- Analytics use case(s): Characterization & Population Level Estimation
- Study type: Clinical Application
- Tags: OHDSI
- Study lead: Jenny Lane
- Study lead forums tag: jenniferlane
- Study start date: 1st December 2020
- Study end date: 1st May 2021
- Protocol: EU PAS 38362
- Publications: ** **
- Results explorer:
A study to evaluate the risk of adverse musculoskeletal events in post-menopausal women taking Aromatase Inhibitors compared to Tamoxifen to treat breast cancer.
If you are undertaking cohort diagnostics for the first time, you may need set up your environment using the instructions given in the HADES installation guide. To run the study you will need to load the package, enter the RProj, and build it. Once built, you will need to open the extras/CodeToRun.R file and enter your database connection details, where you want to save your results locally, and so on (see instructions in the file). In this same file you can then run the study, view the results locally in a shiny application, and share your results.
Please note prior to running (and as detailed in the file Extas/CodetoRun.R), you may also need to install packages in order for packages to run including:
- devtools
- dplyr
- ggplot2
- SqlRender
- DatabaseConnector
- parallel
- rJava
From github:
- OHDSI/FeatureExtraction@v3.1.0
- OHDSI/Andromeda@v0.4.0
- OHDSI/OhdsiSharing@v0.2.2
- edward-burn/CohortDiagnostics; branch = DiagAi
Note, we suggest using the branch of cohort diagnostics from the edward-burn account rather than the OHDSI one so as to ensure consistency in results set (the OHDSI cohort diagnostics package continues to be developed - the version on Ed´s github is just a recent fork from the main OHDSI repo).
This package should run through the regeneration of the appropriate dependencies by using the renv.lock
file. To install these dependencies, run
renv::restore
after opening the project. The file given here requires R4.0
minimum- contact us if you require a renv.lock
file that runs with R3.5.0
or above. You can execute the study using the CodeToRun
file in the package.
The output is contained in a .ZIP file within the Output
folder, in the Export
directory.
We recommend centres review the blinded results in their personal shiny app prior to sharing within OHDSI. We then invite centres to share the results with us via the SFTP server. To share these via the OHDSI SFTP you will need a key file which you will need to be sent separately. To get this please contact jennifer.lane@ndorms.ox.ac.uk.
We look forward to working with you!