Autism among Children in 2019 National Survey of Children’s Health

Autism is a broad range of complex developmental disabilities in social communication and interaction coincided with restricted, repetitive behaviors. According to research from the National Survey of Children’s Health (NSCH), approximately 28.5% of children aged from 0 to 17 have autism, and the percentage has been increasing in recent years. To control the increasing trend of autism, this study aims to examine the predictors of autism and build a predictive model for autism using the logistic regression model. 

We used the 2019 NSCH dataset in this report. After feature normalization, we built a logistic regression model to predict whether a child is likely to develop autism. The predictive model is further validated by an overall evaluation of the model and has achieved an AUROC score of 0.68. By investigating the correlation among variables and the logistic regression coefficients, we found the dependent variable is most positively correlated with the financial situation of the household and most negatively correlated with the gender of the child. The results imply that older children are more likely to develop autism (p=0.0086, OR=1.093), and children in the family whose income is not able to cover basic living are more likely to develop autism (p<0.001, OR=1.77). Besides, children do not have a low birth weight (p=0.033, OR=0.47) and female children are less likely to develop autism (p<0.001, OR=0.82).