Abstract
Introduction
-Sleep difficulties are a common problem in children with autism spectrum disorder (ASD)
Of the children in the Autism Speaks-Autism Treatment Network (ATN) registry who do not have any parent-reported sleep problems at baseline (58\%), a substantial subset have sleep problems reported at first follow-up (20.5\%).
-Developing a predictive model for parent-reported sleep problems using longitudinal data and machine learning could help with treatment and prevention of these problems.
Methods
-A sample of children in the ATN registry without parent-reported sleep problems at baseline and with complete sleep data at first-follow-up was randomly split into training (n=527) and test samples (n=518).
-88 training sample baseline characteristics recommended by a clinician were tested for associations with subsequent sleep problems.
-Model predictors were chosen based on statistical significance and clinical importance, correlation and multicollinearity considerations, and comparison of c-statistics from alternative logistic regression models.
-Given probabilities of sleep problems from the final model, a threshold for classifying children as at risk was selected that yielded at least 85\% sensitivity and maintained maximum associated specificity.
-Each child in the test sample was scored and assigned a predicted sleep problem status based on the model threshold, and comparison of predicted and true status yielded sensitivity, specificity, PPV, NPV, and overall accuracy.
Conclusions
-Among children with ASD, those with ENT problems, asthma, more anxious/depressed and aggressive behavior, and less educated parents at baseline may present with more sleep problems during a follow-up visit.
-In a multivariable model, aggressive behavior independently predicts sleep problems.
-The model’s high sensitivity for identifying children at risk and its accurate prediction of low risk can help with treatment and prevention of sleep problems.
-Further data collection may provide better prediction through methods requiring larger samples.
Users
Please
log in to take part in the discussion (add own reviews or comments).