🩺 AI Health Assistant

This project is a simple medical symptom checker that predicts the most likely diseases based on user-selected symptoms. It uses a Decision Tree Classifier trained on a publicly available dataset and provides basic precautionary measures. The app is built with Python, scikit-learn, and Streamlit, and also lets users download a PDF summary of the results.

🚀 Key Features

• Predicts the top 2 most probable diseases based on chosen symptoms
• Displays a confidence level for each prediction
• Shows precautions and recommended steps for predicted diseases
• Generates a downloadable PDF report for the selected symptoms and outcomes
• Provides an easy-to-use Streamlit web interface

📂 Folder Structure

AIHealthAssistant/
│
├── app.py                  # Streamlit web app
├── train_model.py          # Script for training the model
├── model/                  # Folder where the trained model is saved after running train_model.py
│   └── dt_model.pkl        # Generated after training
├── data/
│   ├── disease_symptoms.csv
│   └── disease_precautions.csv
├── requirements.txt        # Dependencies list
├── README.md               # Documentation for the project
└── concepts.md             # Beginner-friendly explanation of ML concepts used

Note: The model/ folder will be empty when you first clone the project. You must run:

python train_model.py

to train the classifier and create model/dt_model.pkl before using the app.

🗂 Dataset

• Dataset comes from a publicly available source (Kaggle)
• Contains 100+ symptoms mapped to 40+ different diseases
• Includes precautionary suggestions for each disease in a separate CSV

⚙️ Setup Instructions

# 1. Clone this repository
 git clone https://github.com/yourusername/AIHealthAssistant.git
 cd AIHealthAssistant

# 2. (Optional) Create a virtual environment
 python -m venv venv
 source venv/bin/activate   # macOS/Linux
 venv\\Scripts\\activate     # Windows

# 3. Install the required libraries
 pip install -r requirements.txt

# 4. Train the model
 python train_model.py

# 5. Launch the app
 streamlit run app.py

🖥️ How to Use

1. Pick one or more symptoms from the dropdown menu.
2. Click Predict Disease.
3. View the top 2 predicted diseases, their confidence levels, and suggested precautions.
4. Download a PDF report with all the details.

📊 Model & Evaluation

• Uses a Decision Tree Classifier for prediction
• Achieves around 90% accuracy on test data
• Shows a classification report and confusion matrix during training for evaluation

🔧 Tools and Technologies

• Python 3.10+
• scikit-learn for building and training the ML model
• Streamlit for creating the web app interface
• FPDF for generating downloadable reports

📌 Possible Future Enhancements

• Improve accuracy with Random Forest or Gradient Boosting models
• Add real-time medical APIs for more details on diseases
• Enable user accounts and history tracking

🙌 Credits

• Dataset adapted from Kaggle - Disease Symptom Prediction
• Inspired by the idea of simple AI-based health assistants