Jiayi Tong (@JiayiJessieTong), Yong Chen (@chenyong1203) and I (@jreps) would like to invite you to participate in an OHDSI network study to implement a novel one-shot distributed method, dGEM, that can efficiently combine heterogeneous data while preserving the privacy of protected patient-level and hospital-level health information and calculate effect estimates that are equivalent to pooling the data (highly accurate). We would like to implement the dGEM method across the OHDSI network for the hospital profiling problem to i) demonstrate the applicability of the proposed dGEM method and ii) investigate the effect estimates of patient- and hospital-level factors of COVID-19 mortality for the alpha and delta waves.
This study involves running five analysis steps and using the website: https://pda-ota.pdamethods.org/ to download the json settings and easily upload json result files. We will organize two meetings during the first step (after May 17th) to: 1) demonstrate how to use the https://pda-ota.pdamethods.org/ website to register and access the json settings, 2) demonstrate how to run the initial steps in the study R package and 3) show the result object that is generated and demonstrate how to upload this via the website. For full details about how to participate see: https://github.com/ohdsi-studies/dGEMcovid/blob/master/STUDY-PACKAGE-EXECUTE.md .
If you have data in the OMOP CDM format and you’d like to get involved, please check out the study GitHub link here: https://github.com/ohdsi-studies/dGEMcovid and reply to this message.
If you would like to participate, please let us know by May 17th. We hope to have all of the data analysis performed by the end of June and then will submit a manuscript with participants as co-authors.
Please reply to this post with any questions, you may also contact us at the following email:
Yong Chen: ychen123[at]pennmedicine.upenn.edu
Or post an issue on the github issues: Issues · ohdsi-studies/dGEMcovid · GitHub