Kazumi Kajiyama’s research while affiliated with Kitasato University and other places

Hospitals, especially regional disaster base hospitals, play a critical role in saving lives during a disaster. Therefore, it is important for the hospitals to conduct disaster response exercises and thoroughly evaluate the results to understand the current level of response capability and to identify potential problems in disaster response and hospital business continuity. However, because data collection and analysis of disaster exercise requires a lot of manpower and time, such evaluation has not been well conducted so far. Aiming to develop evaluation indices of exercise performance, we collected data on patient and document flow, inter-departmental communication, and exercise participant behavior using video cameras, voice recorders, and NFC tags. Focusing on inter-departmental communication, this paper describes the data collection using voice recorders attached to PHS and reports on an attempt to use a large-scale language model to automatically classify the verbal data into several performative verbs.Methods: We collected data during disaster response exercises conducted at a major hospital in Kanagawa Prefecture, Japan, in 2022 and 2023. This hospital is designated as a regional disaster base hospital, which is expected to play a central role in regional disaster medicine. These two exercises were both designed to accommodate mass casualties caused by a major earthquake. We used a voice recorder that can record a conversation via PHS/smartphone; we can record the voice from both sides with a single voice recorder. We attached this voice recorder to several exercise players in different departments to collect communication data on information sharing and command and control. The conversation data was transcribed and used for the further analysis. We analyzed the conversation content from the viewpoint of performative verbs to calculate the anticipation ratio of inter-departmental communication. Usually, this kind of analysis is done manually, which requires a lot of man-hours. On the other hand, in this analysis, we applied the GPT-4.0 language model to automatically classify the conversations into nine performative verbs: greet, inform, acknowledge, request, query, accept, declare, confirm, and suggest. Results and Discussions: We compare the results obtained by GPT with those of human analysts to evaluate the reliability of the classification. We confirmed that the kappa value is 0.73, which indicates that there is substantial agreement between the GPT and manual classification. Then, we calculated the anticipation ratio, which is the ratio of push to pull information, and is often used as a rough indicator of efficient information sharing. By comparing the ratio between 2022 and 2023, we found that the ratio was higher for the command post in 2023, indicating that the command post in 2023 proactively provided information to other departments in advance before they were asked. Conclusion: Through this study, we confirmed that inter-departmental conversations in the exercise can be clearly recorded with the voice recorder attached to the PHS. We also confirmed that a large language model can be used for the classification by performative verbs, thus saving man-hours in calculating the anticipation ratio.

Disaster drills are effective for understanding and mastering business continuity plans (BCP) and response plans, and for identifying and improving problems, but the time and effort required to collect data is an obstacle. In this study, we developed a smartphone application to easily record the flow of patients and documents: NFC tags are attached to simulated patients and documents used in training, and they are read at each scene using a dedicated application. This tool can record when and where the patient was in the area, and when and where the documents were issued and received. We also propose several visualization and quantitative analysis methods that can be performed using the acquired data.

In this study, we developed a simulation model of detailed in-hospital disaster response to a mass casualty incident based on the analysis of related documents and actual in-hospital disaster response training, aiming to assess the hospital’s response capacity under various disaster situations. This simulation model includes detailed models of patients, floor configurations, resources, and response tasks, which consider resource requirements for the treatment of different patients with various injuries and physical conditions. The model covers patients’ arrivals to hospitalization or discharge. We conducted simulations of the target hospital to test two resource allocation strategies under two patient scenarios. By comparing the results under different resource allocation strategies, we found that the X-ray photography examination capacity could become a fundamental bottleneck in responding to mass casualty incidents. Also, we found that the appropriate resource allocations and quick replenishment could alleviate the negative effect of resource shortages and maintain a higher performance. Furthermore, the results show that the length of stay can be heavily affected by the patients’ configuration. As a result, by monitoring and anticipating the situation, a resilient and responsive resource allocation strategy must be prepared to handle such uncertain disaster situations.

Hospitals need to prepare for disasters. For this purpose, they need to formulate a BCP and a response plan and conduct training and exercise. Through training and exercise, they are expected to learn necessary actions and procedures, as well as to find problems in the BCP and response plan. It is, however, not easy to comprehend and evaluate the entire training and exercise process, particularly as the scale of the training and exercise becomes larger. Besides, it also becomes more difficult to find important problems and their causes in training and exercise. To solve this problem, it is necessary to record the events and activities during training and exercise as much detail as possible and analyze them from various viewpoints. This paper describes the comprehensive data recordings in a disaster response exercise in a big hospital in Kanagawa Prefecture, Japan. This hospital is one of the largest hospitals in the region and is designated as a disaster-base hospital which is expected to play a central role in disaster medicine in the area. The target exercise is an annual exercise for a mass casualty incident assuming a big earthquake, in which 46 injured patients are transported to the hospital. We collected data on the following three aspects of the exercise: judgment and decision-making, such as triage and diagnosis and treatment, the in-hospital flow of patients and medical instruments, and the flow of information in and among different areas and rooms. Regarding the data on judgment and decision-making, we collected the documents used in actual disaster medicine, such as triage tags and medical charts. For the other data, we used action cameras attached to hospital staff, station cameras to observe the activities in specific areas, IC recorders attached to patients and several important hospital staff, PHS-IC recorders attached to area leaders to record inter-area communication, Zoom recordings for online communications, and several observers to record inter-area transfer of patients.This paper also describes preliminary analyses of the data recorded in the exercise, including the accuracy of 1st and 2nd triage, the details of patient flow, and the length of stay in each area.Finally, this paper describes the data analysis planned in the next step, such as network analysis of inter-area communication, task analysis of the activities in each area, and so on. In addition, we will discuss the possibility of utilizing the study to reproduce the exercise using the agent-based simulation we developed for a deeper understanding of the entire process of the in-hospital disaster medicine process.