Towards Context-Aware Clinical Decision Support with Retrieval-Augmented Generation

This research explores the Retrieval-Augmented Generation (RAG) method in the application of the Large Language Model to help Indonesian doctors in clinical decision making. By using the Clinical Practice Guidelines implemented by the Ministry of Health of the Republic of Indonesia, the answers generated by the model are expected to be in accordance with existing standards so that doctors are able to make more precise decisions. Two models are implemented with RAG in this research, GPT-4o mini by OpenAI and Claude 3.5 Haiku by Anthropic. From the evaluation that has been done in this research, it is proven that the models implemented with RAG are better in terms of the accuracy of disease diagnoses for the symptoms given to conformity with the Clinical Practice Guidelines.