As seen in the Usage example,
growthcleanr creates a vector of flags identifying which
measurements should be included or excluded, and why. This vector can be
added to the original dataset for further review. The flags
growthcleanr sets are text labels, most of which are
descriptive. The text does not always map directly back to the precise
step in the algorithm.
For convenience, the following table maps which algorithm steps in
the pediatric algorithm, as defined by Supplemental File 3 within the Supplementary
Material of the Daymont et al. paper, define exclusions in the
algorithm by the exclusion identifier also prescribed in the algorithm
(“algorithm exc id”). The exclusion type
column identifies the text of the exclusion flag set by
growthcleanr. The notes column at right
identifies minor discrepancies between the algorithm’s step labels and
labels used in comment text in growthcleanr.
| algorithm step | algorithm exc id | exclusion type | notes |
|---|---|---|---|
| 2d | 0 | Include | - |
| 2d | 1 | Missing | - |
| 5b | 2 | Exclude-Temporary-Extraneous-Same-Day | - |
| 7d | - | Swapped-Measurements | - |
| 8f | - | Unit-Error-High | - |
| 8f | - | Unit-Error-Low | - |
| 8f | - | Unit-Error-Possible | Not set in R code |
| 9c | 3 | Exclude-Carried-Forward | - |
| 10c | 4 | Exclude-SD-Cutoff | 10d, 10e |
| 11d | 5 | Exclude-EWMA-Extreme | 11e |
| 11f.ii | 6 | Exclude-EWMA-Extreme-Pair | 11i (R only) |
| 12d.i | 7 | Exclude-Extraneous-Same-Day | 12diii, 12ei, 12f |
| 14f.i | 8 | Exclude-EWMA-8 | Set in 14h (in R) |
| 14f.ii | 9 | Exclude-EWMA-9 | Set in 14h (in R) |
| 14f.iii | 10 | Exclude-EWMA-10 | Set in 14h (in R) |
| 14f.iv | 11 | Exclude-EWMA-11 | Set in 14h (in R) |
| 14f.v | 12 | Exclude-EWMA-12 | Set in 14h (in R) |
| 14f.vi | 13 | Exclude-EWMA-13 | Set in 14h (in R) |
| 14f.vii | 14 | Exclude-EWMA-14 | Set in 14h (in R) |
| 15s | 15 | Exclude-Min-Height-Change | Set in 15q (in R), 15s, 15t |
| 15s | 16 | Exclude-Max-Height-Change | Set in 15q (in R), 15s, 15t |
| 16b | 17 | Exclude-Pair-Delta-17 | - |
| 16b | 18 | Exclude-Pair-Delta-18 | - |
| - | - | Exclude-Pair-Delta-19 | Added flag.both mode |
| 16d | 19 | Exclude-Single-Outlier | - |
| 17a | 20 | Exclude-Too-Many-Errors | - |
| 17a | 21 | Exclude-Too-Many-Errors-Other-Parameter | - |
A researcher wanting to compare two datasets – for example, one
including only measurements growthcleanr flags for
inclusion, and a second which also includes carried forward values –
could extract only measurements where the flag is set to
“Include” for the first set and either
“Include” or “Exclude-Carried-Forward” for the
second set. This flexibility gives each researcher control over how to
incorporate growthcleanr results into their own research.
For the adult algorithm, the following exclusions apply. Note that a
few types are described with similar exclusion names. To distinguish
which of these exclusions were flagged using the adult algorithm,
“-Adult-” has been inserted into each of these exclusion
names.
While formal publication of the adult algorithm is in progress, a detailed narrative describing the algorithm is available at Adult algorithm.
| algorithm step | implementation step | exclusion type | notes |
|---|---|---|---|
| 1h | H BIV | Exclude-Adult-BIV | - |
| 3h | H temp extraneous | - | Nothing excluded in this step |
| 5h | H 100 metric | Exclude-Adult-Hundreds | - |
| 6h | H unit errors | Exclude-Adult-Unit-Errors | - |
| 7h | H transpositions | Exclude-Adult-Transpositions | - |
| 8 | swaps | Exclude-Adult-Swapped-Measurements | - |
| 9h | H same day extraneous | Exclude-Adult-Extraneous-Same-Day | Multiple error codes depending on outcome |
| 9h | H same day extraneous | Exclude-Adult-Identical-Same-Day | Multiple error codes depending on outcome |
| 10ha | H distinct pairs | Exclude-Adult-Distinct-Pairs | - |
| 10hb | H distinct 3+ | Exclude-Adult-Distinct-3-Or-More | - |
| 11h | H mean heights | - | Nothing excluded in this step |
| 13 | distinct 1 | Exclude-Adult-Distinct-Single | - |
| 14h | H error load | Exclude-Adult-Too-Many-Errors | - |
| 1w | W BIV | Exclude-Adult-BIV | - |
| 2w | W repeated values | - | Nothing excluded in this step |
| 3w | W temp extraneous | - | Nothing excluded in this step |
| 4w | W weight cap | Exclude-Adult-Weight-Cap | - |
| 4w | W weight cap | Exclude-Adult-Weight-Cap-Identical | - |
| 5w | W hundreds | Exclude-Adult-Hundreds | - |
| 6w | W unit errors | Exclude-Adult-Unit-Errors | - |
| 7w | W transpositions | Exclude-Adult-Transpositions | - |
| 9w | W extreme EWMA | Exclude-Adult-EWMA-Extreme | - |
| 10w | W same day extraneous | Exclude-Adult-Extraneous-Same-Day | Multiple error codes depending on outcome |
| 10w | W same day extraneous | Exclude-Adult-Identical-Same-Day | Multiple error codes depending on outcome |
| 11wa | W distinct ordered pairs | Exclude-Adult-Distinct-Ordered-Pairs | - |
| 11wb | W moderate EWMA | Exclude-Adult-EWMA-Moderate | - |
| 12w | W weight cap influence | Exclude-Adult-Possibly-Impacted-By-Weight-Cap | - |
| 14w | W error load | Exclude-Adult-Too-Many-Errors | - |
| Most weight steps | W repeated value exclusion | paste(proposed, "-RV") |
When a first repeated value is excluded, all other repeated values are excluded |