-
1
Across 10 Chinese hospitals, logistic regression was selected for the poststroke anxiety nomogram because it generalized better and had the lowest test-set Brier score among 7 models.
-
2
The 10-variable nomogram combines neurological, metabolic, and psychosocial factors, including NIHSS score, Barthel index, LDL cholesterol, sleep quality, and income.
-
3
SHAP ranked contributions to the fitted logistic regression model, while variable selection also considered redundancy, cost, and bedside feasibility.
-
4
The test partition contained approximately 36 patients, and evaluation sets came from the same multicenter cohort, so the reported performance reflects internal rather than external validation.
-
5
The article contains inconsistencies in cohort descriptions and reported metrics, including conflicting test-set Brier scores and stacking-model AUCs.
