The digital transformation of healthcare services is driving the development of electronic medical record (EMR) systems capable of supporting structured and efficient clinical data recording. A critical process within EMR systems is ICD-10 diagnosis coding; in conventional practice, this is often performed manually, requiring a high level of precision from coding staff. This study aims to develop a web-based EMR system featuring an artificial intelligence-driven ICD-10 coding recommendation tool. A Research and Development approach utilizing the Waterfall model was employed, encompassing requirements analysis, design, implementation, testing, and evaluation. The system was developed using Laravel, PHP, and MySQL, with integration of the Gemini AI API. Black-box testing across nine scenarios demonstrated that all key features functioned correctly, achieving a 100% success rate. Testing of the recommendation feature against nine diagnoses showed 100% alignment between the system-generated ICD-10 codes and reference codes for clear, specific diagnoses. The study contributes an EMR prototype that integrates diagnosis input, ICD-10 code recommendations, an ICD-10 library, recommendation rationales, and coding history into a unified workflow. This system offers advantages over conventional methods by accelerating and guiding the initial coding process, while still ensuring that final verification is performed by the coder.
Copyrights © 2026