Entropy as a Method for Identifying Treatment Resistant Autism Spectrum Disorder
Background: Individuals diagnosed with Autism Spectrum Disorder (ASD) experience lifelong challenges which can impact peer relationships, adaptive functioning, and independent living. Verbal intelligence has proven to be the strongest indicator of outcomes and responsiveness to behavioral intervention, but this property only stabilizes in children between 6 and 8 years of age. Behavioral treatment is the primary intervention for individuals diagnosed with ASD, but it is most effective when delivered as an early intervention strategy for toddlers and very young children. A biomarker which could distinguish treatment resistant subgroups of ASD from would allow for the development and implementation of alternative treatments in an attempt to improve long term outcomes.
Methods: Our study used data from 49 participants made available through the National Database for Autism Research (NDAR). The sample group contained children between 4 and 11 years of age diagnosed with ASD and typically developing peers. Our study used EEG and behavioral measures to explore whether sample entropy analysis of EEG, as developed by Bosl et al. (2011), could distinguish between individuals with ASD and low verbal IQ from their average verbal and typically developing peers.
Results: The analysis we performed found that higher levels of sample entropy were correlated with lower ASD symptoms and better adaptive functioning. ANOVA analysis also suggested that sample entropy could distinguish ASD and typically developing children. Sample entropy was not correlated with verbal IQ and could not distinguish the ASD low verbal IQ group from both ASD with average verbal IQ and typically developing groups.
Conclusion: Researchers interested in identifying biomarkers for treatment resistant ASD should look beyond sample entropy for reliable measures. Sample entropy does appear to play a role in autistic symptomatology, and greater research into its role as a possible indicator of underlying neurological abnormalities should be explored. Researchers may also find value in including sample entropy in longitudinal studies to see how this measure changes with behavioral improvements as a result of behavioral treatment.