This project provides an ensemble machine learning model to predict the risk of heart disease based on various health indicators. It utilizes a combination of feature selection techniques, resampling methods, and ensemble learning to achieve robust and accurate predictions.
The Ensemble Risk Predictor is designed to assist healthcare professionals and individuals in assessing the risk of heart disease. It utilizes a publicly available dataset from Kaggle, and employs a combination of feature selection (RFE), a resampling (SMOTETomek) technique to handle class imbalance, and an ensemble model (EasyEnsembleClassifier with RandomForestClassifier base estimators) to make predictions.
This project also includes the ability to analyze data from CSV files and provide tailored health advice using the Google Gemini API.
The model is trained and evaluated on the Heart Disease Health Indicators Dataset from Kaggle. This dataset contains a variety of health-related features collected from individuals, and includes a binary indicator for whether a person has experienced a heart attack.
The following steps were taken in building the Ensemble Risk Predictor:
Recursive Feature Elimination (RFE) was used with a Decision Tree classifier to select the most relevant features for predicting heart disease risk.
Because the dataset contains an imbalanced target class, a combination of SMOTE (Synthetic Minority Over-sampling Technique) and Tomek links were used to address the imbalance issue, leading to more robust model training.
An Easy Ensemble Classifier was trained with multiple Random Forest Classifier base estimators.
The model was evaluated using Stratified K-Fold cross-validation to get robust performance metrics.
Before using the project, ensure that you have the following:
- Python 3.6+
- The required Python packages:
You can install these using pip:
kagglehub==0.3.5 pandas==2.2.2 numpy==1.26.4 matplotlib==3.8.0 seaborn==0.13.2 scikit-learn==1.5.2 plotly==5.24.1 imbalanced-learn==0.12.4 joblib==1.4.2 google-generativeai==0.8.3
pip install kagglehub==0.3.5 pandas==2.2.2 numpy==1.26.4 matplotlib==3.8.0 seaborn==0.13.2 scikit-learn==1.5.2 plotly==5.24.1 imbalanced-learn==0.12.4 joblib==1.4.2 google-generativeai==0.8.3
- A Gemini API key from Google AI Studio or Google Cloud Console. See instructions in the source code or below.
This project is designed to run in Google Colab, using the provided notebook (Ensemble_Risk_Predictor.ipynb). To use it:
- Open the notebook in Google Colab.
- Run each cell in order. This will:
- Clone the GitHub repository
- Download the dataset
- Install necessary libraries
- Train the model using the provided notebook. / Use already trained model to make predictions.
- Allow you to interact with the model.
Once the notebook has been executed, you'll be prompted to input data or use a csv file, then you can input patient data directly via the console. You will be asked to provide the name of a patient and values for the features. The model will then generate a prediction on the risk score.
Alternatively, you can also use a CSV file that includes all required features, with an additional "PatientName" column. The model will iterate over each row, generate risk scores, and add a new column RiskScore to the CSV.
The project also supports a functionality to combine and analyze data from multiple CSV files. It will prompt you for the CSV files and saves the results into one output file.
The project uses the Google Gemini API to generate a basic advice based on the risk score.
Important Security Warning: Never hardcode your API key directly into any notebook.
You can acquire your API key using one of the following methods:
- Navigate: Visit the Google AI Studio.
- Create API Key: Follow the on-screen instructions to create a new API key.
- Copy Key: Once created, copy the API key. Keep this key secure and do not share it publicly.
- Create Project: Begin by creating a new project in the Google Cloud Console.
- Enable API: Enable the "Vertex AI API" for your project.
- Create API Key: Create a new API key following the instructions within the console.
- Copy Key: Copy the generated API key. Treat this key as confidential.
To ensure the security of your API key, we will use Google Colab's built-in Secrets feature:
-
Access Secrets: In your Google Colab notebook, locate the left sidebar and click on the key icon (labeled "Secrets").
-
Add Secret: Click on the button labeled "+ Add Secret".
-
Name the Secret: In the "Name" field, enter the following (exactly as shown):
GEMINI_API_KEY. -
Paste API Key: In the "Value" field, carefully paste the Gemini API key you copied in the previous step.
-
Add: Finally, click the "Add Secret" button.
-
Key Storage: This process will securely store your API key within Colab's secret management system. Your key will not be directly visible within your notebook's code.
Contributions to this project are welcome! Feel free to fork the repository, make changes, and submit a pull request. You can contribute by:
- Improving the performance of the model.
- Adding more sophisticated risk analysis tools.
- Enhancing the user interface and experience.
- Adding support for other datasets
- Adding different model architectures
- Improve the documentation
This Ensemble Risk Predictor is intended for research purposes only and should not be used as a substitute for professional medical advice. The predictions provided by this model are based on an analysis of data and should not be considered a diagnosis or a guarantee of health outcomes. Always consult with a qualified healthcare professional for any health concerns or before making any decisions related to your health or treatment.
This project is licensed under the MIT License - see the LICENSE file for details.