COVID-19 DREAM Challenge: Baseline Model

Overview

This repository describes how to build and run locally the baseline model of the COVID-19 DREAM Challenge. The goal of this DREAM Challenge is to develop models that take as input the electronic health records (EHRs) of a patient and outputs the probability of this patient tested positive for COVID-19.

Description of the model

This baseline model takes 13 features including clinical symptoms and vital signs and refers to research conducted by Feng et al. link This baseline model isn't trained on any real COVID patient data. Each patient is given a risk score based on the presence of the 14 features. A threshold of risk score is chosen basing on 10% test positive rate in Washington state. Patients whose risk scores are above threshold are assigned test-positive probability as 1, otherwise 0.

Feature	OMOP Code	Domain	Threshold
age	-	person	>60
temperature	3020891	measurement	>37.5'
heart rate	3027018	measurement	>100n/min
diastolic blood pressure	3012888	measurement	>80mmHg
systolic blood pressure	3004249	measurement	>120mmHg
hematocrit	3023314	measurement	>52
neutrophils	3013650	measurement	>8
lymphocytes	3004327	measurement	>4.8
oxygen saturation in artery blood	3016502	measurement	<95
cough	254761	condition	-
pain in throat	259153	condition	-
headache	378253	condition	-
fever	437663	condition	-

Dockerize the model

1. Clone this GitHub repository
2. docker build -t docker.synapse.org/syn12345/my_model:v0.1 example/app

Run the baseline model locally on synthetic data

This section describes how to run the model locally, that is, without using the IT infrastructure of the COVID_19 DREAM challenge(need updates).

Description of the data

Learn more about OMOP Synpuf data(need updates)

Download the data

The Synpuf data are available here. After downloading them, uncompress the archive and place the data folder where it can later be accessed by the dockerized model (see below).(need updates)

Run the model locally on Synpuf data

Once the baseline model has been dockerized (see above), run the following command to train the model on Synpuf data:

docker run -v <path to data folder>:/data:ro
-v <path to scratch folder>:/scratch:rw
-v <path to output folder>:/output:rw
 docker.synapse.org/syn12345/my_model:v0.1 bash /app/COVID_baseline.sh

where

• <path to data folder> is the absolute path to the data (e.g. /home/charlie/ehr_experiment/synpuf_data/data).
• <path to scratch folder> is the absolute path to the scratch folder (e.g. /home/charlie/ehr_experiment/scratch).
• <path to output folder> is the absolute path to where the the predictions will be exported (e.g. /home/charlie/ehr_experiment/output))

If the docker model runs successfully, the prediction file predictions.csv file will be created in the output folder. This file has two columns: 1) person_id and 2) test-positive probability. Note: make sure the column 2) contains no NA and the values are between 0 and 1.

Make a submission to COVID-19 DREAM challenge

Please see this Synapse page for instructions on how to make a submission link