Abstract:

Social communication impairment is one of the core symptoms of autism spectrum disorder (ASD). Previous research has predominantly interpreted this impairment through individual-level factors, such as deficits in social attention or empathy, while overlooking the inherently interpersonal nature of social communication. The perspective of interpersonal synchrony offers a novel framework to understand this core deficit. According to the bio-behavioral synchrony theory, successful social interaction depends on the synchronization between interaction partners at behavioral, physiological, and neural levels. Moreover, existing ASD screening tools (e.g., M-CHAT, ADOS-2) mainly rely on subjective rating scales, which are time-consuming and require extensive professional training. Although emerging technologies such as eye-tracking and EEG provide more objective screening possibilities—such as detecting abnormal gaze patterns or EEG power differences—laboratory-based tasks often lack ecological validity and fail to capture behavior in real social contexts.

Therefore, this project aims to investigate the mechanisms of behavioral (motor and facial) and neurophysiological synchrony during parent-child interactions in children with ASD, and to examine their relationships with social communication impairments. In addition, this project explores the feasibility of using parent-child synchrony as an objective indicator for early ASD screening.

Study 1 will use behavioral experiments to assess facial and motor synchrony during free play in parent-child dyads involving children with ASD. Emotional synchrony will be objectively evaluated using the facial expression coding system through the FaceReader software, while motor synchrony will be automatically quantified through AlphaPose, a regional multi-person pose estimation model.

Study 2 will employ physiological recording systems and functional near-infrared spectroscopy (fNIRS) hyper-scanning to examine synchrony in respiratory sinus arrhythmia (RSA)—an index of autonomic nervous system functioning—and inter-brain neural synchrony during parent-child interactions. Dyads will participate in three interactive tasks—free play, cooperative drawing, and conflict discussion—to simulate common real-life scenarios. The goal is to elucidate the neurophysiological mechanisms underlying behavioral synchrony and their associations with social communication impairments in children with ASD.

Study 3 will adopt a longitudinal design to explore whether early behavioral and neurophysiological synchrony between high-risk toddlers and their parents can predict ASD diagnosis one year later. This study employs a classic machine learning algorithm—Support Vector Machine (SVM)—to predict children’s social communication skills and diagnostic outcomes. To address the issue of a small sample size, leave-one-out cross-validation (LOOCV) will be used, in which each participant is iteratively left out for prediction to enhance the robustness of the model. The trained model will be evaluated using several metrics, including accuracy, sensitivity, specificity, and the receiver operating characteristic (ROC) curve. The central hypothesis of this study is that that an SVM model built on parent-child emotional, behavioral, physiological, and neural synchrony data can accurately classify high-risk autistic toddlers and typically developing toddlers, and reliably predict whether high-risk toddlers will be diagnosed with autism one year later.

This project offers a novel theoretical framework for understanding the pathological mechanisms of social communication impairments through the lens of parent-child synchrony. We propose that the core nature of social communication deficits in children with ASD constitutes a "synchrony disorder," wherein impairments in synchrony may lead to failures in social interactions, subsequently hindering the development of social relationships and adaptive functioning. Furthermore, this project innovatively proposes the use of parent-child synchrony as an objective biomarker, combined with artificial intelligence techniques, to develop an early screening model for ASD. By integrating synchrony features across behavior (facial expressions, motor actions), autonomic regulation (RSA), and neural activity (inter-brain connectivity), the project aims to develop a machine learning-based predictive model to achieve automated, objective, and precise early detection of autism.

Key words: autism spectrum disorder, social communication impairment, parent-child synchrony, early screening