Dynamic risk assessment in healthcare based on Bayesian approach

Abstract

Risks related with healthcare are always dynamic, and they are affected by situations of patients, human errors in treatment and even the states of medical devices. This paper proposes a dynamic medical risk assessment model, for capturing the impacts of factors on the occurrence of adverse events. In this model, a static fault tree is established to show risk scenarios. Dynamic Bayesian network and Bayesian inference are introduced to analyze the operations of medical devices, in consideration of their failures, repairs, and human errors over time. Hemodialysis infection is taken as the case to verify that the proposed method is helpful to demonstrate the changes of medical risks with time, and to identify the critical events contributing to the occurrence of the adverse event at different moments. These findings can act as the basis to assign and adjust safety measures.