MediSure Risk Analyzer using Random Forest Classifier
CHAPTER 1
INTRODUCTION
1.1 Overview of the Study:
The rapid advancement of Artificial Intelligence (AI) and Machine Learning
(ML) technologies has created new possibilities in healthcare, especially in the realm of
disease prevention and early diagnosis. Non-communicable diseases such as
cardiovascular conditions and diabetes have become the leading causes of death
globally. Early prediction and timely intervention can significantly mitigate the risk and
reduce the burden on healthcare systems. Despite the existence of diagnostic tools and
medical records, there remains a gap in accessibility, personalized insight, and
continuous health monitoring— especially for individuals without regular access to
medical professionals.
This project, Medisure Risk Analyzer using Random Forest Classifier, addresses
that gap. It is a real-time health risk prediction platform that integrates AI models,
conversational AI (Gemini), explainable AI visualizations (SHAP), and Firebase-based
cloud support into a single, accessible web application. The system enables users to
input basic medical parameters and lifestyle information to get risk assessments for
diseases like heart disease and diabetes. It combines the power of machine learning
models with the capabilities of generative AI to deliver meaningful insights,
personalized recommendations, and downloadable reports. The solution is particularly
beneficial in remote or under-resourced areas where access to specialist healthcare is
limited.
1.2 Objectives of the Project:
The primary objective of this project is to provide a comprehensive, user-friendly
platform that allows individuals to assess their risk of chronic illnesses using AI. The
goals are:
• Develop robust machine learning models for heart disease and diabetes
prediction based on standard datasets.
DEPT OF CSE, KIT, Tiptur 2024 - 2025 Page 1
MediSure Risk Analyzer using Random Forest Classifier
• Build an intuitive web interface using Streamlit that allows users to input data
and receive predictions in real-time.
• Integrate Google Gemini AI to interpret results and provide lifestyle-based
recommendations tailored to user health inputs.
• Ensure transparency of predictions using SHAP values and graphical
explanations, helping users understand why a certain prediction was made.
• Allow users to download a professionally formatted PDF report summarizing
their risk and health suggestions.
• Enable secure user authentication and save historical predictions using Firebase
Authentication and Firestore.
• Incorporate voice-based interaction and accessibility options to support
differentlyabled users.
• Add contextual features such as real-time weather and air quality data using
geolocation APIs to enhance health context awareness.
• Maintain modular architecture for future expansion to include more diseases like
cancer, kidney disease, or mental health conditions.
1.3 Scope of the Project :
This project serves as a proof of concept and potential base model for AI-integrated
preventive healthcare solutions. The scope includes the following dimensions:
• Technical Scope:
o The project is built using Python, Streamlit, Scikit-learn, Firebase, and
Google Gemini API. o Models used are trained and evaluated on standard
datasets like the UCI Heart Disease and Pima Indians Diabetes datasets.
o It supports real-time interactions, PDF
generation, AI-driven recommendations, data visualizations, and
Firestore-based history tracking. Functional Scope:
o The app can predict heart disease and diabetes risk with high accuracy
based on user-input parameters such as age, blood pressure, cholesterol
DEPT OF CSE, KIT, Tiptur 2024 - 2025 Page 2
MediSure Risk Analyzer using Random Forest Classifier
levels, and more. o It provides AI-generated health tips based on additional
lifestyle data such as diet, stress, smoking habits, and sleep.
o The system logs predictions along with confidence scores and visualizes
trends over time using plots and charts.
• User Scope:
o The application is designed for individual users, healthcare practitioners,
and wellness consultants. o The system includes Firebase-based login and
supports multiple user profiles and personalized dashboards.
• Scalability Scope:
o Designed for cloud deployment, the application can scale to support
thousands of users simultaneously. o The architecture allows future inclusion
of additional diseases and integration with wearable devices for real-time
health tracking.
• Limitations & Assumptions:
o The current model predictions are based on structured input data and do
not cover medical imaging or genetic data. o The system assumes users can
interpret basic health metrics and seek medical advice based on the AI
suggestions provided. o It is not intended to replace professional diagnosis
but to supplement early awareness and self-monitoring.
DEPT OF CSE, KIT, Tiptur 2024 - 2025 Page 3
