README.md

In ohdsi-studies/ExistingStrokeRiskExternalValidation: OHDSI network large-scale external validation study

Proof of concept demonstrating the utilization of the OHDSI data network to perform external model validation of existing stroke models

• Analytics use case(s): Patient-Level Prediction
• Study type: Methods Research
• Tags: External validation
• Study lead: Jenna Reps
• Study lead forums tag: jreps
• Study start date: Jan 1, 2019
• Study end date: Feb 20, 2019
• Protocol: -
• Publications: Here
• Results explorer: -

A package for running the OHDSI network study to externally validate 5 existing stroke risk prediction models using the PatientLevelPrediction framework

Features

• Creates target cohort of females newly diagnosed with atrial fibrilation between ages 35-65
• Creates 4 outcome cohorts of various stroke definitions
• Implements 5 existing stroke risk prediciton models and validates on data in OMOP CDM for each target and outcome cohort combination
• Sends summary results

Technology

ExistingStrokeRiskExternalValidation is an R package.

System Requirements

Requires R (version 3.3.0 or higher).

Dependencies

• PatientLevelPrediction
• PredictionComparison

Getting Started

1. In R, use the following commands to download and install:

```r #================================= STEP 1: INSTALL PACKAGES ================================== install.packages("devtools") devtools::install_github("OHDSI/PatientLevelPrediction") devtools::install_github("OHDSI/PredictionComparison") devtools::install_github("ohdsi-studies/ExistingStrokeRiskExternalValidation")

library('ExistingStrokeRiskExternalValidation')

view the models - this should pop up a View with the model info

viewModels()

================================= STEP 2: MAIN STUDY ==================================

options(fftempdir = 'T:/yourFftemp') dbms <- yourDbms user <- yourUsername pw <-yourPassword server <- Sys.getenv('server') port <- Sys.getenv('port') connectionDetails <- DatabaseConnector::createConnectionDetails(dbms = dbms, server = server, user = user, password = pw, port = port)

databaseName = 'friendlyDatabaseName' cdmDatabaseSchema <- 'yourCdmDatabaseSchema' cohortDatabaseSchema <- 'yourCohortDatabaseSchema' cohortTable <- 'existingStrokeVal' outputLocation <- 'C:/existingStrokeVal'

ExistingStrokeRiskExternalValidation::main( connectionDetails=connectionDetails, oracleTempSchema = NULL, databaseName=databaseName, cdmDatabaseSchema=cdmDatabaseSchema, cohortDatabaseSchema=cohortDatabaseSchema, outputLocation=outputLocation, cohortTable=cohortTable, createCohorts = T, runAtria = T, runFramingham = T, runChads2 = T, runChads2Vas = T, runQstroke = T, summariseResults = T, packageResults = T, N=10)

After checking the compressed folder containing the shareable results submit the results

either email them to study admin or run

submitResults(exportFolder = file.path(outputLocation,'export',paste0(databaseName,'.zip') ), keyLocation, userName)

where keyLocation, userName are provided by request

```

License

ExistingStrokeRiskExternalValidation is licensed under Apache License 2.0

Development

ExistingStrokeRiskExternalValidation is being developed in R Studio.

ohdsi-studies/ExistingStrokeRiskExternalValidation documentation built on Aug. 23, 2020, 1:52 p.m.