Skip to content

Latest commit

 

History

History
152 lines (78 loc) · 4.04 KB

NEWS.md

File metadata and controls

152 lines (78 loc) · 4.04 KB

FeatureExtraction 3.2.0

New Features:

  • Adds the ability to create a "temporalSequence" to create temporal features by specifying a fixed time window. (#134)

  • Identify inpatient events using visit_occurrence to support CDM 5.4 (#141)

Bugfixes:

  • Fixes MeasurementRangeGroup analysis bug (#133) to use proper look back days

Other:

  • Additional unit tests and improvement of code coverage

FeatureExtraction 3.1.1

Bugfixes:

  1. Removes 'size' column name since this is a reserved keyword for Oracle.

  2. Age group covariate name fix for Spark SQL

  3. Removes extraneous concepts from CCI score

  4. Correct output of multiple cohort IDs in print() function.

  5. Fix external class caching issue

  6. Fix standardized difference calc

  7. Fixing duplicates in counts when generating server-side aggregated measurement-range-group covariates.

FeatureExtraction 3.1.0

Changes:

  1. Added ability to compute aggregated statistics for multiple cohorts at once: The cohortId argument of the getDbCovariateData() function now accepts a vector of IDs. The aggregated statistics now contains a new field called cohortDefinitionId that can be used to distinguish between the various cohorts.

  2. Added filterByCohortDefinitionId() function to select covariates of one cohort from a covariateData object containing multiple.

  3. The cohortId argument now also supports integers greater than 32-bits.

Bugfixes:

  1. If a person has multiple measurements with the same measurement_concept_id, the selection of which measurement value to include is now deterministic.

FeatureExtraction 3.0.1

Changes:

  1. Adding timeRef table to CovariateData object for temporal covariates.

  2. Throw meaningful error when createCovariateSettings() is called without specifying any covariates.

  3. getDbCovariateData() returns empty covariates instead of NULL covariates when no covariates are specified.

Bugfixes:

  1. Time ID is now retrieved when aggregating binary temporal covariates.

FeatureExtraction 3.0.0

Changes:

  1. Switching from ff to Andromeda for storage of large data objects.

  2. Adding option to createTable1 to (also) show absolute counts.

  3. For analyses that restrict to inpatient diagnoses, the string '(inpatient)' is now appended to the covariate name.

Bugfixes:

  1. Dropping spurious 'analysis_name' field in temporary covariate reference table.

  2. Fixed covariate name of age groups over 100 years old.

FeatureExtraction 2.2.5

Changes:

  1. Added Korean translation of vignettes.

Bugfixes:

  1. Fixing aggregation of covariates generated using the cohort_attribute table.

  2. Fixed error when calling tidyCovariates without removing redundancy, when there are no infrequent covariates.

  3. Fixed server-side computation of median and interquartile range for measurement values.

  4. Restricting by concept ID now works for measurement values.

FeatureExtraction 2.2.4

Bugfixes:

  1. Removing redundant covariates now precedes removing infrequent covariates when calling tidyCovariates. Analyses where the most prevalent (redundant) covariate was removed are now exempt from removal of infrequent covariates.

  2. Fixed some typos in the vignette

FeatureExtraction 2.2.3

Changes:

  1. Also removing descendants when excluding condition concepts from condition groups. Condition groups therefore now work similar to drug groups.

Bugfixes:

  1. Updated workaround for ff bug causing chunk.default error on R v3.6.0 on machines with lots of memory.

FeatureExtraction 2.2.2

Changes:

  1. Using new SqlRender (v1.6.0) functions.

FeatureExtraction 2.2.1

Changes:

  1. Added option to specify number of digits for continuous variables in createTable1 function.

Bugfixes:

  1. Added missing space cause SQL error when both include and exclude concept are specified.

FeatureExtraction 2.2.0

Changes:

  1. Added the Hospital Frailty Risk Score.