Rcpchgrowth

Updates

Moves the midparental height calculations into the package and out of the server

Implements the regression to the mean concept (Wright et al.)

Adds tests

Update

Reverting previous change to `create_chart` - creates curves from all LMS values

Update

* Trisomy 21 (AAP) reference added
* reorganisation of `create_chart` function - previously this would create centile lines by calculating a measurement for a given reference, measurement method, sex and z score at fixed decimal age intervals. This generated often uneven lines, particularly when growth velocity is quite steady. This update generates measurement for each interval in the data set meaning no interpolation between ages is necessary, making it quicker. It applies the `default_youngest_reference` flag more clearly, so that where references overlap and two values exist for a given age (eg at 2 y where infant are measured both lying and standing), a measurement for each is returned, allowing lines with clean breaks to be plotted with no gap. It also reduces the number of decimal places in the return object, again speeding things up.

**NOTE:** It is likely this latter change will break the tests for chart-coordinates in the server, so for the moment that is pinned to v4.1.1

**NOTE:** that although the T21 AAP reference has been added, there are no tests for this as there is currently no published gold standard - if anyone knows of one do please get in touch. The accuracy of results relating to this reference for the moment therefore cannot be guaranteed.

Age bands

Since the introduction of CDC, which uses a different centile format collection, the centile band advice is nolonger accurate.
UK-WHO use 2/3 SDS intervals between centiles, so that the collection becomes 0.4th (-2.67SDS), 2nd (-2SDS), 9th (-1.33), 25th (-0.67), 50th (0 SDS), 75th, 91st, 98th, 99.6th - these are termed 'cole-nine-centiles' as a collection name for shorthand
CDC uses a mixture of centiles in their charts - standard is termed 'five-percent-centiles', comprising 5th, 10th, 25th, 50th, 75th, 90th, 95th. For depicting more extremes, 3rd centile is added and is termed 'three-percent-centiles'. The extended BMI release of 2022 included here recognised the significant impact of obesity on population health and sought to reflect this in the charting. It recognised the difficulties of using SDS at higher values discussed elsewhere and introduced a further parameter (sigma) in the LMS dataset, which defines dispersion or spread of all BMI measurements above the 95th centile (p95).
`sigma = sqrt(2) * S * exp(L * (log(BMI/M) - log(P95/M)))`
This has allowed the introduction of a new centile collection, termed 'eight-five-percent-centiles' which include all those in the five-percent-centiles with the additional 85th, 90th, 95th, 99.9th, 99.99th.

While this is all implemented for CDC in RCPCHGrowth, the `centile_band` is text for lay users to know how far the measurement is from the nearest centile. The Measurement class though cannot know what sort of chart it is being plotted on so the solution here is to produce advice based on the standard chart for `reference` and `measurement_method`.

CDC (Centers for Disease Control and Prevention)
Version 4.1 is the first version to include the CDC charts. CDC is made up of WHO (0-2y) and CDC (2y+). The WHO version used by CDC though has been tweaked, presumably to make the disjunction between the references less marked. The effect is that the code base now has 2 versions of WHO - the WHO one and the US interpretation. CDC BMI is included; for BMI measurements > 95th centile some clever maths is introduced to prevent some of the issues that have been detailed in conversations about the accuracy of z scores.

Note that there is no preterm data for the moment. The Canadian Fenton data is what is typically used, but this is licensed and conversations about the legals are ongoing. This means that for all corrected ages < 0 no values are returned, but a meaningful error is included.

There is also a fix to the confusing error raised when Estimated Date of Delivery (EDD) has not yet been reached in a preterm baby.

There are also tests for the same test subjects used in UK-WHO, but recalculated for CDC using the excellent R package [cdcanthro] (https://github.com/CDC-DNPAO/CDCAnthro) by David Freedman, author of the CDC BMI reference.

Tweak to birth date wording - replace "Birth date is today." with "Birth date"