The Diagnostic Efficacy of a Facial Recognition Screening Tool for Children with Autism Spectrum Disorder

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by impairments in social communication and by restrictive and repetitive patterns of behavior. An important quantitative social deficit, which is oftentimes not discussed but seen in those with ASD is impaired facial recognition, or the ability to identify and discriminate faces based on memory of one’s unique facial features. Despite the absence of this impairment in the criteria of the Diagnostic and Statistical Manual of Mental Disorders, it is arguably a core deficit in those with ASD. Evidence of this impairment is seen across the board – in behavioral, neuroimaging and eye-tracking studies. The purpose of this study was to examine whether a standardized facial recognition tool, the Faces Immediate and Faces Delayed subtests of the Children’s Memory Scale (CMS; Cohen, 1997), could be used to distinguish between those with and without the disorder.
Participants consisted of 26 children reliably diagnosed with ASD and 62 age-matched controls between the ages of five to sixteen. Data were utilized from the National Institute of Mental Health (NIMH)’s National Database for Autism Research (NDAR) as well as from the Psychological Evaluation Research and Counseling Clinic (PERCC) and the Diagnostic and Research Institute for Autism Spectrum Disorders at Hofstra University. Receiver operating characteristic (ROC) analysis was conducted to assess the diagnostic efficacy of this tool. Research questions were explored to consider whether areas of social functioning were associated with facial recognition scores.
The results of this study indicated that the CMS Faces Immediate subtest showed acceptable discrimination between those with and without ASD based on an area under the curve (AUC) of .706 (p < .05). The sensitivity of the tool for the general sample ranged from 65% to 77% and the specificity ranged from above 50% to 61%. Upwards of 55 children were correctly identified based on a selected cut-point of 10. The CMS Faces Immediate subtest was also found in an acceptable range when discriminating between males and females with and without ASD with AUC’s that ranged from .738 to .735 (p < .05). The CMS Faces Delayed subtest produced an AUC of .657 (p < .05), which indicated that the CMS Faces Immediate was a more discriminative tool than the CMS Faces Delayed. Facial recognition scores were positively correlated with willingness to join group activities. The implications of these results for detection and for the use of a time and cost effective future screening tool are discussed. Based on these findings, the importance of facial recognition interventions for children with ASD is also explored.