应急情景模拟演练的即时评估与反馈:经验与启示OA

Emergency simulation exercises are widely used in emergency management training to improve participants'response capacities.The evaluation and feedback component of these exercises,however,has not re-ceived sufficient attention.Two types of approaches are prevalent in the assessment and feedback of emergency simulation practice in China,structured evaluation methods and expert-led post-exercise debriefs.The former mainly uses pre-designed scales and scoring criteria to measure performance quantitatively,while often compro-mising the authenticity of scenarios.The latter relies on unstructured observation and commentary provided by experts,which is very challenging to standardize comments from experts.This creates a tension between the au-thenticity of scenarios and the structuration of feedback in assessing and providing instant feedback for partici-pants in drills.To address the gap,we review existing debrief and assessment methods,including post-exercise de-brief models such as Plus-Delta,the 3D model,PEARLS,and GAS;self-report scales assessing individual psy-chological states and group dynamics;behavioral observation tools;and recent applications of multimodal data analysis in the fields of medicine,aviation,and education.We also report our instant assessment practices used in two scenario-based tabletop exercises.The first scenario is a public health emergency,and participants from na-tional and provincial disease control agencies are assigned roles as expert group members with tasks on assessing the situation and providing recommendations on control measures in the face of an unknown viral variant.The second scenario is an interagency funding allocation and strategy-making with roles as leaders from agencies in a county.In each exercise,we collected video data of each participant.About one hour after the simulation,we pro-vided each participant with an assessment report documenting their performance in the exercise.This instant feedback in the report has three types of analytical results.The first assessed individual discursive capability,a participant's competence to construct persuasive and institutionally grounded arguments in the simulation.The second mapped out group interaction dynamics via network analysis.The third captured decision-making and conflict resolution dynamics.We also identified three remaining challenges in the assessment and feedback.First,synchronizing multimodal data from different sources is challenging in such a short analytical time win-dow.Second,natural language processing models cannot reliably decode the decision rationales embedded in domain-specific terminology.Third,behavioral network analysis alone cannot capture the complex dynamics of interpersonal interactions.We conclude by proposing future directions of instant assessment and feedback.Tech-nically,we call for new methods of integrating data sources and developing domain-specific language models for emergency management contexts.Meanwhile,more rigorous testing is required for competency and performance assessment frameworks fitting exercise scenarios,and support teams including observers,analysts,and domain experts are another critical factor that might influence data collection and analysis.