Towards Automated Medical Pre-Diagnosis: Disease Prediction Chatbot Using Ensemble Learning

Abstract

Recent breakthroughs in artificial intelligence are truly transforming the landscape of healthcare diagnostics, particularly in automating the detection of diseases based on the symptoms patients report. This paper presents an innovative and interactive disease prediction chatbot that merges Natural Language Processing (NLP) with ensemble machine learning models, making medical assessments both precise and userfriendly. The system leverages BERT-based embeddings to interpret unstructured symptom descriptions and features a stacked ensemble model that incorporates MLPClassifier, XGBoost, LightGBM, and CatBoost for dependable disease classification. The model exhibits outstanding performance across a number of parameters, including accuracy, precision, recall, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{F 1}$</tex>-score, following training and testing on publicly accessible medical datasets. Through web and YouTube integration, the chatbot also offers customers visual metrics, individualized health advice, and even helps them find local experts and educational materials. With a Tkinter-based graphical user interface (GUI) that utilizes symbols and color codes, it facilitates easy and iterative symptom input, enhancing user-friendliness. This work underscores the thrilling potential of AI-driven conversational agents in supporting early disease prediction, increasing health awareness, and improving access to initial healthcare services.

Ähnliche Arbeiten