Competing Risks Model

A Multi State Competing Risks model to predict COVID-19 medical conditions

Prerequisites

• R >= 4.0
• Node >= 13.0

Quickstart

cd competing-risks/client
npm run build
cd ..
docker-compose up

Development

First, clone this repository:

git clone https://github.com/idocal/competing-risks.git
cd competing-risks

Now add your CSV data file to server/data directory.

To start the R server:

cd server
Rscript serve.R

R should mow be listening to port 8000.

To start the client:

cd client
npm install
npm run start

The React server should now be running on port 3000. The website should be available through the browser at: http://localhost:3000

Server Requests

The R server runs 20,000 Monte Carlo simulations for a requested patient and returns:

• Death probability
• Time at hospital quantiles (0.1, 0.25, 0.5, 0.7, 0.9)
• Time at Severe condition in days

Each requests expects:

• Age
• Gender (Male 1 or Female 2)
• Start State Type
• Start State
• Visited States

Example request:

curl --data '{"age": 45, "gender": 1, "start_state_type": 3, "start_state": 23, "states": [{"medicalState": 2, "hospital": 5}]}'  "http://localhost:8000/patient"