SqlRender 1.8.0 has been released. The most important update is the addition of translation to Spark SQL, provided by @Ajit_Londhe.
3 Likes
SqlRender 1.8.1 has been released. This update is just to prevent SqlRender from being kicked out of CRAN for requiring a Java version newer than Java 6.
1 Like
Cohort Diagnostics 2.1.2 was released yesterday. This is a hotfix release that fixed a few minor issues identified in shiny app Diagnostics Explorer.
2 Likes
EmpiricalCalibration 3.0.0 has been released. This new version brings several important changes, including:
- Ability to combine empirical calibration with MaxSPRT correction for sequential testing. See the new vignette
- Improved computation of calibrated p-values and CIs, making them more stable, and increasing agreement between p-values and CIs when calibrating CIs based on negative controls alone.
Thanks to the hard work of contributors during the HADES Unit-Test-a-Thon, we now also have a whole battery of unit tests for EmpiricalCalibration, bringing code coverage from 0% to 89%!
3 Likes
CohortMethod 4.2.1 has been released. The most important change is the ability to downsample the cohorts right before computing covariate balance. This is helpful when the exposure cohorts are really large (think millions), where this option can reduce the compute time from days to hours, with pretty much the same result.
3 Likes
DatabaseConnector 5.0.0 has been released. The most important change is reverting a change of v4: integers in the database will once again be imported as numeric (real) values in R. The two main reasons are that (1) R and especially dplyr have issues with R integers, causing errors downstream in the HADES stack, and (2) Oracle doesn’t really support integers, leading to different behaviors between platforms.
CohortDiagnostics 2.1.4 has been released with changes logged here. The most important change is a bug fix with privacy protecting feature (min cell count). Thank you @msuchard for identifying the issue.
ROhdsiWebApi 1.2.1 has been released with changes logged here. This release involves a new and much request function to update a published definition that already exists in the webapi/atlas. Thank you @jpegilbert for leading this release.
1 Like
Introducing CohortGenerator v0.2.0, a new package for generating cohorts against the OMOP CDM. At the moment, many of our analytics and study packages contain code to generate cohorts and so we wanted to create a package to centralize this core task for use across HADES. This package contains the core elements for generating cohorts including creating the cohort tables, functions for generating the cohorts and gathering cohort counts. We’ve provided a vignette to walk through how to use CohortGenerator and explain some features of the package.
Thanks to @schuemie and @Gowtham_Rao since most of this code originated from the CohortDiagnostics package.
4 Likes
FeatureExtraction v3.2.0 has been released with changed detailed here. Most notably, this release will allow for compatibility with CDM v5.4.
SelfControlledCaseSeries v3.2.0 has been released. The most important changes include the ability to model the effect of calendar time using a spline, and a diagnostic for whether the rate of the outcome is constant over calendar time (before or after adjustments).
SqlRender v1.8.3 has been released. The most important change is a fix for errors in unit tests of other packages that use Eunomia, caused by a change in the latest version of SQLite.
SelfControlledCohort v1.6.0 has been released. The main changes are the usage of batched queries for very large data sets with optional callbacks to handle data in chunks as it is retrieved from the database, the addition of time at risk statistics for the unexposed and exposed populations and how time at risk statistics are stored (in separate data frames).
SelfControlledCaseSeries v.3.2.1 has been released. This fixes the issue that v3.2.0 referenced the develop version of SqlRender.
CohortMethod v4.2.2 has been released. Most important change is the ability to exclude specific analyses when running multiple analyses.
Capr v1.0.1 (includes minor bug fixes since initial release) have been released. Capr is a new package in HADES focusing on designing and manipulating cohort definitions from an R interface. Capr leverages circe-be in designing cohorts. Cohort definitions designed in Capr may be exported as an equivalent json to one designed using ATLAS, thus it is compliant with the rest of the OHDSI analytics stack.
Capr also specializes in the creation and saving of cohort components, sub-items of the cohort definition that may be reused across multiple cohort definitions (i.e. primary criterias or inclusion rules). Therefore, Capr syntax can be used to template cohort definitions and build multiple like cohort definitions iteratively.
To properly use Capr, a user must have access to a CDM vocabulary schema to lookup concepts and have a location to save cohort definition results. Capr relies on CirceR to help compile the cohort definition into ohdisql. To learn more about Capr, check out the vignettes.
Thanks to many people for their comments, suggestions, and encouragement throughout the development of this package. Particular shout out to @lee_evans who originated this project.
2 Likes
CohortDiagnostics version 2.2.0 has been released. This version now supports the use of the CohortGenerator package, it’s usage is documented here. As this is a minor release, backward compatibility is maintained for instantiating cohorts and using the runCohortDiagnostics interface. However, please note that these will be removed in a future version; it is strongly recommended that you alter any study packages/scripts.
Please check the release notes for for details of changes.