SelfControlledCaseSeries 5.2.0 has been released. This contains several bug fixes and a minor change to the results model output.
SelfControlledCaseSeries 5.2.1 has been released. This fixes a critical bug in the results model introduced in 5.2.0.
SelfControlledCaseSeries 5.2.2 has been released. This fixes another critical bug, this time in the exportToCsv() function.
Friends:
KEEPER 0.2.0 has been released (marking its incorporation into the HADES family).
Before I proceed with describing what it is (you may have already seen it at the OHDSI Global 2023 Plenary here and here), I want to thank @schuemie big time for support and major help with this release.
In a nutshell, KEEPER is a package that enables you to do “chart review” on OMOP-ed data. It takes the cohort of patients you want to review as well as codes you want to look at for a condition of interest (relevant diagnosis, lab tests, procedures, alternative diagnosis) and outputs patient profiles that you can review in the same way you review charts to determine if a patient has a disease of interest or not. This vignette has more details on how to select codes.
Using KEEPER with human review was shown to be as accurate and twice as fast compared to full chart review in our paper.
What is more is that KEEPER can be paired with LLMs to review patients in an automated fashion and compute all phenotype performance characteristics (PPV, sensitivity and specificity). The author of this piece, @schuemie showed that such review is as good as human review. This vignette describes how to do it.
Many thanks to our wonderful collaborators: @zhuk, @Vlad_Korsik, Seung In Seo and of course even more thanks to @Patrick_Ryan and @hripcsa.
1 Like
DeepPatientLevelPrediction v2.1.0 has been released. This release includes a number of additions. The biggest one are transfer learning support and gradient accumulation. Now as well there is a docker image published accompanying each release.
DataQualityDashboard v2.6.1 has been released. This release includes a ton of new documentation (nearly all checks now have a full-page guide to how to interpret and resolve failures!). Check it out here: Index • DataQualityDashboard
We’ve also fixed several bugs and added a numeric execution time field to the results output (that’s for you @Frank - finally!).
Huge, huge thanks to @MaximMoinat for significant contributions to this release 
and a shoutout to Nitesh Balakrishnan for adding his first DQD feature 
@Dymshyts @clairblacketer and @Jared all contributed documentation pages as well. Great work team!
1 Like
CohortGenerator v0.10 has been released. This update includes a new function for performing all cohort generation tasks in a single call, adopts the ResultModelManager for managing results produced by the package and expands unit tests to run on all OHDSI test database servers in support of our HADES 2024 OKRs. There are also some additional bug fixes as detailed in the release notes.
OhdsiShinyModules 2.1.5 has been released. This contains a hotfix for the CohortDiagnostics app when using sqlite databases.
FeatureExtraction v3.6.0 has been released. This contains a couple bug fixes as well as the option to create covariates based on value_as_concept (for observations/measurements)
CohortIncidence v4.0.0 has been formally accepted into HADES. Although this package has been available to create incidence rates and proportions, as of version 4.0.0, it meets HADES requirements.
CohortIncidence is used to calculate incidence rates and proportion given a Target - Outcome - TimeAtRisk specification. It stratifies by age, gender and index year, and produces output in a portable CSV format.
Please visit the package documentation for reference and usage guide.
ResultModelManager v0.5.9 has been released. This package has now been accepted in to CRAN, and the changes to the package largely reflect this.
Unless using a development release, please install with install.packages('ResultModelManager') and remove remote references from and R package DESCRIPTION files.
SelfControlledCaseSeries 5.2.3 has been released. Fixes some edge cases and has some quality of life improvements.
SelfControlledCaseSeries 5.3.0 has been released. This adds the option to select whether you are using exposure or outcome (negative) controls.
SqlRender 1.18.1 has been released. This adds some fixes related to date function translations for Databricks and DuckDB.
Andromeda 0.6.7 has been released. This adds a minor bugfix.
ParallelLogger 3.3.1 has been released. This adds a minor bugfix.
FeatureExtraction v3.7.0 has been released. This release adds the tempEmulationSchema parameter, oracleTempSchema has been made deprecated.
CohortGenerator v0.11.1 has been released and includes better support for censoring based on a minimum cell count.
CohortGenerator has also been accepted in to CRAN.
CohortMethod 5.4.0 has been released. Most important changes are compatibility with the latest OhdsiShinyModules API, and adding power calculations for logistic regression models.
[CohortDiagnostics version 3.3.0] has been released. This includes a number of fixes for bugs and an optional parameter to run characterization only on a sample of cohorts.
Please check the release notes for full details of changes.