HealthGuardAI

Inspiration

Diabetes and other lifestyle diseases often remain undetected until serious complications occur. Many people lack access to early screening tools and continuous health monitoring. We wanted to build a solution that uses AI + health data to: Detect risks early Visualize health insights Help users take preventive action Our goal was to shift healthcare from reactive → preventive.

What it does

HealthGuardAI is an intelligent health-risk prediction platform that: 🔐 Provides secure user login & signup 🧠 Uses Machine Learning to predict diabetes risk 📊 Displays interactive health analytics dashboard 📝 Stores user feedback in a database 📈 Shows data visualizations like: Risk distribution Correlation heatmap Health metrics HealthGuardAI is an AI-powered preventive healthcare platform that: 🔐 Provides secure user authentication (login & signup) 🧠 Predicts diabetes risk using Machine Learning 📊 Displays an interactive health dashboard 🧾 Stores user data & feedback using a database 📈 Visualizes: Outcome distribution (diabetic vs non-diabetic Correlation heatmap of medical parameters Key health metrics Users enter their health details (Glucose, BMI, Age, etc.), and the system: Calculates the risk Shows probability Gives instant health insight This enables **early detection and preventive

How we built it

We developed HealthGuardAI as a complete end-to-end AI web application.

🔹 Data Processing & Machine Learning

Used the Pima Indians Diabetes dataset

Cleaned and analyzed the data using Pandas & NumPy

Trained a Random Forest Classifier for accurate predictions

Evaluated the model using train-test split and accuracy score

The model learns a function:

𝑓 ( 𝑥 ) → 𝑦 f(x)→y

Where:

𝑥 x = health parameters

𝑦 y = diabetes risk (0 or 1)

🔹 Frontend & Dashboard

Built using Streamlit, which provides:

Interactive UI

Real-time prediction forms

Dynamic charts & health metrics

Multi-page navigation (Login → Dashboard → Predictor → Feedback)

Challenges we ran into

database integration We used SQLite for lightweight and fast local storage. It stores: 👤 User credentials (for login & signup) 💬 User feedback 📊 (Optional future scope: prediction history) This made the application stateful and dynamic, allowing different users to securely access their dashboard. 🔹 Visualization We implemented: 📊 Bar chart – diabetes outcome distribution 🥧 Pie chart – risk ratio 🔥 Correlation heatmap – relationship between medical features These visual insights help users understand their health data, not just see a prediction. ⚙ Running Streamlit in Codespaces At first we ran: python app.py This caused: missing ScriptRunContext session state

Accomplishments that we're proud of

✅ Built a complete end-to-end AI healthcare web app ✅ Implemented secure authentication system ✅ Integrated Machine Learning with real-time prediction ✅ Designed an interactive analytics dashboard ✅ Connected a working database for users & feedback ✅ Successfully ran the project in Codespaces environment ✅ Created a hackathon-ready deployable solution Most importantly, we transformed raw medical data into actionable health insights that support preventive care

What we learned

🧠 AI & Data Science How to train and evaluate real-world healthcare models Feature importance in medical prediction Avoiding overfitting using ensemble methods