Early Prediction of Multiple Diseases and LLM-Based Recommendation System

Abstract

Preventive patient care is the significant need in the present day scenario. Existing ML based healthcare systems lack in identifying the multiple diseases simultaneously and accurately so that supporting early prediction and further patient cares. The proposed system predicts multiple diseases with a proven machine learning approach that predicts diabetes, heart, and Parkinson’s disease. The system is developed in Python and makes use of libraries like Streamlit for the construction of an interactive web application. Pickle is employed for model serialization. Therefore, by integrating advanced machine learning algorithms for each disease this system helps in enhancing the predictive accuracy and aids in prior identification of vulnerable individuals. Support Vector Machine (SVM) algorithm is used for predicting diabetes and Parkinson’s disease, wherein linear kernel is employed to strengthen classification tasks by finding optimal hyperplanes in high-dimensional data. Pregnancies, glucose levels, blood pressure, insulin, BMI, and age is evaluated by the diabetes prediction model, whereas the Parkinson’s model uses clinical features to predict the presence of disease. Logistic Regression is used for heart disease prediction, providing probability- based predictions which is based on key clinical features that also adopts a binary classification threshold. In addition to predictions, this system also promotes a comprehensive approach to health management by incorporating dietary recommendations that aids in prevention, care and overall well-being using LLM. With a user-centric interface for health data input, integration of diverse patient data and robust pre-processing, this system showcases its effectiveness in early detection, strategies for prevention and management of diseases.