# Development and comparison of multivariate diagnostic models for rapidly progressive central precocious puberty in girls: the role of serum osteocalcin

## TL;DR

This study develops a diagnostic model for rapidly progressive central precocious puberty in girls, showing that including serum osteocalcin improves prediction accuracy.

## Contribution

The study introduces a novel diagnostic model for RP-CPP that incorporates serum osteocalcin, enhancing predictive performance.

## Key findings

- A model including osteocalcin achieved an AUC of 0.973 with high sensitivity and specificity.
- Models with osteocalcin had lower AIC and prediction error rates compared to those without.
- The model performed well in both internal and external validation sets.

## Abstract

To develop a diagnostic prediction model for rapidly progressive central precocious puberty (RP-CPP) and evaluate the contribution of osteocalcin(OC) to the model.

For a total of 411 girls who met the criteria for central precocious puberty were selected. Of these, 219 were included in the training set, 87 in the internal validation set, and 105 in the external validation set. Binary logistic regression was used to construct the model. The model fit and diagnostic accuracy were assessed using the Akaike Information Criterion (AIC), calibration curves, and the area under the receiver operating characteristic curve(AUC). The model was presented in the form of a nomogram. Internal and external validations of the model were performed.

Diagnostic models for RP-CPP were developed both with and without the inclusion of OC. Among all models, those that included OC consistently demonstrated smaller AIC values, higher AUC values, and lower prediction error rates. A model incorporating the duration of breast development, serum OC levels, mean ovarian volume, endometrial presence/absence, and breast Tanner staging demonstrated superior performance. The AUC for diagnosing RP-CPP was 0.973, with a sensitivity of 91.6% and specificity of 92.5%. The model performed well in the internal and external validation sets, demonstrating good clinical application value.

The inclusion of OC helps improve the predictive performance of the model. For the diagnosis of RP-CPP in girls, a model can be chosen that includes the duration of breast development, serum OC levels, mean ovarian volume, endometrial presence/absence, and breast Tanner staging. However, all samples were from a single center, and multicenter validation is still needed.