Analysis on Performance of Semi-Real-Time Transcription on Health App

The fragmentation of health information systems has hindered seamless data exchange in healthcare services, posing challenges in achieving efficient, real-time access to patient health data. This study addresses these issues by evaluating the implementation of a semi-real-time transcription system using a distributed architecture with a message broker to enhance the documentation of clinical examination processes in primary health services. The research focuses on assessing the feasibility of this approach by measuring key performance indicators such as latency and word error rate (WER) during clinical conversation transcription. Audio recordings were segmented into shorter fragments to simulate real-time transcription, and data transmission was conducted through the Advanced Message Queuing Protocol (AMQP). The findings demonstrate that the average latencys achieved by semi-real-time transcription system is 7.7s, indicating its potential suitability for clinical documentation applications that do not require strict real-time processing. However, glitches observed in the message broker's forwarding mechanism suggest the need for further optimization to ensure consistent performance. Future work aims to develop strategies to minimize transcription errors, enhance reliability, and improve the overall integration of this system in healthcare settings. This study contributes to the advancement of health information systems by offering architecture into scalable and effective methods for semi-real-time data transcription and interoperability.