A RAG-based Medical Assistant Especially for Infectious Diseases

• Published in: 2024 International Conference on Inventive Computation Technologies (ICICT) pp. 1128 - 1133
• Main Authors: S, Stewart Kirubakaran, G, Jasper Wilsie Kathrine, E, Grace Mary Kanaga, J, Mahimai Raja, Singh A, Ruban Gino, E, Yuvaraajan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24-04-2024
• Subjects: Artificial intelligence; chatbot; Chatbots; COVID-19; Infectious diseases; knowledge graph; large language model; natural language processing; Recording; Reliability; retrieval augmented generation
• ISSN: 2767-7788
• DOI: 10.1109/ICICT60155.2024.10544639

Infectious diseases like COVID-19 have gained international attention recently. Furthermore, there are significantly fewer doctors per capita in densely populated nations like India, which hurts those in need. Under such circumstances, natural language processing techniques might make it feasible to create an intelligent and engaging chatbot system. The primary objective of the effort is to develop an interactive solution that is entirely open source and can be easily installed on a local computer using the most recent data. Even though there are numerous chatbots on the market, proposed solutions highlight the need to provide individualized and sympathetic responses. Getting Back While the data is stored in the graph database as nodes and relationships, and the knowledge graph is constructed on top of it, augmented generation is utilized to extract the pertinent content from the data. To improve the generator's context, pertinent sections are collected during the question-answering process. This reduces hallucinations and increases the correctness of abstractions by providing external knowledge streams. Furthermore, the research study employs a text-to-speech model that was replicated from a physician's voice recording to narrate the produced responses, thereby augmenting user confidence and interaction. Academic institutions and healthcare organizations can benefit from this work by better understanding the value and effectiveness of applying NLP techniques to infectious disease research.