A COMPARATIVE STUDY OF SYNCHRONOUS AND ASYNCHRONOUS COMMUNICATION IN IMPLEMENTING FHIR STANDARDS FOR ELECTRONIC MEDICAL RECORD (EMR) INTEROPERABILITY ON SATUSEHAT INDONESIA PLATFORM

• Main Author: Dewi, Sari
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/83738
• Institution: Institut Teknologi Bandung

Amid advances in digital technology, interoperability as a concept that allows different systems to communicate and share data has become crucial, especially in the health sector to increase efficiency in health service and improve patient care. In the health sector's interoperability context, standards such as Fast Healthcare Interoperability Resources (FHIR) play an important role. In Indonesia, health data interoperability has become mandatory with Minister Health Regulation number 24 of 2022 concerning Medical Records and the launch of the SATUSEHAT platform as a health data exchange ecosystem. However, with the need to quickly implement interoperability in Indonesia, a large amount of health data from Electronic Medical Records (EMR) will need to be exchanged. The communication method used in interoperability to the SATUSEHAT server is a synchronous communication that uses the Representational State Transfer Application Programming Interfaces (REST APIs). Using REST APIs has risks when handling large amounts of data, namely in the form of high latency, risk of server overload, and lack of stability to handle spikes in requests effectively. This research aims to design an asynchronous communication architecture for health data interoperability and conduct a comparative study between synchronous and asynchronous communication in implementing FHIR standards for EMR interoperability in the SATUSEHAT platform. In the design of the asynchronous communication architecture, this research incorporates RabbitMQ as a message broker to create a message queue before sending messages via the REST protocol to the SATUSEHAT server. Performance testing of response times, throughput, and error rate for the two interoperability architectures was conducted by sending requests using GET, POST, and PUT methods to several resource endpoints of the SATUSEHAT server. This testing involved four test scenarios with varying data and numbers of users. The tools used for performance testing include Apache Jmeter, Gatling, and k6. iv The test result indicates that asynchronous communication's response time performance is better than synchronous for data transfer up to 200 data in scenario 1. The response time shows a reduction of 23% for 50 data, 6,52% for 100 data, 2,68% for 150 data, and 4,94% for 200 data, demonstrating that asynchronous communication has a better response time compared to synchronous communication. Similarly, in scenario 2, testing with 10 users shows a response time reduction of 6,05%. However, with larger amounts of data or more users, the response time increases, indicating that asynchronous communication is slower compared to synchronous communication. This increase is due to the message queuing mechanism and resending messages after request failure due to rate limits, which lead to increased response times. JMeter performance test results show that asynchronous communication throughput is slightly higher than synchronous communication, with increases of 30.43% for 50 data, 6.67% for 100 data, 10% for 150 data, and 66.67% for 250 data. Similarly, in scenario 2, throughput increased by 7.13% for 10 users, 24.85% for 20 users, and 3.53% for 30 users. However, with more data or users, JMeter throughput in asynchronous communication is lower than in synchronous communication. A varying error rate is observed in synchronous communication due to the rate limit. When the rate limit is reached, the request will be rejected by the SATUSEHAT server. Under high load and surge requests sent immediately after the rate limit is reached, the error can reach 100 %. Performance in asynchronous communication is better, achieving an error rate of 0% across all test scenarios due to the mechanism of returning messages to the queue and resending them after a delay to the SATUSEHAT server following request failures. The overall results indicate that asynchronous communication is more reliable in handling high load and request spikes.