Building a Predictive Model of ADHD among Children

Attention Deficit/Hyperactivity Disorder (ADHD) is one of the most common neurodevelopmental disorders of childhood. According to the Centers of Disease Control and Prevention (CDC), the estimated number of children ever diagnosed with ADHD nationalwide is 6.1 million (9.4%). Among those children, 6 in 10 with ADHD had at least one other mental, emotional, or behavioral disorder that may have long-lasting impacts on their development.

 In this research, we investigated possible risk factors related to development of ADHD among children and identified the most significant positive and negative factors through logistic regression. We used the 2020 National Survey of Children’s Health survey data containing 42,777 complete data samples with features ranging from demographic information to the child’s family condition. The response variable is whether a child has ever been diagnosed with ADHD.

After processing the dataset, we built a logistic regression model to predict whether a child will develop ADHD. By investigating the logistic regression coefficients, we found that parents’ physical and mental health, the family’s financial ability to cover basic living expenses, and whether the parents are divorced are all risk factors. The logistic regression model has achieved an AUROC score of 0.73, with 0.67 true positive rate (TPR) and 0.324 false positive rate (FPR). This predictive model is helpful for healthcare professionals to identify and reduce the risk for the children that are prone to the development of ADHD.