Large Language Models (LLMs) have significant potential for facilitating intelligent end-user applications in healthcare. However, hallucinations remain an inherent problem with LLMs, making it crucial to address this issue with extensive medical knowledge and data. In this work, we propose a Retrieve-and-Medically-Augmented-Generation with Knowledge Reduction (ReMAG-KR) pipeline, employing a carefully curated knowledge base using cross-encoder re-ranking strategies. The pipeline is tested on medical MCQ-based QA datasets as well as general QA datasets. It was observed that when the knowledge base is reduced, the model’s performance decreases by 2-8%, while the inference time improves by 47%.

This paper describes the work undertaken as part of the SMM4H-2024 shared task, specifically Task 5, which involves the binary classification of English tweets reporting children’s medical disorders. The primary objective is to develop a system capable of automatically identifying tweets from users who report their pregnancy and mention children with specific medical conditions, such as attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorders (ASD), delayed speech, or asthma, while distinguishing them from tweets that merely reference a disorder without much context. Our approach leverages advanced natural language processing techniques and machine learning algorithms to accurately classify the tweets. The system achieved an overall F1-score of 0.87, highlighting its robustness and effectiveness in addressing the classification challenge posed by this task.