Strategus v1.4.1 has been released which fixes a bug in the PatientLevelPrediction & PatientLevelPredictionValidation modules to handle covariate cohort tables.
CohortGenerator 0.12.2 has been released. This includes a bug fix and updates for unit tests.
ResultModelManager v0.6.0 has been released.
This release adds experimental support for using reticulate/python to upload results into postgres databases with the COPY functionality, which should significantly increase upload times when using data cleaning checks due to reduced IO.
In addition, this includes messages to notify developers that R6 will be removing public use of $finalize methods. Instead, package maintainers should replace all calls with $closeConnection.
CohortMethod 5.5.1 has been released. This includes a refactoring of the Kaplan Meier curve code, with as main benefit that the exportToCsv() function no longer throws an error when one of the cohorts is empty and using stratification.
DatabaseConnector 7.0.0 has been released. This is a major version update.
When DatabaseConnector was created over 10 years ago, DBI was still in its infancy. Over the years, both have evolved and matured. We asked ourselves if we still need DatabaseConnector, and have have concluded the answer is ‘yes’. DBI is not sufficient to guarantee ‘write code once, run on all platforms’, but DatabaseConnector aims to be. An example is how DatabaseConnector provides temp table emulation while DBI does not. DatabaseConnector also prescribes specific drivers for each platform, leading to greater consistency.
That being said, DatabaseConnector and DBI did have inconsistencies that didn’t make sense. The goal of this new version is to bring DatabaseConnector closer to DBI. A lot has changed under the hood, which should also make it easier to slot in new drivers in DatabaseConnector in the future.
But the most important change is to switch the dbplyr backend from one that relied on SqlRender to the native backends for the respective platforms. This provides much greater reliability and compatibility with other DBI drivers. Many thanks to @Adam_Black for implementing this important change, which should allow us to better bridge the gap between HADES and the tools developed by the Oxford team, which rely on dbplyr.
This new version has been extensively tested with all the HADES packages, but there may be some issues remaining, so please be patient.
2 Likes
DataQualityDashboard 2.8.0 has been released. This release includes an important bugfix to the isStandardValidConcept check, which previously was not flagging valid non-standard concepts! The updated check will now fail for all classification concepts, invalid concepts, and non-standard concepts.
1 Like
SqlRender 1.19.4 has been released. This contains a single but rather critical bug fix. OHDSISql FLOAT was translated to FLOAT on DataBricks, but the former has 64-bit precision whereas the latter has 32-bits. As a consequence, if for example you uploaded a data frame with concept IDs those IDs would have silently changed because of the loss of precision.
ResultModelManager v0.6.1 has been released. This is a hotfix to resolve an issue with Pooled connection handling compatibility with DBI. Upgrading requires the use of DatabaseConnector v7.0.0
HADES-wide release 2025Q3 has been released. This is a snapshot of the HADES packages and their dependencies, thoroughly tested and confirmed to be mutually compatible. It is intended to be a stable environment for studies and execution engines. Currently, the release is only available as an renv lock file, but folks are working on containers as well.
PatientLevelPrediction v6.5.1 has been released. This release has bugfixes for deprecations in xgboost and ResultModelManager as well as a bugfix for compatibility with DatabaseConnector 7.
EvidenceSynthesis 1.1.0 has been released. This adds prediction intervals to the Bayesian random-effects meta-analysis.
SelfControlledCaseSeries 6.1.1 has been released. This includes a more robust implementation of the time-trend diagnostic, which sometimes threw an error (especially with very large data).
FeatureExtraction v3.12.0 has been released. It fixes non reproducible behaviour in the removal of redundant covariates.
Andromeda 1.2.0 has been released. This includes several changes that make Andromeda behave better for very large data (many GBs), e.g. avoiding running out of memory.
CohortMethod 5.5.2 has been released. The main functions were refactored to reduce peak memory requirements. Together with the new Andromeda and ParallelLogger releases this allows running CohortMethod on very large cohorts (> 10 million persons per cohort).
OhdsiReportGenerator 2.0.1 has been released. The main changes are improved prediction result exporting and a bug fix with the full report generation.
OhdsiShinyModules 3.4.0 has been released. This package now uses OhdsiReportGenerator 2.0.1 for extracting results and has large improvements in the interface for characterization, estimation and prediction result viewing.
1 Like
CohortGenerator v1.01 has been released. This is a major release with breaking changes:
- Information about which cohort generations have completed successfully has moved from the file system (v0.x behavior) to the database to allow for incremental execution in distributed environments.
- Adds greater support for creating cohorts using SQL and bulk cohort creation as detailed in the Using Template Cohorts vignette.
Big thanks to @jpegilbert who led the development of the new features in this release.
PheValuator 2.2.16 has been released. It fixes a bug that occurred due to new version of PLP.
DataQualityDashboard 2.8.2 has been released. This release contains some minor bugfixes:
- Adds missing units to the plausible units check
- Fixes schema references that were throwing errors for users with vocab/cohort tables stored outside their CDM schema
It also contains all the changes needed (I HOPE!) to submit DQD to CRAN. Big thanks to @MaximMoinat who slashed a bunch of unused UI code and got the package under the CRAN size limit 
1 Like
DatabaseConnector 7.1.0 has been released. This includes 2 bugfixes and the ability to configure DuckDB (e.g. set its memory limit).