# A Clinical Prediction Model for Genetic Risk in Children with GDD/ID: A Retrospective Study

## TL;DR

This study creates a model to predict which children with developmental delays are likely to have genetic causes, helping doctors decide who needs genetic testing.

## Contribution

A novel clinical prediction model using Lasso and logistic regression to identify high genetic risk in children with GDD/ID.

## Key findings

- Four clinical predictors (craniofacial, visceral, growth abnormalities, and family history) were identified for genetic risk.
- The model achieved an AUC of 0.734 with good calibration and clinical utility in validation.
- The model is proposed as a screening tool to prioritize genetic testing in GDD/ID patients.

## Abstract

Objectives: Global Developmental Delay (GDD) and Intellectual Disability (ID) are prevalent neurodevelopmental disorders with significant disability burden, and genetic factors play a crucial role in their etiology. This study aimed to develop and validate a clinical prediction model for identifying children with GDD/ID at high genetic risk, facilitating targeted genetic testing. Methods: We retrospectively analyzed clinical data of children with GDD/ID treated at Nanjing Children’s Hospital from January 2019 to December 2023. Children with comorbid Autism Spectrum Disorder (ASD) were excluded. The dataset was randomly split into training and validation sets (7:3 ratio). Lasso regression was used to identify potential predictive factors for positive genetic test results, followed by multivariable logistic regression to select independent predictors, which were incorporated into a nomogram. Model performance was evaluated by discrimination, calibration, and clinical utility using decision curve analysis in both sets. Results: Four independent predictors—craniofacial abnormalities, visceral abnormalities, physical growth abnormalities, and family history of ID—were identified. The resulting nomogram demonstrated an area under the curve (AUC) of 0.734., with good calibration and positive net benefit on decision curve analysis. Validation confirmed the reliability of the model. Conclusions: We developed a clinically applicable prediction model to identify high genetic risk among children with GDD/ID without ASD. This model may serve as a preliminary screening tool to assist clinicians in prioritizing genetic testing and improving diagnostic efficiency in clinical practice.