Interactive AI Medical Assistant Through Natural Conversations – Disease Prediction And More

Abstract

Artificial intelligence is revolutionizing healthcare across the world. This study introduces the development of a smart medical chatbot that can understand symptom descriptions in casual everyday language and accurately maps them to medical conditions. It can even interpret slang, spelling errors, typographical errors and non-medical terms. Built on a huge database covering with over 444 different symptoms and 837 diseases, it is capable of predicting multiple possible diseases at once just like how real doctors think about overlapping symptoms. The chatbot also consider user specific factors and tailor responses based on age and health history. It can also answer general health questions in plain English using advanced language models. Designed with a user-friendly web interface, it is compatible with both smartphones and computers. Evaluation shows 91.6% accuracy in disease prediction and an 83.1% success rate in answering general health questions, all with an average response time under 200 milliseconds., making it feel like a real conversation. This tool has a strong potential to provide preliminary health guidance, especially in areas and circumstances where consulting a doctor is not easy.