Growth charts for small sample sizes using unsupervised clustering: Application to canine early growth
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Breed-specific growth curves (GCs) are needed for neonatal puppies, but breed-specific data may be insufficient. We investigated an unsupervised clustering methodology for modeling GCs by augmenting breed-specific data with data from breeds having similar growth profiles. Puppy breeds were grouped by median growth profiles (bodyweights between birth and Day 20) using hierarchical clustering on principal components. Median bodyweights for breeds in a cluster were centered to that cluster's median and used to model cluster GCs by Generalized Additive Models for Location, Shape and Scale. These were centered back to breed growth profiles to produce cluster-scale breed GCs. The accuracy of breed-scale GCs modeled with breed-specific data only and cluster-scale breed GCs were compared when modeled from diminishing sample sizes. A complete dataset of Labrador Retriever bodyweights (birth to Day 20) was split into training (410 puppies) and test (460 puppies) datasets. Cluster-scale breed and breed-scale GCs were modelled from defined sample sizes from the training dataset. Quality criteria were the percentages of observed data in the test dataset outside the target growth centiles of simulations. Accuracy of cluster-scale breed GCs remained consistently high down to sampling sizes of three. They slightly overestimated breed variability, but centile curves were smooth and consistent with breed-scale GCs modeled from the complete Labrador Retriever dataset. At sampling sizes ≤ 20, the quality of breed-scale GCs reduced notably. In conclusion, GCs for neonatal puppies generated using a breed-cluster hybrid methodology can be more satisfactory than GCs at purely the breed level when sample sizes are small.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it