The EMT MDS consists of several forms that must be filled out during disasters. The set includes a medical records form combined with a tick-box section and the daily reporting form, both of which are sent to the EMT Coordination Cell (EMTCC). In the medical records section and tick-box form, EMT officers fill in patient data such as identity, medical history, vital signs, physical examination, and therapy management, which must be signed by a doctor. In this study, the tick box only helps to highlight the patient’s condition, which is divided into age, sex, health events, procedures and outcomes, and context, without the patient’s identity to maintain anonymity. Afterward, the form will be collected by the EMTCC to aggregate the data into the daily reporting form. The daily reporting form is part of the EMT MDS designed by the WHO to be used in the EMTCC office. The daily reporting form itself has already been tested during disasters in some countries [13,18,19]. It contains basic information regarding the team, the location to which the team was assigned, daily summaries of the facility, accumulation of data from the tick boxes as MDS statistics, and information regarding crucial needs and risks [20].

This study extracted data elements from the medical records form and daily reporting form. We obtained 2 variations of medical records from the WHO EMT MDS and the ASEAN EMT MDS, which are adaptations for ASEAN countries developed by the ARCH Project. These forms are illustrated in Figure 1.

We chose this medical record form for Indonesia since the Indonesian government, as a member of ASEAN states, has already been trained to use the EMT MDS during regional collaboration drills on the ARCH Project [21,22]. This project aims to strengthen disaster health management in ASEAN countries and meet the WHO EMT MDS as a standard operational procedure [23]. Thus, digitizing this set will be an appropriate choice.

Next, we manually mapped these data elements onto the base resources. Based on the results, we conducted a gap analysis between the data elements and FHIR resources. We divided the EMT MDS elements into the medical record and daily reporting form, including the tick-box section. Next, we examined whether each data element and the FHIR base resources matched.

Profiling is the process of defining the profile created by setting the cardinality and creating an extension to the data elements that have definitions in the resource base. We used the Forge application since it is a common tool used in profiling research [24,25]. Next, we validated the profiles using the SIMPLIFIER.NET website [26].

We implemented the profiles in the HAPI framework using an open-source FHIR server [27]. We used the Insomnia application, an open-source application that performs the API testing and development process, to perform the POST request to the server with body content using JSON [28].

Since our research did not involve human subjects and no actual patient data were included in this study, ethics approval was not required.

We created a list and group of variables extracted from the WHO and ASEAN EMT MDS medical records. The data were divided into 2 categories: medical and daily reporting forms. Subsequently, the base FHIR was mapped. The list of variables and mapping process results from the WHO EMT MDS and ASEAN EMT MDS medical records forms are shown in Table 1. Meanwhile, Table 2 shows the variables in the daily reporting form mapped to the FHIR resources.

From the 40 resources available in the FHIR ID core, we selected 10, 14, and 9 resources for the WHO EMT MDS, ASEAN EMT MDS, and daily reporting form, respectively, as displayed in Table 3. The total numbers of variables from the WHO EMT MDS medical record form, the ASEAN EMT MDS medical record form, and the daily reporting form with variables that matched the FHIR resources are listed in Table 4.

From the gap analysis, we found several data elements that did not match the FHIR resources. Most data came from the daily reporting form. The list is presented in Table 5 . We created extensions for the age, relationship, and estimated date departure elements for this problem. Age did not contain a data element in the FHIR resources. Because age is related to patients, we created an extension of the patient resource to define it. The JSON file is shown in Figure 2.

Relation status is defined as whether the patient’s condition is related to a disaster event. Thus, an extension of the condition resource is deemed appropriate to explain this condition. By definition, the estimated date of departure is when the team ends its service at specific locations. This allows the EMTCC to plan another EMT if necessary. Because this definition differs from the operational service hours in which the data element is available in the Location resource, we added an extension under the Organization resource to define the end period of the EMT service.

Sex in medical records uses a different value set from FHIR, divided into 3 groups: male, pregnant female, and nonpregnant female. Although sex is only divided into 2 groups, it adds pregnancy status to the female criteria. Since the pregnancy status already has a code in Logical Observation Identifiers Names and Codes (LOINC; “pregnancy status reported,” code 11449‐6), we bundled the Patient and Observation resources for this variable.

We validated the resources for the FHIR ID core using SIMPLIFIER.NET. Example results are shown in Figure 3. The proposed disaster FHIR profile is implemented using the HAPI framework. We used the Insomnia application to perform requests on the FHIR REST server.

We created several mock datasets based on the SATUSEHAT public example [29]. We managed to have samples for each resource. However, we added extensions that are not defined in the SATUSEHAT platform. Afterward, we validated the disaster profile using the HAPI FHIR server. We created a JSON file for each resource to implement the POST protocol. The POST protocol validates the request body and stores the resources in a database. We successfully validated 15 resources used in this profile (an example message is shown in Figure 4).

Data collection plays a significant role during disasters. With major calamities requiring support from international aid workers, implementing a standard data format can greatly facilitate the collection of health-related information. We chose the WHO and ASEAN EMT MDS forms to map onto the FHIR profiles. Although the EMT MDS is a standard developed by the WHO for data collection during disasters, the technical implementation designs for disaster situations in which data are transferred manually using an Excel spreadsheet [18]. As daily reports from various sources may be used by the MoH or EMTCC to make decisions, shifting from paper-based documentation to a digital process may increase the efficiency and effectiveness of collecting daily reports from various sources. In this study, we successfully created and validated the profile to the FHIR ID core profile using base resources. A FHIR profile has generally been created for specific use cases; however, few studies have performed disaster-related profiling. Several studies have created profiles related to disasters, such as COVID-19 [30,31].

To the best of our knowledge, this is the first study that has developed FHIR profiles for natural disasters. Our profiles may standardize health data collection, resulting in real-time actionable insights during disaster scenarios. For example, the interoperability facilitated by FHIR allows instant data sharing between disparate health systems, ensuring that international aid workers and medical facilities can access up-to-date patient records, treatment histories, and socioeconomic backgrounds. This seamless exchange of information is essential for coordinating effective and timely responses, significantly affecting survival rates and recovery outcomes.

Another advantage of using the FHIR format is that it optimizes the usability and quality of the health data. The EMT MDS form only records the number of disaster incidents. For example, if a patient visited twice in a day, 2 reports were generated [20]. Using FHIR, we can retrieve existing patient data from the database. The data entered during a disaster can be saved on the SATUSEHAT platform and reused when referring disaster survivors to nearby health care facilities [32]. By doing so, the quality and safety of patient care can be enhanced. These interoperability features have also enabled us to improve the efficiency of clinical data utilization during and after disasters. Ayaz et al [33] discussed the implementation of FHIR applications in their systematic literature review. Balch et al [34] defined the use of FHIR data standards in machine learning–enabled clinical information systems. Thus, using the data available on the SATUSEHAT platform would be helpful for researchers to perform studies related to disasters, such as identifying patterns of disease outbreaks and improving decision-making processes.

We mapped over 80% of the WHO and ASEAN EMT MDS medical record variables onto FHIR resources. Meanwhile, the variables that cannot be mapped to FHIR resource are mainly in the tick-box sections. Variables that did not have data elements, such as age and sex, were due to the characteristics of the EMT MDS form itself. The EMT MDS form was created to facilitate the collection of data and reports on the conditions occurring in the field [35]. This report form is an accumulation of the various daily report forms used by the EMTCC and MoH to provide care and treatment. Thus, age and sex were classified in a classification format to simplify grouping.

This study faced a notable challenge in developing a comprehensive value set for EMT MDS tick-box variables. FHIR uses a standard value set to define medical terminology, such as Systematized Nomenclature of Medicine–Clinical Terms (SNOMED-CT) or LOINC codes. We could not define the value options for health events and procedure variables because the value options from these variables were a group of several conditions. For example, health events have a “major extremity injury” option, which by definition indicates any upper- and lower-extremity injury requiring hospitalization and/or spinal or general anesthesia [36]. However, SNOMED-CT did not have codes for this category; it did have codes to define each of the upper- and lower-extremity injuries (the detailed list is available in Multimedia Appendix 1). Some studies have reported similar findings when mapping the values to SNOMED-CT or LOINC codes [37,38]. A previous study also mentioned that SNOMED-CT requires disaster codification adjustments [39]. One solution that has been done is to submit new codes to the SNOMED-CT or LOINC committee. However, this process is often prolonged without a guarantee of acceptance of the proposed code [40]. An alternative, more immediate solution we considered was to create a new value set that aligns with the WHO EMT MDS standards [41]. This approach aims to bridge the current gap in codification and facilitate more accurate data representation and analysis for disaster health management.

Finally, because this study was based on the concept of the mapping process, we did not perform a functionality test using real data. We used a mock-up dataset to test the POST and GET functions on the FHIR server. In the future, we will conduct tests using data from the SATUSEHAT platform by opening up opportunities for collaboration with the Indonesian MoH. In addition, future work should focus on FHIR-based applications using disaster profiles for health data collection during disaster events.

This research aims to facilitate data collection using international standards, such as the WHO EMT MDS, and transfer it to the SATUSEHAT platform using FHIR as an interoperability standard. The proposed disaster profile was successfully implemented and validated using the FHIR server. This research will be beneficial for facilitating data collection using international standards during disasters and transferring it to a national platform for health care activity using FHIR as an interoperability standard.