Approaches for Predicting the Occurrence of Challenging Behaviors in Individuals with Autism Spectrum Disorder: A...

The presentation of autism spectrum disorder can be highly heterogeneous, and a significant percentage of individuals with autism exhibit challenging behaviors. However, the frequency and severity of challenging behaviors can vary significantly over time, even for the same person, and it is often unclear what triggers a behavior. Being able to predict the occurrence of challenging behaviors has significant potential for improving the safety of individuals with ASD, as well as quality of life for them and their caregivers. Given the large heterogeneity found in the presentation of autism, most predictions need to be personalized to the individual or at least a subgroup of individuals. This work reviews the state of research in the field for predicting behaviors, ranging from short-term predictions just a few moments before a behavior occurs to predicting the probability of challenging behaviors a day in advance. Several reviewed approaches provide promising initial results, but few studies have been conducted on populations large and diverse enough to determine their generalizability.