# Risk factors and a prediction model for ASD symptoms in Chinese preschool children

## TL;DR

This study identifies risk factors and creates a prediction model for autism symptoms in young Chinese children, aiming to improve early screening in community settings.

## Contribution

A practical nomogram model for predicting ASD symptoms in Chinese preschoolers using familial and child-level factors.

## Key findings

- Lower parental fondness, inconsistent parenting beliefs, poor sleep quality, and family history of mental disorders are risk factors for ASD symptoms.
- Higher parental education and balanced caregiving time are protective factors against ASD symptoms.
- The nomogram model achieved moderate discrimination (AUC = 0.757) and demonstrated good calibration and clinical utility.

## Abstract

The global prevalence of autism spectrum disorder (ASD) is rising, creating an urgent need for practical early screening tools, especially in community and resource-limited settings. This study aimed to identify key risk factors and develop an individualized prediction model for ASD symptoms in Chinese preschool children.

A cross-sectional study was conducted in 2024, involving 13,641 children aged 3–6 years from 32 kindergartens in Guizhou Province, China. ASD symptoms were screened using the Autism Behavior Checklist. Predictor variables were selected via LASSO regression with 10-fold cross-validation. A multivariable logistic regression model was constructed and presented as a nomogram. Model discrimination was evaluated by the area under the receiver operating characteristic curve (AUC) with bootstrapped 95% confidence intervals (CI). Calibration was assessed using calibration curves and the Hosmer-Lemeshow test, and clinical utility was measured by decision curve analysis.

Among the participants, 324 (2.38%) screened positive for ASD symptoms. Multivariable analysis identified several independent risk factors: lower degree of fondness for the child (OR = 1.53, 95% CI: 1.29–1.81), inconsistency in parenting beliefs (OR = 1.17, 95% CI: 1.06–1.30), poorer sleep quality (OR = 1.55, 95% CI: 1.33–1.80), and a family history of mental disorders (OR = 2.80, 95% CI: 1.81–4.32). Higher parental education (OR = 0.86, 95% CI: 0.78–0.94) and balanced caregiving time (OR = 0.82, 95% CI: 0.76–0.88) were protective factors. The nomogram demonstrated moderate discrimination (AUC = 0.757, 95% CI: 0.731–0.782), was well-calibrated, and provided a net clinical benefit for threshold probabilities between 0.1% and 19.6%.

We successfully developed and validated a practical nomogram that integrates multiple familial and child-level factors for predicting ASD symptoms. This tool exhibits good performance and clinical applicability, offering a valuable approach for early community-based screening of preschool children.