叶斯网络方法
Abroon Qazi, Mecit Can Emre Simsekler & M. K. S. Al-Mhdawi
摘要：
本研究探讨了一种贝叶斯信念网络（BBNs）方法，开发了两个不同的模型，用于在特定类别和全球卫生安全指数（GHS 指数）的情景下优先考虑与“快速响应和减缓流行病传播”类别相关的七个指标。本文使用的方法包括严格的预处理、增强朴素贝叶斯算法的结构化学习和k折交叉验证，并使用2021年GHS指数的数据进行研究。研究显示了两个BBN模型中的独特视角的重要发现。在交互信息量分析中，“连接公共卫生和安全部门”成为“快速响应和减缓流行病传播”类别中的关键预测指标，而“应急准备和响应规划”指标则在GHS指数情景中占据首要地位。敏感性分析突出了“应急准备和响应规划”和“连接公共卫生和安全部门”在极端状况下的关键作用，而“获取通信基础设施”和“贸易与旅行限制”则在不同模型中表现出不同的含义。BBN模型表现出较高的预测准确性，在“快速响应和减缓流行病传播”类别的极端状态下达到了83.3%的准确率，而在GHS指数情景极端状态中则达到了82.3%的准确率。本研究通过对GHS指数中快速响应维度的各种指标之间的依赖关系建模，并基于交互信息量和敏感性分析强调其相对重要性，为GHS研究相关文献做出了贡献。
关键词：
贝叶斯信念网络；全球卫生安全；指标；减缓；政策影响；快速响应
Prioritizing Indicators for Rapid Response in Global Health Security: A Bayesian Network Approach
Abroon Qazi, Mecit Can Emre Simsekler & M. K. S. Al-Mhdawi
Abstract：
This study explored a Bayesian belief networks (BBNs) approach, developing two distinct models for prioritizing the seven indicators related to the “rapid response to and mitigation of the spread of an epidemic” category within the context of both the specific category and the Global Health Security Index (GHS index). Utilizing data from the 2021 GHS index, the methodology involves rigorous preprocessing, the application of the augmented naive Bayes algorithm for structural learning, and k-fold cross-validation. Key findings show unique perspectives in both BBN models. In the mutual value of information analysis, “linking public health and security authorities” emerged as the key predictor for the “rapid response to and mitigation of the spread of an epidemic” category, while “emergency preparedness and response planning” assumed precedence for the GHS index. Sensitivity analysis highlighted the critical role of “emergency preparedness and response planning” and “linking public health and security authorities” in extreme performance states, with “access to communications infrastructure” and “trade and travel restrictions” exhibiting varied significance. The BBN models exhibit high predictive accuracy, achieving 83.3% and 82.3% accuracy for extreme states in “rapid response to and mitigation of the spread of an epidemic” and the GHS index, respectively. This study contributes to the literature on GHS by modeling the dependencies among various indicators of the rapid response dimension of the GHS index and highlighting their relative importance based on the mutual value of information and sensitivity analyses.
Keywords：
Bayesian belief networks, Global health security, Indicators, Mitigation, Policy implications, Rapid response
文章链接：
https://link.springer.com/article/10.1007/s13753-024-00570-w