This is the work of Nufail Vhora (2k21/CO/319) and Paras Arun (2k21/CO/323) of Delhi Technological University
Welcome to MedWise, a web application for predicting Diabetes, Parkinson's Disease, and Heart Disease. This project utilizes custom machine learning models seamlessly integrated into a Streamlit web app to provide efficient predictions.
MedWise is designed to predict three different diseases using specific machine learning models:
- Parkinson's Prediction using a Stacking Classifier with KNN, SVC and Random Forest as base Models, Logistic Regression as the meta model
- Heart Disease Prediction using a Bagging Ensembler built on Logistic Regression Model
- Diabetes Prediction using a Soft Voting Classifier
- Parkinson's Prediction: 92.3%
- Heart Disease Prediction: 90.11%
- Diabetes Prediction: 79%
The application is deployed and can be accessed here.
The user interface is intuitive and user-friendly, leveraging the Streamlit Option Menu for easy navigation.
-
streamlit: Enabled the creation of a user-friendly web application. Leveraged Streamlit's simple and intuitive API to design the user interface.
-
scikit-learn: used in Parkinson's Prediction. The Random Forest Classifier, Linear SVM Model, KNN and Logistic Regression Model are all implemented and stacked using scikit-learn.
-
pycaret: used in Heart Disease and Diabetes Prediction. Soft Voting Ensembler and Bagging are implemented using it
-
pickle: Employed in MedWise for model persistence. After training the machine learning models, they are serialized using pickle and saved to disk. This allows for quick and easy loading of pre-trained models within the Streamlit app, ensuring that predictions can be made without the need for retraining each time the application is launched.
-
pandas: Used in handling and processing the datasets used in MedWise.