基于FTA和Bayesian网络的血液透析感染风险评估

Infection Risk Assessment of Hemodialysis Based on Fault Tree Analysis and Bayesian Network

血液透析感染风险评估有助于提高血透质量以及保证患者安全。血液透析感染风险受大量不确定性、多状态因素的影响而呈现出动态性,传统故障树分析(FTA)的确定性和二态性使其在评估该类风险时存在缺陷。通过将血液透析感染FTA静态模型映射为Bayesian网络对血液透析感染风险中存在的不确定性关系和多状态变量进行建模,利用Bayesian定理进行概率更新从不同角度识别易导致感染的关键事件,并且考虑患者异质性实现血液透析感染风险的动态评估。通过实例验证了该方法的可行性和有效性,与单一FTA方法相比该方法能够从多个角度对风险进行全面地分析与评估。

Infection risk assessmem of hemodialysis contributes to hemodialysis quality and patient safety. Influenced by a large number of uncertain and multi-state factors, hemodialysis infection risk is therefore characterized with dynamic nature. Traditional fault tree analysis is unable to solve these problems due to its two-state hypothesis and determinacy. A corresponding Bayesian network is mapped firstly from the process-oriented hemodialysis infection fault tree to model uncertain relationships and multi-state variables. Bayesian theorem is then used for probability updating to identify the weak points of hemodialysis infection from different angles and conduct dynamic risk assessment given patient heterogeneity. An instance is illustrated for the feasibility and effectiveness of the proposed method. The outcomes verified that the proposed method can analyze and assess risk from various angles more comprehensively compared to the single FTA.