yogeshrajkumar/cardiointel

CardioIntel AI is a web application that assesses a person's cardiovascular risk based on clinical measurements they enter. A patient or healthcare professional fills in values like blood pressure, cholesterol, and other biomarkers, and the tool produces a risk estimate along with an explanation written in plain language. The risk prediction comes from three machine learning models working together: Random Forest, XGBoost, and LightGBM. These models were trained to detect patterns in clinical data linked to heart disease. Combining their outputs through a soft-voting approach is intended to produce more reliable estimates than any single model alone. To help explain why a particular risk level was assigned, the tool uses two interpretability techniques called SHAP and LIME, which surface which measurements contributed most to the result and how much weight each one carried. The written explanations and chat responses are generated by Google Gemini. Gemini takes the raw risk scores and turns them into a clinical narrative: it describes how different factors interact (for example, how elevated blood pressure compounds the risk from high cholesterol), assigns an urgency level such as immediate, moderate, or routine, and suggests next steps. The app also includes a chat interface where a user can ask follow-up questions about diet, exercise, and lifestyle choices based on their specific results. Past assessments can be saved and revisited through a MongoDB database, and the tool can produce a printable summary report. The frontend is built with React and the backend runs on Flask. The README includes a disclaimer stating that CardioIntel is intended for educational and research purposes only and is not a substitute for professional medical advice or diagnosis.