Midterm Project - Heart Failure Prediction

Motivation

Cardiovascular diseases (CVDs) are the number 1 cause of death globally, taking an estimated 17.9 million lives each year, which accounts for 31% of all deaths worldwide. Heart failure is a common event caused by CVDs and this dataset contains 12 features that can be used to predict mortality by heart failure.

Most cardiovascular diseases can be prevented by addressing behavioral risk factors such as tobacco use, unhealthy diet and obesity, physical inactivity, and harmful use of alcohol using population-wide strategies.

People with cardiovascular disease or who are at high cardiovascular risk (due to the presence of one or more risk factors such as hypertension, diabetes, hyperlipidemia, or already established disease) need early detection and management wherein a machine learning model can be of great help.

That's why in this project, we will try several machine learning classification models to come up with a solution. We'll try to predict DEATH_EVENT of a person based on a set of attributes.

Original source: https://github.com/mdrkb/mlcamp/tree/main/07-midterm-project

Dataset

The dataset is collected from Kaggle:
https://www.kaggle.com/andrewmvd/heart-failure-clinical-data

Feature Description

• age: Person age.
• anaemia: Decrease of red blood cells or hemoglobin (boolean).
• creatinine_phosphokinase: Level of the CPK enzyme in the blood (mcg/L).
• diabetes: If the patient has diabetes (boolean).
• ejection_fraction: Percentage of blood leaving the heart at each contraction (percentage).
• high_blood_pressure: If the patient has hypertension (boolean).
• platelets: Platelets in the blood (kiloplatelets/mL).
• serum_creatinine: Level of serum creatinine in the blood (mg/dL).
• serum_sodium: Level of serum sodium in the blood (mEq/L).
• sex: Woman or man (binary).
• smoking: If the patient smokes or not (boolean).
• time: Follow-up period (days).
• DEATH_EVENT: If the patient deceased during the follow-up period (boolean).

Repo Structure

The following files are included in the repo:

heart-failure-prediction
├── Dockerfile <- Instructions to build Docker image
├── README.md <- Getting started guide of the project
├── heart_failure_clinical_records_dataset.csv <- Dataset
├── model.bin <- Exported trained model
├── notebook.ipynb <- Jupyter notebook with all codes
├── predict.py <- Model prediction API script for local deployment
├── preduct_test.py <- Model prediction API script for testing
├── requirements.txt <- Package dependency management file
└── train.py <- Final model training script

Create a Virtual Environment

Clone the repo:

git clone <repo>
cd heart-failure-prediction

For the project, virtualenv is used. To install virtualenv:

pip install virtualenv

To create a virtual environment:

virtualenv venv

If it doesn't work then try:

python -m virtualenv venv

Activate the Virtual Environment:

For Windows:

.\venv\Scripts\activate

For Linux and MacOS:

source venv/bin/activate

Install Dependencies

Install the dependencies:

pip install -r requirements.txt

Run the Jupyter Notebook

Run Jupiter notebook using the following command assuming we are inside the project directory:

jupyter notebook

Run the Model Locally

The final model training codes are exported in this file. To train the model:

python train.py

For local deployment, start up the Flask server for prediction API:

python predict.py

Or use a WSGI server, Waitress to run:

waitress-serve --listen=0.0.0.0:8080 predict:app

It will run the server on localhost using port 8080.

Finally, send a request to the prediction API http://localhost:8080/predict and get the response:

python predict_test.py

Run the Model in Cloud

The model is deployed on Pythonanywhere and can be accessed using:

https://mdrkb.pythonanywhere.com/predict

The API takes a JSON array of records as input and returns a response JSON array.

How to deploy a basic Flask application to Pythonanywhere can be found here. Only upload the heart_failure_clinical_records_dataset.csv, train.py, and predict.py files inside the app directory. Then open a terminal and run train.py and predict.py files. Finally, reload the application. If everything is okay, then the API should be up and running.

To test the cloud API, again run predict_test.py from locally using the cloud API URL. We'll get output similar to this:

[
  {'heart_failure': True, 
    'heart_failure_probability': '99.99%', 
    'id': 0
  }, 
  {
    'heart_failure': False, 
    'heart_failure_probability': '99.886%', 
    'id': 1
  }
]

Notes:

• Pythonanywhere instance has almost all the necessary packages installed. So, we don't need to install packages manually.
• Some packages are not up to date. So, there might be compatibility issues, especially with Scikit-Learn and XGBoost. For example changing dv.get_feature_names_out() to dv.get_feature_names() for lower versions of Scikit-Learn.

Build Docker Image

To build a Docker image:

docker build -t heart-failure-prediction:1.0 .

TO run the image as a container:

docker run --name heart-failure-prediction -it --rm -p 8080:8080 heart-failure-prediction:1.0 

To test the prediction API running in docker, again run predict_test.py locally.