In this paper, we introduce an end-to-end machine learningbased system for classifying autism spectrum disorder (ASD) using facial attributes such as expressions, action units, arousal, and valence. Our system classifies ASD using representations of different facial attributes from convolutional neural networks, which are trained on images in the wild. Our experimental results show that different facial attributes used in our system are statistically significant and improve sensitivity, specificity, and F1 score of ASD classification by a large margin. In particular, the addition of different facial attributes improves the performance of ASD classification by about 7% which achieves a F1 score of 76%.
Maternal history of Major Depressive Disorder (MDD) dramatically increases children's risk for developing depression, highlighting the critical need for further research on the specific processes involved in the intergenerational transmission of depression. Although previous research suggests that maternal depression may adversely affect the quality of mother-child interactions, less is known about the role of maternal MDD in the moment-to-moment changes in affect that occur during these interactions. The goal of this project, therefore, was to examine synchrony of facial displays of affect during a positive (Vacation Planning) and a negative (Issues Discussion) mother-child interaction, and how this synchrony may be impacted by maternal history of MDD. In doing so, we examined both concurrent and lagged synchrony of facial affect. We recruited 341 mother-child dyads (child average age ϭ 9.30 years; 50.1% girls; 71.6% Caucasian) with and without a maternal history of MDD. Facial electromyography (EMG), continuously recorded during those tasks, was used to index mother and child facial affect. We found that a maternal history of MDD was associated with reduced concurrent synchrony and lagged synchrony (mother facial affect predicting changes in child facial affect) of positive affect during Vacation Planning. Reduced concurrent mother-child synchrony of positive affect during the discussion was also associated with an increase in child self-reported sad affect from before to after the discussion. These findings provide promising initial evidence for how the dynamic exchange of positive affect during mother-child interactions may be disrupted in families with maternal MDD history.
The way individuals process socio-affective information is thought to impact their responses to social interactions, but research testing the relation between these processes is scarce, particularly among children. This study examined if children's attention to socio-affective stimuli was associated with their autonomic nervous system (ANS) reactivity during parent-child interactions. Children's sustained attention to facial expressions of emotion (afraid, happy, sad) was indexed using the late positive potential (LPP) event-related potential (ERP) component during a computer-based task. To measure ANS reactivity, children's respiratory sinus arrhythmia (RSA) was assessed at baseline and during positive and negative parent-child discussions. Enhanced LPP amplitudes in response to all emotional facial expressions, reflecting greater sustained attention to socio-affective stimuli, were associated with increased RSA reactivity during parent-child discussions. These results show correspondence between two psychophysiological substrates of emotion processing in healthy children and highlight how these systems may be synergistic forces contributing to emotion reactivity.The Research Domain Criteria (RDoC; https://www.nimh.nih.gov/researchpriorities/rdoc/) initiative grew out of the 2008 National Institute of Mental Health (NIMH) Strategic Plan to link new classifications of psychiatric disorders to recent advances in neurobiology (Cuthbert, 2014; NIMH, 2008). Importantly, this initiative has provided researchers with opportunities to identify core processes that cut across traditional diagnostic boundaries and to examine the full range of functioning from normal to abnormal across human development (Cuthbert & Insel, 2013). To do so, the RDoC framework divides processes of human behavior into five domains of functioning (i.e.,
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.