Surgery Outcome Prediction

This project is a Streamlit application for predicting surgery outcomes based on doctor's notes and patient details. It uses machine learning models to classify surgery types and predict the outcome based on various features. The Surgery Outcome Prediction model predicts four types of surgeries: DALK, EK, PK, and THPK and the outcome based on those surgeries.

Features

• Predict Surgery Types: Classify the type of surgery based on doctor's notes.
• Predict Surgery Outcome: Predict whether a surgery will pass or fail based on patient details and predicted surgery types.

Models Used

• BiLSTM Model: For predicting surgery types from doctor's notes.
• Gradient Boosting Model: For predicting the surgery outcome.

Libraries

This project uses the following Python libraries:

• pandas
• numpy
• joblib
• gensim
• tensorflow
• streamlit
• scikit-learn

How to Run

1. Clone the Repository:

   git clone https://github.com/mishra-krishna/Surgery-Outcome-Prediction

2. Install Dependencies: Install the required libraries using the requirements.txt file.

   pip install -r requirements.txt

3. Run the App: Start the Streamlit app with:

   streamlit run main.py

Deployed App

You can access the deployed Streamlit app here.

Model Development

For details on the model development process and dataset insights, please refer to the model_development.ipynb notebook.

Model Files

Ensure that the following files are present in the project directory for the app to work:

• bilstm_model.h5 - BiLSTM model for surgery type prediction.
• gradient_boosting_model.pkl - Gradient Boosting model for outcome prediction.
• word2vec_model.bin - Word2Vec model for text processing.

Contributing

Feel free to open issues or submit pull requests for improvements.