hmankodiya / covid-19-vs-influenza

COVID-19 vs Influenza

This repository contains code and ML models used for distinguishing between COVID-19 and Influenza.

Models

• Influenza vs COVID-19 Positive vs COVID-19 Negative
• COVID-19 Negative vs COVID-19 Positive
• Others vs Influenza
• Influenza vs COVID-19

Data

The columns present in the dataset (in order) are:

• sex_F – 1 if Female, 0 if otherwise

• race_White – 1 if White, 0 otherwise

• race_AA – 1 if African American, 0 otherwise

• race_Other – 1 if neither White nor African American, 0 otherwise

• ethnicity_Hispanic_YN – 1 if Hispanic or Latino, 0 otherwise

• Age – Age in years (integer)

• patient_class

• encounter_type

• reason_for_visit

• SBP – Systolic BP

• DBP – Diastolic BP

• Temp_C – Temperature in ºC

• HR – Heart Rate

• RR – Respiratory Rate

• SPO2 – Blood Oxygen (SpO_2)

• BMI

• BSA

• Month – Month (integer) 0 - 11

• Class – Ground Truth

Instructions

To run the code, ensure you have Python installed. All our code was written in Python 3.8.0, but should be backwards-compatible to some earlier versions of Python 3. Follow the instructions below to get started.

Clone or Download Repository

Clone the repository by running this command.

git clone https://github.com/ynaveena/COVID-19-vs-Influenza

You may also just download the repository from GitHub.

Requirements

To install all requirements, simply run this in the Terminal.

pip install -r requirements.txt

Note: On some systems, you may have to use pip3 instead.

Jupyter Notebooks

Additionally, to run our notebooks, you need to have Jupyter Notebook installed. You can do this using the command below.

pip install jupyter

Note: On some systems, you may have to use pip3 instead.

Running the Code

To use the notebooks, simply start up a Jupyter server using the command below.

jupyter notebook

Our code is also available as Python scripts. To use them, follow the instructions below.

Sample Validation

Run the command given below from the root of the repository to validate the models on sample data.

python scripts/validation.py [MODELS]
# For help: python scripts/validation.py --help

[MODELS] can either be empty (this would mean all 4 models are validated) or a space-separated list of specific models to validate. The arguments can be specified as follows:

• 3Class – Influenza vs COVID-19 Positive vs COVID-19 Negative

• CPosvCNeg – COVID-19 Negative vs COVID-19 Positive

• FvOthers – Others vs Influenza

• FvC – Influenza vs COVID-19

Citation

If you use the code and models provided in this repository, please do cite us.

@misc{covid19vsinfluenza,
  title = {COVID 19 vs Influenza},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub Repository},
  howpublished = {\url{https://github.com/ynaveena/COVID-19-vs-Influenza}}
}