The family and acronym arguments are used to select a specific growth standard.
The classify_*() family of functions no longer take vectors as input, but instead take data.frame-like objects.
NEW FEATURES:
Use classify_growth() to get data on multiple growth indicators at the same time. It will try to compute as many growth outcomes (i.e. size-for-GA, SVN, stunting, wasting, weight-for-age and head size) as possible based on the data you provide, and will tell you which analyses were performed.
The GIGS z-scoring functions are now explicitly user-facing. Check out gigs_waz() and friends for easy z-scoring using growth standards suggested by GIGS.
An id parameter for classify_growth() and friends, as well as the gigs z-scoring functions. This parameter informs the functions about who each data point is from, so that birth measurements for each individual can be identified and assessed appropriately.
The compute_*() functions for growth analysis. These functions take several vectors containing anthropometric measurements, age, and sex; they return a factor of growth categories. They are equivalent to the classify_*() functions from v0.4.1 and earlier.
The categorise_*() functions for growth analysis. These functions take vectors of growth centiles/z-scores as input, and return a factor of growth categories.
Added the INTERGROWTH-21st standard for estimated fetal weight based on the Hadlock formula. The is available through value2zscore() and friends provided that family = "ig_fet" and acronym = "hefwfga".
Added a new package-level option - "handle_unused_levels". This option matters when you’re categorising growth data, but want gigs to drop unused levels in your newly-made factors.
By default, .gigs_options$handle_unused_levels is "keep_warn". So gigs will tell you that there are unused factor levels, but not drop them. Change it to "drop_warn" or "drop_silent"to drop the unused factor levels automatically.
Added extended versions of the INTERGROWTH-21st Newborn Size standards for weight-for-GA, length-for-GA, and head circumference-for-GA. These extended versions have a GA range from 154 days (22+0 to 44+6 weeks). You can access these extended standards by setting family = "ig_nbs_ext" in GIGS conversion functions.
Z-scoring based on GIGS-recommended standards now accounts for measurements from different individuals, whereas in earlier versions the z-scoring logic treated data as if it all came from the same individual.
.gigs_options is now actively exported by gigs and available to end-users.
INTERNAL/DOCS CHANGES:
Errors, warnings, and messages from GIGS are now handled with cli::cli_abort() and friends, and look much prettier.
Updates to the ‘Getting Started’ vignette and ‘Benchmarking’ article.
Other non-breaking changes to documentation and test suite.
gigs 0.4.1
Changes to documentation to facilitate autotest::autotest_package() pass.
Minor changes to vignettes for new patched version.
Added INTERGROWTH-21st Fetal standards, including the Fetal Growth standards, Fetal Doppler standards, Brain Development, and more!
The INTERGROWTH-21st Postnatal Growth (ig_png) conversion functions now accept post-menstrual age in weeks (pma_weeks) as either whole weeks or decimals, e.g. 37 or 37.142857, where 37.142857 is equal to 37 weeks + 1 day, so 37 + 1/7 weeks.
Added checkmate-based verification of inputs, with customised warning() and stop() mechanisms to exported functions. See documentation for allowed data types/lengths in each function argument.
gigs package options via .gigs_options, which lets you customise warning and error behaviour as you see fit.
Better documentation, for compliance with SRR tags.
Internal, non-breaking changes to test suite and documentation.
gigs 0.3.1
Internal, non-breaking changes to test suite and documentation.
Classification functions now use gigs_**z() functions internally for z-scoring. These z-scoring functions, found in gigs-zscoring.R, will be made user-facing once the GIGS guidance documentation is made public.
gigs 0.3.0
This update includes several large changes to the GIGS interface, including changed argument names for some classification and conversion functions. We have only not used a major release as the package may still change with the release of the GIGS guidance document.
BREAKING CHANGES:
Replaced gest_age parameter with gest_days in all circumstances, including ig_nbs growth curve/coefficient tables.
classify_wasting() now takes in gestational age/age data, and uses this to apply the INTERGROWTH-21st Postnatal Growth standard for weight-for-length where appropriate.
Changed out percentile for centile in function names and documentation. Argument names left unchanged, so centiles are still passed into functions as p.
NEW FEATURES:
Added classify_svn() function, which classifies newborns into the categories used in the Lancet’s 2023 Small Vulnerable Newborn (SVN) Series.
INTERNAL/DOCS CHANGES:
Updated coefficient retrieval functions to run ~ 2 times as fast as in version 0.2.3.
Added type checking functions to conversion functions, along with additional unit tests to make sure they work.
Extensive changes to documentation structure. These changes appear internally, where the roxygen documentation uses more @inherit/@inheritParams tags.
Added vignette displaying performance differences between gigs and other, similar R packages, as well as Stata implementations.
gigs 0.2.3
Updated regression equations for the normative body composition standards in ig_nbs, namely fmfga, ffmfga, and bfpfga. These functions give outputs much closer to the published centiles, though a degree of error is still present.
Updated and improved body composition vignette to include the new derivation process.
The coefficient-based ig_nbs/who_gs functions are now 30 to 10x quicker depending on the length of the function inputs (more inputs = slower).
This comes from removal of the data.table dependency - all coefficient retrieval/interpolation now occurs using matrices. The move from using data.table.merge() to subsetting operators has sped up the code significantly.
More consistent subsetting operators and terminology in package source code.
gigs 0.2.2
Optimised interpolation for coefficient-based who_gs_...()/ig_nbs_...() functions. No more for-loops with rbind - instead, the approx() function for interpolation is called on vectors.
Began using data.table for retrieving/merging coefficient tables, in a new helper function named retrive_coefficients(). This approach is slower for smaller inputs than the old data.frame approach, but maintains a much lower time as the number of inputs are increased. It also concentrates all the coefficient-retrieval logic in one group of common functions.
Minor changes to README.md and small refactors for readability.
gigs 0.2.1
Renamed coarse parameter in classify_sga() to severe to better reflect its function; now setting severe to FALSE will cause classify_sga() to specify which SGA cases are below the 3rd percentile.
Documentation updates:
Full use of <sup> tags for pkgdown website
Standardised titles in documentation for conversion + classification functions
Updated “Introduction to gigs” article to include all available classification functions
Extra docs changes which can be viewed in the diffs for this commit (e.g., replacing all @returns Roxygen tags with @return)
gigs 0.2.0
Added INTERGROWTH-21st Postnatal Growth tables + functions for a new weight-for-length standard. These include ig_png_wfl_value2zscore() and ig_png_wfl_percentile2value().
Refactored check_params.R to be more standard-specific. Added some comments as the purpose of this code was unclear. May refactor later to make more self-evidently important.
gigs 0.1.2
classify_sga() now uses < and >, not <= and >= to determine LGA/SGA cut-offs. This aligns this function with the ‘correct’ methodology for calculating size for GA.
gigs 0.1.1
Major documentation updates
Added life6mo dataset
Added ‘Getting Started’ vignette
gigs 0.1.0
Created package
Added a NEWS.md file to track changes to the package.
