Revolutionizing Healthcare Triage: A Comparative Analysis of Machine Learning-Driven Symptom Checkers and Triage Bots for Common Diseases and Skin Conditions

Abstract

Symptoms Checker and Triage Bot application built using Streamlit, offering users a comprehensive tool for diagnosing common diseases and skin conditions. Leveraging machine learning models, the application provides two main modules: the Common Diseases Checker and the Skin Disease Classifier. The Common Diseases Checker utilizes a decision tree classifier trained on symptom-disease datasets to predict potential health conditions based on user-provided symptoms, offering detailed disease descriptions and precautionary measures. Meanwhile, the Skin Disease Classifier employs a pretrained ResNet-50 convolutional neural network to analyze uploaded images of skin lesions and classify them as either melanoma or allergy, aiding in early diagnosis. The Streamlit interface enables seamless navigation between modules via sidebar buttons, ensuring a user-friendly experience. Overall, the application aims to empower users with efficient and accurate health assessments, facilitating informed decision- making and timely medical intervention. ResNet50 outperformed and attained 97.23% accuracy than other baseline models in skin disease prediction. In common disease prediction, K- nearest neighbour model attained 98.87% accuracy than other models.

Ähnliche Arbeiten