Caregiver-Child Interaction as an Effective Tool for Identifying Autism Spectrum Disorder: Evidence from EEG Analysis

doi:10.21203/rs.3.rs-3240924/v1

Download PDF

Research Article

Caregiver-Child Interaction as an Effective Tool for Identifying Autism Spectrum Disorder: Evidence from EEG Analysis

https://doi.org/10.21203/rs.3.rs-3240924/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 14 Dec, 2023

Read the published version in Child and Adolescent Psychiatry and Mental Health →

You are reading this latest preprint version

Background

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder that affects individuals across their lifespan. Early diagnosis and intervention are crucial for improving outcomes. However, current diagnostic methods are often time-consuming, and costly, making them inaccessible to many families. In the current study, we aim to test caregiver-child interaction as a potential tool for screening children with ASD in clinic.

Methods

We enrolled preschool children, including ASD children with or without developmental delay (DD), and typical development (TD) children, along with their caregivers. Behavioral indicators were derived from video encoding of caregiver-child interaction, including social involvement of children (SIC), interaction time (IT), response of children to social cues (RSC), time for caregiver initiated social interactions (GIS) and time for children initiated social interactions (CIS)). Power spectral density (PSD) values were calculated by EEG signals simultaneously recorded. ASD core symptoms were evaluated by Childhood Autism Rating Scale (CARS) and Autism Diagnostic Observation Schedule-Calibrated Severity Scores (ADOS-CSS). Partial Pearson correlation analysis was used in both ASD groups to investigate the correlation between behavioral indicators scores and ASD symptom severity/PSD values. Receiver operating characteristic (ROC) analysis was used to describe the discrimination accuracy of behavioral indicators.

Results

We enrolled 85 participants (Mean age: 4.90 ± 0.65 years, 70.6% male). SIC scores negatively correlated with CARS (p = 0.006) and ADOS-CSS (p = 0.023) in the ASD with DD group. PSD values elevated in ASD groups compared to TD (all p < 0.05), associated with SIC (theta band: p = 0.005; alpha band: p = 0.003) but not IQ levels. SIC was effective in identifying both ASD groups (sensitivity/specificity: ASD children with DD, 76.5%/66.7%; ASD children without DD, 82.6%/82.2%).

Conclusion

Our results verified the behavioral paradigm of caregiver-child interaction as an efficient tool for early ASD screening, further validated by the PSD of alpha and theta bands.

Autism Spectrum Disorder

Caregiver-child Interaction

Social Interaction

Electroencephalography

Autism spectrum disorder (ASD), a complex neurodevelopmental disorder with a current prevalence rate of 2.3% in the United States, is characterized by social skills deficits [1, 2]. Early identification is a crucial step in improving the prognosis [3] and ensuring timely access to early intervention strategies for children with ASD [4]. ASD diagnosis typically occurs at the age of five years [5]; however, signs of social abnormalities in autistic children [6], such as difficulty in perceiving and recognizing other people's faces, emotional expressions [7], eyes [8], movements [9], and mental states [10], can manifest as early as infancy and imped their daily functioning. Current diagnosis of ASD is built on time-consuming assessment by specialized developmental-behavioral pediatrician. With 90% of people with ASD residing in low- and middle-income countries and regions, there is a critical need for low-cost screening tools that do not require trained professionals [11].

The widely used caregiver-reported measures for scanning ASD, like the Modified Checklist (M-CHAT) [12] have limitations in terms of accuracy despite being cost-effective, such as relying on subjective reporting, potentially introducing biases due to caregivers’ beliefs and experiences, as well as education differences [13]. Relying solely on screening questionnaires may therefore overlook subtle or nuanced symptoms and identify only the most apparent developmental and behavioral issues [14]. Furthermore, diagnostic scales such as Autism Diagnostic Observation Schedule (ADOS) [15] and Autism Diagnostic Interview-Revised (ADI-R) [16], require trained experts to conduct lengthy interviews, which limits their applicability given the high incidence of ASD. Therefore, screening assessments of social skills in children with ASD should be focused on real-life social situations, which would lead to a more objective and reliable identification of ASD and can be conducted in a cost-effective manner.

Caregiver-child interaction is an important foundation for children's cognitive, linguistic, and socio-emotional development [17], and serves as a crucial starting point for acquiring interactive skills, including social communication skills [18, 19]. This interaction provides language and social stimuli that support the development of social skills [20], as caregivers provide feedback [21] on their children's behavior to aid in their developmental process. As the most familiar social environment for children [22], caregiver-child interaction is applied in the early screening of ASD children to assess the performance of interaction and create conditions that maximize social interaction [23]. The content of caregiver-child interaction provides a direct source for clinician or other therapists to guide family intervention for ASD, making it an important tool for extensive early screening, diagnosis, and intervention of ASD children with significant health and economic implications.

Advancements in electrophysiology tools have led to increased ecological validity of research on social interaction [24, 25], such as the use of EEG. EEG allows researchers to examine the natural electrical activity of the brain during different stimuli and conditions with high time resolution, portability, and tolerance to movement. Moreover, EEG signals reflect postsynaptic activity, while EEG power indicates the excitability of neuronal groups [26]. Studies [27–29] have demonstrated a connection between behaviors observed during caregiver-infant interactions and infants' EEG activity. A Bernier, SD Calkins and MA Bell [29] found that higher maternal positive affect and lower physical stimulation during play with infants at 5 months predicted increased frontal alpha and theta EEG resting-state power. Researchers also have found that infants who later develop autism exhibit higher alpha power and lower theta power for static faces relative to objects [30], in contrast to typical developmental infants [31]. While EEG has been widely used to study ASD, current studies do not reflect naturalistic social interaction with their caregivers. Therefore, using EEG to investigate the early social interaction of ASD children with caregivers has the potential to investigate the neural mechanism.

In this study, we utilized free play derived from the ADOS assessment as a caregiver-child interaction task to collect EEG signals and introduce reliable electrophysiological indicators into natural face-to-face interactions between children with ASD and their caregivers. To account for variability in interactions, we employed micro-coding to identify participation status offline. As ASD is a complex and heterogeneous clinical syndrome that includes individuals with varying levels of intellectual disability, language, and cognitive skills [2, 32], and individuals with good cognitive skills may use scripted social behaviors to navigate social interactions [33, 34]. Thus, we plan to recruit preschool ASD children with different IQ level and typically developing (TD) children, along with their caregivers, to engage in free play while simultaneously recording EEG and video signals. This will enable us to explore and evaluate effective indicators of atypical social patterns of ASD children and investigate the neural mechanism and also investigated whether the IQ level of young children affects the social performance assessment of caregiver-child interaction.

Based on previous evidence, we hypothesize that:

Based on caregiver-child interaction, the novel paradigm of behavioral coding can effectively differentiate between typically developing children and ASD children with varying levels of IQ.

Regardless of IQ, ASD group has increased alpha power and theta power in ASD group than TD children, and these PSD values are correlated with the performance of caregiver-child interactions in ASD group.

2.1 Participants

85 children aged between 3–5 years were recruited from the Xinhua Hospital Affiliated to Shanghai Jiaotong University, including 40 ASD children (23 ASD children with developmental delay (DD), 17 ASD children without DD), and 45 typically developing (TD) children (Fig. 1). All ASD children were evaluated and scored by experienced clinicians who specialize in identifying ASD, based on the Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition, DSM-5). And ASD children scored above the threshold on both the Childhood Autism Rating Scale (CARS) and the Autism Diagnostic Observation Schedule (ADOS). Children with a history of neurological or genetic disorders, or those unable to engage in 3–5 minutes of interaction with a caregiver while wearing an EEG cap, were excluded from the study. TD children had no history of developmental disease and did not have any first-degree relatives with ASD. Caregivers of TD children were asked to complete the Chinese validated version of Social Responsiveness Scale (SRS) and only those who screened negative were included in this study. Intelligence quotient (IQ) of both ASD and TD children were assessed using the Wechsler Preschool and Primary Scale of Intelligence (WPPSI)[35]. The ASD children with DD group had Full Scale IQ (FSIQ) score at least two standard deviations below the average, while the ASD children without DD group were matched with the TD group in terms of IQ. The study was conducted in accordance with the Declaration of Helsinki, with approval from the Ethical Committee of the Xinhua Hospital Affiliated to Shanghai Jiaotong University, and in compliance with all applicable laws and regulations. Written informed consent was obtained from the children’s caregivers. All necessary biosecurity and institutional safety protocols were followed during the study.

2.2 Measures

2.2.1 Caregiver-child dyads coding

To encourage natural and spontaneous interaction between the children and their most familiar caregivers, we provided a 3–5 minutes session of free play involving puzzles and blocks [36, 37]. Caregivers and children were seated at a table and wearing EEG caps. Their interaction was recorded by a camera while EEG was captured simultaneously (see Fig. 1B). Due to potential distractions, we used the ELAN (EUDICO Linguistic Annotator) program to capture and categorize the organic and social interactions between caregivers and children in recorded videos offline and analyzed using an event-based coding system that calculated 5 behavioral indicators per second (see Supplementary). These indicators included Social Involvement of Children (SIC), Interaction Time (IT), Response of Children to Social Cues (RSC), time for Caregiver Initiated Social interactions (GIS) and time for Children Initiated Social interactions (CIS). Table S1 provides definitions and examples for each code.

2.2.2 Behavior assessment

The ADOS is a widely utilized standardized diagnostic tool for ASD in both clinical and research settings [38] by evaluating social interaction, communication, and play in individuals with high-risk ASD. To account for variability in score across different modules of the ADOS, a mapping of ADOS module total scores to Calibrated Severity Scores (CSS) has been suggested [39]. The CSS system transforms the ADOS total score into a standardized score ranging from 1 to 10, with higher scores indicating greater severity of autistic features, based on the child's actual age and language abilities. This standardized scoring system helps to provide a more accurate representation of the severity of autistic features in individuals with ASD.

The CARS [40] is a tool used to diagnose and assess the severity of ASD in children and consisted of 15 items rated on a 7-point scale from one to four; higher scores indicating a higher level of impairment. We further categorized these items into three subscales [41]: Social Impairment (SI), Negative Emotionality (NE), and Distorted Sensory Response (DSR). The criterion validity for CARS with a cut-off of 30 resulted in sensitivity of 0.86 and specificity of 0.79 [42].

The SRS is a commonly used to evaluate social deficits associated with ASD and other developmental disorders in clinical and research settings [43, 44]. The SRS provides a total score and scores on five subscales: social awareness, social cognition, social communication, social motivation, and autistic mannerisms. Multiple studies [45, 46] have reported high reliability and validity of the SRS, including the Chinese Mandarin version, which showed internal consistency for the total scale of 0.871–0.922, and test-retest reliability of 0.81–0.94. Receiver operating characteristic (ROC) analyses revealed that the SRS accurately classified 69.2–97.2% of youth with ASD [47].

2.2.3 EEG recording and pre-processing

During a 3–5 minutes free play session between caregivers and children, we recorded EEG signals using a high-density 128-channel Electrical Geodesics, Inc (EGI) system with a vertex reference (channel Cz) and a sampling rate of 1000 Hz. To ensure high-performance data, we kept impedances below 100KΩ [48]. MATLAB [49] and the EEGLab [50] toolbox were used to process offline. A bandpass filter between 0.5 and 45 Hz and a 50 Hz notch filter were consistently applied to the continuous EEG data. We retained 82 channels for analysis after excluding 46 peripheral "skirt channels" to reduce noise and muscle artifact, and interpolated any noisy electrodes [51]. EEG segments were excluded from further analysis in the following cases: when the caregiver-child interaction disengaged or when the segment contained artifacts. Afterward, 1-s epoch segments are created from the preprocessed EEG data. Independent component analysis (ICA) [52] is used to identify and eliminate eye blink, movement, and muscle activity artifacts after physically confirming artifacts rejection by visual examination. Prior to spectral analysis, the data was re-referenced to the average of the mastoids.

2.2.4 Alpha and Theta Power Spectral Density (PSD)

The power spectral density (PSD) of the theta (4–7 Hz) and alpha (8–13 Hz) frequency bands are computed for each EEG channel by applying the FFT algorithm, squaring the resulting signal to obtain amplitude, transforming the bilateral spectrum into a unilateral spectrum, and dividing it by the frequency resolution.

2.3 Statistical Analysis

The statistical analyses employed SPSS software, version 23, with the Wilcoxon rank sum test and Analysis of Variance (ANOVA) used to compare mean ± standard error of continuous variables between the ASD and TD groups. Significance level (α) was set at 0.05. Prior to analysis, behavioral data and PSD values were transformed by square root to meet normal distributional assumptions. General linear model (GLM) was used to compare groups on EEG PSD.

The application of Receiver Operating Characteristics (ROC)[53] analysis was implemented with the intent of computing the Area Under the Curve (AUC)[54], serving as a metric for the discriminative capability of behavioral indicators and PSD values in distinguishing between ASD groups and TD group. The ROC analysis, by providing sensitivity and 1 - specificity outcomes across a comprehensive range of potential thresholds, empowers the selection of a putatively "optimum" cutoff for each distinct comparative grouping. To further evaluate the diagnostic performance, we scrutinized the accuracy, sensitivity, and specificity of these behavioral indicators and PSD values within distinct ASD subgroups (i.e., ASD children with DD group, ASD children without DD group) and the TD group to explore potential discrepancies pertaining to IQ.

A series of partial Pearson correlation analyses were conducted to evaluate the dimensional relationships among EEG PSD and IQ level and behavioral indicators (SIC, IT, RSC, GIS, CIS) in all groups.

3.1 Participant Information

The demographic and clinical characteristics of the three groups were shown in Table 1. The average age of participants was 4.87 (0.18) years in ASD children with DD group, 4.80 (0.26) years in ASD children without DD, and 4.98 (0.15) years in TD children group. There were no significant differences in age among the three groups. The ASD children with DD group (82.6%, p = 0.036) and the ASD children without DD group (94.1%, p < 0.001) had a significant higher proportion of male participants than TD children group (51.1%). Both verbal (p < 0.001) and performance IQ (p < 0.001) were significantly lower in the ASD children with DD group than in the other two groups. After adjusting for the effect of gender, there were no differences in symptom severity scores between ASD children with DD group and ASD children without DD group (p = 0.123).

Table 1

The demographic and clinical (mean (SE)) characteristics of three groups
	ASD children with DD group	ASD children without DD group	TD children group	F-Value	P
N	23	17	45
Female (%)	4 (17.4)	1 (5.9)	22 (48.9)	6.906	0.002^ab
Age	4.87 (0.18)	4.80 (0.26)	4.98 (0.15)	0.692	0.504
P_IQ	55.13 (1.91)	85.88 (3.74)	87.53 (3.92)	30.311	< 0.001^ac
V_IQ	47.91 (1.38)	67.12 (4.85)	77.65 (2.41)	31.117	< 0.001^ac
FSIQ	48.52 (1.33)	82.22 (4.55)	91.30 (2.26)	112.132	< 0.001^ac
ADOS-CSS	6.96 (0.45)	7.06 (0.47)		-0.158	0.875
SRS	93.22 (4.86)	85.31 (4.80)		2.061	0.159
Social awareness	11.13 (0.58)	10.88 (0.46)		0.103	0.750
Social cognition	19.17 (1.14)	16.62 (0.76)		2.846	0.100
Social communication	32.96 (1.64)	30.50 (1.88)		0.953	0.335
Social motivation	15.00 (0.97)	13.88 (0.89)		0.666	0.420
Autistic mannerisms	14.96 (1.38)	13.44 (1.42)		0.559	0.459
CARS	35.87 (3.67)	34.15 (3.04)		2.485	0.123
SI	23.46 (0.66)	22.63 (1.63)		0.305	0.584
NE	6.63 (0.187)	6.19 (0.452)		1.439	0.238
DSR	6.89 (0.26)	6.75 (0.50)		0.038	0.847
Notes: Values for age and all clinical test scores are presented as mean (SE). Abbreviations: ASD, autism spectrum disorder; DD, developmental delay; TD, typically developing; P_IQ, performance Intelligence Quotient; V_IQ, verbal Intelligence Quotient; FSIQ, full scale Intelligence Quotient; ADOS-CSS, Autism Diagnostic Observation Schedule- Calibrated Severity Score; SRS, Social Responsiveness Scale; CARS, Childhood Autism Rating Scale; SI, Social Impairment; NE, Negative Emotionality; DSR, Distorted Sensory Response. ^a, Significant difference between ASD children with DD group and TD group. ^b, Significant difference between ASD children without DD group and TD group. ^c, Significant difference between ASD children with DD group and ASD children without DD group.

3.2 Caregiver-child dyads coding

Figure 2 presents a comparison of caregiver-child interaction performance among the groups, after gender adjustment. As shown in Table 2, the SIC, IT and CIS scores in both ASD groups were significantly lower than those in TD group (all p values < 0.05). Among the groups, only ASD children with DD group demonstrated a significant decrease in RSC (p = 0.001) compared to the TD children group. In both ASD groups, GIS were significantly higher than in TD children group (all p values < 0.01).

Table 2

The behavioral indictors and PSD values (mean (SE)) of three groups
Indicators		Mean (SE)	gender-adjusted mean	P value
SIC	ASD children with DD group	0.424(0.031)	-0.214^a	< 0.001^c
	ASD children without DD group	0.467(0.048)	-0.170^b	0.001^c
	TD children group	0.638(0.020)
IT	ASD children with DD group	0.694(0.036)	-0.161^a	< 0.001^c
	ASD children without DD group	0.729(0.056)	-0.126^b	0.021^c
	TD children group	0.855(0.013)
RSC	ASD children with DD group	0.569(0.043)	-0.176^a	0.010^c
	ASD children without DD group	0.640(0.056)	-0.111^b	0.289
	TD children group	0.730(0.033)
GIS	ASD children with DD group	0.850(0.017)	0.195^a	< 0.001^c
	ASD children without DD group	0.797(0.033)	0.142^b	0.003^c
	TD children group	0.650(0.023)
CIS	ASD children with DD group	0.212(0.029)	-0.225^a	< 0.001^c
	ASD children without DD group	0.241(0.038)	-0.195^b	< 0.001^c
	TD children group	0.439(0.022)
Theta PSD	ASD children with DD group	1.730(0.102)	0.310 ^a	0.016 ^c
	ASD children without DD group	1.741(0.107)	0.309 ^b	0.041 ^c
	TD children group	1.394(0.506)
Alpha PSD	ASD children with DD group	1.214(0.065)	0.218 ^a	0.012 ^c
	ASD children without DD group	1.178(0.742)	0.173 ^b	0.041 ^c
	TD children group	0.973(0.036)
Notes: Abbreviations: ASD, autism spectrum disorder; DD, developmental delay; TD, typically development; SIC, Social Involvement of Children; IT, Interaction Time; RSC, Response of children to Social Cues; GIS, Caregiver Initiated Social interaction; CIS, Children Initiated Social interactions. ^a, gender-adjusted mean between ASD children with DD group and TD group; b, gender-adjusted mean between ASD children without DD group and TD group; ^c, p < 0.05. The p values are adjusted for multiple comparison (Bonferroni).

We investigated whether the behavioral indicators were associated with severity scores among autistic children (see Fig. 3 and Table S2). The SIC (r = -0.587, p = 0.004; r = -0.728, p = 0.002), IT (r = -0.639, p = 0.001, r = -0.725, p = 0.002), RSC (r = -0.609, p = 0.003, r = -0.668, p = 0.006) and GIS (r = 0.586, p = 0.004; r = 0.754, p = 0.001) showed significant correlations with the total score of CARS in both ASD groups. And in ASD children (with or without DD) group, besides the total score of CARS, SIC (r = -0.371, p = 0.022), IT (r = -0.425, p < 0.001), RSC (r = -0.342, p = 0.033), GIS (r = 0.394, p = 0.013) were also significantly correlated with the social impairment subscale of CARS.

Additionally, in the ASD children without DD group, significant correlations were found between SRS cognitive subscale and SIC (r = -0.519, p = 0.039), GIS (r = 0.549, p = 0.028), CIS (r = -0.510, p = 0.043). In ASD children with DD group, SIC was not only significantly correlated with CARS (r = -0.605, p = 0.006) total scores but also correlated with ADOS-CSS score (r = -0.493, p = 0.023).

Examining the ROC for each of the 3 groups in differentiating ASD from TD for 5 behavioral indicators (see Table 3). CIS (sensitivity = 70%, specificity = 91.1%) was most effective in the full samples. Although IT (sensitivity = 88.2%, specificity = 44.4%) and RSC (sensitivity = 87.0%, specificity = 46.7%) are the most sensitive indicators in in differentiating ASD children without DD and ASD children with DD respectively, their specificity is too low. SIC (sensitivity = 76.5%, specificity = 66.7%; sensitivity = 82.6%, specificity = 82.2%) has the second highest sensitivity in identifying both groups of ASD children, and the specificity is higher than both of the above two indicators (see Fig. 4).

Table 3

Results of AUC analyses in 3 groups
	Indicators	Sensitivity	Specificity	Accuracy
Full samples (N = 85)	SIC	67.5%	82.2%	82.1%
	IT	57.5%	82.2%	74.6%
	RSC	60.0%	77.8%	77.8%
	CIS	70.0%	91.1%	86.2%
	GIS	67.5%	91.1%	83.2%
ASD children without DD group and TD children group (N = 62)	SIC	76.5%	66.7%	75.4%
	IT	88.2%	44.4%	66.2%
	RSC	41.2%	86.7%	62.3%
	CIS	64.7%	91.1%	83.1%
	GIS	52.9%	91.1%	76.3%
ASD children with DD group and TD children group (N = 68)	SIC	82.6%	82.2%	87.1%
	IT	69.6%	82.2%	80.8%
	RSC	87.0%	46.7%	74.4%
	CIS	78.3%	86.7%	88.5%
	GIS	78.3%	93.3%	88.3%
Notes: Abbreviations: ASD, autism spectrum disorder; DD, developmental delay; TD, typically development; SIC, Social Involvement of Children; IT, Interaction Time; RSC, Response of children to Social Cues; GIS, Caregiver Initiated Social interaction; CIS, Children Initiated Social interactions.

3.3 EEG Power and its correlations with behavioral indicators and IQ level

PSD was calculated for the theta and alpha bands in three groups (ASD children with DD, ASD children without DD, and TD children). Channel-to-channel comparisons were conducted between the groups (see Fig. 5), there were significant differences in the PSDs of theta and alpha band between the ASD children with DD group and TD children group, as well as between the ASD children without DD group and TD children group, after adjusting for gender. Specifically, the gender-adjusted mean difference was 0.310 (p = 0.005) for theta and 0.218 (p = 0.004) for alpha between the ASD children with DD group and TD children group, and 0.309 (p = 0.014) for theta and 0.173 (p = 0.041) for alpha between the ASD children without DD group and TD children group.

We also calculate the accuracy, sensitivity, and specificity of PSD in the theta and alpha frequency bands (see Fig. 4). PSD of Alpha band (sensitivity = 75.0%, specificity = 60.0%; sensitivity = 52.9%, specificity = 86.7%; sensitivity = 73.9%, specificity = 62.2%) is more sensitive than theta band (sensitivity = 50%, specificity = 88.9%; sensitivity = 52.9%, specificity = 89.9%; sensitivity = 69.6%, specificity = 68.9%), and PSD of theta band is more specific in all of 3 groups.

We investigated the association between PSD in the theta and alpha bands and behavioral indicators while adjusting for verbal IQ levels and sex in full sample (performance IQ did not correlate with PSD values across all groups). Details were shown in Table 4. Specifically, we observed statistically significant correlations (all FDR-p < 0.05) between the five primary behavioral indicators of cognitive intelligence and the PSD values of alpha and theta brain waves. Notably, these correlations remained significant even after adjusting for verbal IQ and gender. The SIC (FDR-p = 0.003, FDR-p = 0.005), IT (FDR-p = 0.003, FDR-p = 0.005), and CIS (FDR-p = 0.003, FDR-p = 0.010), exhibited robust associations with PSD. Among these indicators, SIC demonstrated the strongest correlation with PSD (see Table 4).

Table 4

Results of correlation analyses in full sample
	Alpha PSD		Theta PSD
	Partial Pearson correlation, controlled for sex	Partial Pearson correlation, controlled for sex and Verbal IQ	Partial Pearson correlation, controlled for sex	Partial Pearson correlation, controlled for sex and Verbal IQ
SIC
r	-0.528^a	-0.414^a	-0.522^a	-0.410^a
p	0.002	0.003	< 0.001	0.005
IT
r	-0.474 ^a	-0.419 ^a	-0.476 ^a	-0.401 ^a
p	0.002	0.003	< 0.001	0.005
RSC
r	-0.353 ^a	-0.200	-0.346 ^a	-0.194
p	0.006	0.137	0.008	0.148
CIS
r	-0.505^a	-0.404 ^a	-0.464 ^a	-0.537 ^a
p	0.001	0.003	< 0.001	0.010
GIS
r	-0.404 ^a	-0.269	-0.410 ^a	-0.279 ^a
p	0.003	0.054	0.001	0.044
Notes: r presented as Pearson’s correlation coefficients, p, FDR p value. Abbreviations: PSD, Power Spectral Density; IQ, Intelligence Quotient; SIC, Social involvement of the child; IT, Interaction Time; RSC, Response of children to Social Cues; GIS, Caregiver Initiated Social interaction; CIS, Children Initiated Social interactions. ^a, FDR. p < 0.05.

In this study, we introduce a novel screening method that combines behavioral indicators and EEG measurements during real-life caregiver-child interactions to effectively identify atypical social patterns in ASD children. We found that the behavioral indicators of both ASD groups were worse than TD group and SIC is the most effective indicator in screening ASD children with or without DD respectively. And both theta and alpha band PSD values can distinguish between children with ASD and TD children in natural interactive settings, and these indicators are primarily correlated with social behavior rather than IQ level.

Consistent with our first hypothesis, ASD children exhibited decreased levels in behavioral indicators, including dyadic interaction (IT), participation to social cues (SIC), initiation (CIS), and responsiveness to social cues (RSC) during caregiver-child interaction tasks, regardless of the presence of intellectual disability. Furthermore, the more severe the social impairment symptoms were, the lower levels of the child's initiation of social interactions, response to social cues, and engagement with social cues were. While IT and RSC respectively exhibited the highest sensitivity in identifying both groups of children with ASD, they presented a specificity under 50%. This significantly elevated the risk of misdiagnosis[53, 54]. Conversely, SIC though second in sensitivity, offered higher specificity than the two highest sensitivity indicators. This suggests that SIC serves as a more consistent indicator of social deficits in children with ASD, irrespective of their IQ levels. SIC reflects both level of active social initiation during caregiver-child interaction and response behaviors to social cues initiated by caregivers. Previous studies on high-risk ASD population (e.g., ASD’s siblings) also highlighted the synchrony and infant/maternal responsiveness computed by frequency and duration of gaze, positive affect and vocalizations during infant-mother interaction can help predict the outcome of autism [55, 56]. Our findings indicated that such behavior indicators especially SIC in our study could effectively distinguish ASD children from TD children.

The second finding in the present study was that significant differences in EEG PSD during caregiver-child interaction was observed between the ASD children and TD children. The PSD of alpha band in the full sample group was the most sensitive indicator in screening ASD children and negative correlations between PSD of alpha band values and behavioral indicators, particularly SIC in both ASD groups. This highlights the potential of EEG measures in assessing social functioning in ASD. Increased alpha PSD values during caregiver-children interaction in our study indicating that ASD children may have atypical neural responses to social interaction. Some researchers [57, 58] proposed that this atypical neural activity may contribute to the atypical social impairments observed in ASD, as it may reflect a decreased ability to process and respond to social cues effectively [59]. And previous studies [60, 61], also reported higher alpha power are in response to social stimuli such as faces and emotions in ASD children group, compared to TD children and positively associated with autistic trait expression [62].

Similar to the alpha frequency band, our findings revealed a higher left parietal theta power in both ASD children groups compared to the TD children group, and a negative correlation between left parietal theta power and SIC and CARS scores. Previous studies have found that parietal lobe is connected to the reception of social stimuli [63], which are among the most potent triggers of theta rhythms [64]. The increased theta band power in children with ASD may be indicative of atypical neural processing during social interactions[65, 66], reflecting difficulties in integrating information from multiple sources[67]. And the atypical theta activity in ASD could potentially result in inadequate processing and interpretation of social cues[67].

Furthermore, significant correlations between behavioral indicators and PSD values were even persistent after controlling for verbal IQ and gender differences. However, after controlling for behavior indicators (SIC, IT, RSC, GIS, CIS) and gender, relationship between verbal IQ and PSD values was not significant in any groups, which indicates that the PSD values is mainly associated with social function. This finding was different with previous researches indicating cognitive ability related to alpha power [68–70], however, most of these studies examined the performance of Alzheimer's patients in the related cognitive paradigm rather than ASD children. Our findings of PSD of alpha and theta band provides further neural mechanism for the ability of behavioral indicators employed in this study to identify social impairments in children with ASD.

There are several limitations that must be taken into consideration. First, the sample size was relatively small, larger sample sizes may be necessary to further validate the effectiveness of these screening assessments. Furthermore, the correlation trend between social function and brain activity may be more pronounced in male ASD participants. This finding may be attributed to the limited number of female ASD children included in our study. Thus, future studies with a larger and more balanced sample of male and female participants could provide insight into the gender differences in social interaction patterns among children with ASD in real-world settings. Second, the present study only focused on caregiver-child interactions, rather than interactions with peers. Future research could explore the use of screening assessments and EEG indicators in peer interactions to gain a better understanding of the social difficulties in ASD children. Lastly, it is important to note that our participants were exclusively Chinese school-age children aged 3–5 years. Therefore, our findings may not be generalizable to other age groups, cultures, or regions. Including participants from different age groups, countries, and regions could provide opportunities for developing new ASD screening paradigms and further validating the generalizability of our findings.

In conclusion, the behavioral paradigm of caregiver-child interaction in our study has been verified as an efficient method for clinical screening young children with ASD. And the effectiveness of this approach was further validated by the examination of the PSD of alpha and theta bands. Indeed, further study with large sample, longitudinal design and multi-modal data are welcomed for the validation and embedded mechanism.

ASD, autism spectrum disorder; DD, developmental delay; TD, typically developing; ROC, receiver operating characteristic; AUC, area under the receiver operating curve; P_IQ, performance Intelligence Quotient; V_IQ, verbal Intelligence Quotient; FSIQ, full scale Intelligence Quotient; ADOS-CSS, Autism Diagnostic Observation Schedule- Calibrated Severity Score; SRS, Social Responsiveness Scale; CARS, Childhood Autism Rating Scale; SI, Social Impairment; NE, Negative Emotionality; DSR, Distorted Sensory Response. SIC, Social Involvement of Children; IT, Interaction Time; RSC, Responding to Social Cues; GIS, caregivers Initiated Social interaction; CIS, children Initiated Social interaction; PSD, power spectral density.

Ethics approval and consent to participate

The studies involving human participants were reviewed and approved by Ethical Committee of Shanghai Jiao Tong University School of Medicine Affiliated Xinhua Hospital. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

Consent for publication

We have obtained consent to publish from the legal caregiver or guardian for children to report individual patient data.

Availability of data and materials

Access to the deidentified participant research data must be approved by the research ethics board on a case-by-case basis, please contact the corresponding authors ([email protected],, [email protected] ) for assistance in data access request.

Competing interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

This study was supported by grants from the National Natural Science Foundation of China (82125032, 81930095 and 81761128035), the Science and Technology Commission of Shanghai Municipality (19410713500 and 2018SHZDZX01), the Shanghai Municipal Commission of Health and Family Planning (GWV-10.1-XK07, 2020CXJQ01, 2018YJRC03), the Shanghai Clinical Key Subject Construction Project (shslczdzk02902), Innovative research team of high-level local universities in Shanghai (SHSMU-ZDCX20211100), the Guangdong Key Project (2018B030335001), the National Nature Science Foundation of China (82204048), the National Natural Science Foundation of China (82001771) the Shanghai Municipal Commission of Health and Family Planning (20214Y0125).

Authors’ contributions

FL, J-BG, and LW conceived the study. LD, W-ZH, Q-LZ, LW, YD, Z-LC, Y-QL, TR, L-LZ, J-BG, and FL contributed to the acquisition of research data. LD and W-ZH conducted the literature review, data analysis and drafted the manuscript. All authors contributed in revising the article critically for important intellectual content, and approved the final version for publication. FL is the guarantor who takes full responsibility for the work as a whole.

Acknowledgements

We gratefully acknowledged all research staff who had contributed to patient recruitment and data collection in the Xinhua Hospital Affiliated to Shanghai Jiaotong University.

Battle DE. Diagnostic and Statistical Manual of Mental Disorders (DSM). Codas. 2013;25(2):191–2.
Maenner MJ, Shaw KA, Bakian AV, Bilder DA, Durkin MS, Esler A, Furnier SM, Hallas L, Hall-Lande J, Hudson A et al. Prevalence and Characteristics of Autism Spectrum Disorder Among Child ren Aged 8 Years - Autism and Developmental Disabilities Monitoring Ne twork, 11 Sites, United States, 2018. Morbidity and mortality weekly report Surveillance summaries (Washing ton, DC: 2002), 70(11):1–16.
Gliga T, Jones EJ, Bedford R, Charman T, Johnson MH. From early markers to neuro-developmental mechanisms of autism. Dev Rev. 2014;34(3):189–207.
Dover CJ, Le Couteur A. How to diagnose autism. Arch Dis Child. 2007;92(6):540–5.
Neimy H, Pelaez M, Carrow J, Monlux K, Tarbox J. Infants at risk of autism and developmental disorders: Establishing early social skills. In : Educational Publishing Foundation; 2017: 6–22.
Zwaigenbaum L, Bryson S, Rogers T, Roberts W, Brian J, Szatmari P. Behavioral manifestations of autism in the first year of life. Int J Dev Neurosci. 2005;23(2):143–52.
Baron-Cohen S. The development of a theory of mind in autism: deviance and delay? Psychiatr Clin North Am. 1991;14(1):33–51.
Mundy P, Sigman M, Kasari C. The theory of mind and joint-attention deficits in autism. Understanding other minds: Perspectives from autism. Oxford University Press; 1994: 181–203.
Williams JHG, Whiten A, Singh T. A Systematic Review of Action Imitation in Autistic Spectrum Disorder. J Autism Dev Disord. 2004;34(3):285–99.
Baron-Cohen S. Mindblindness: An Essay on Autism and Theory of Mind. In.: The MIT Press; 1995.
Tomlinson M, Swartz L. Imbalances in the knowledge about infancy: The divide between rich and poor countries. Infant Mental Health Journal. 2003;24(6):547–56.
Kleinman JM, Robins DL, Ventola PE, Pandey J, Boorstein HC, Esser EL, Wilson LB, Rosenthal MA, Sutera S, Verbalis AD, et al. The modified checklist for autism in toddlers: a follow-up study investigating the early detection of autism spectrum disorders. J Autism Dev Disord. 2008;38(5):827–39.
Magnitude. and moderators of bias in observer ratings: A meta-analysis. In., vol. 4. US: American Psychological Association; 1999: 403–424.
Heunis T, Aldrich C, Peters JM, Jeste SS, Sahin M, Scheffer C, de Vries PJ. Recurrence quantification analysis of resting state EEG signals in autism spectrum disorder – a systematic methodological exploration of technical and demographic confounders in the search for biomarkers. BMC Med. 2018;16(1):101.
Lord C, Rutter M, Goode S, Heemsbergen J, Jordan H, Mawhood L, Schopler E. Autism diagnostic observation schedule: a standardized observation of communicative and social behavior. J Autism Dev Disord. 1989;19(2):185–212.
Chojnicka I, Ploski R. [Polish version of the ADI-R (Autism Diagnostic Interview-Revised)]. Psychiatr Pol. 2012;46(2):249–59.
Norton ES, Manning BL, Harriott EM, Nikolaeva JI, Nyabingi OS, Fredian KM, Page JM, McWeeny S, Krogh-Jespersen S, MacNeill LA, et al. Social EEG: A novel neurodevelopmental approach to studying brain-behavior links and brain-to-brain synchrony during naturalistic toddler-parent interactions. Dev Psychobiol. 2022;64(3):e22240.
Peacock-Chambers E, Ivy K, Bair-Merritt M. Primary Care Interventions for Early Childhood Development: A Systematic Review. Pediatrics 2017, 140(6).
Tomasello M, Farrar MJ. Joint attention and early language. Child Dev. 1986;57(6):1454–63.
Olson J, Masur EF. Mothers' labeling responses to infants' gestures predict vocabulary outcomes. J Child Lang. 2015;42(6):1289–311.
Bornstein MH, Tamis-Lemonda CS, Hahn CS, Haynes OM. Maternal responsiveness to young children at three ages: longitudinal analysis of a multidimensional, modular, and specific parenting construct. Dev Psychol. 2008;44(3):867–74.
Sigman M, Mundy P. Social attachments in autistic children. J Am Acad Child Adolesc Psychiatry. 1989;28(1):74–81.
Corona LL, Weitlauf AS, Hine J, Berman A, Miceli A, Nicholson A, Stone C, Broderick N, Francis S, Juárez AP, et al. Parent Perceptions of Caregiver-Mediated Telemedicine Tools for Assessing Autism Risk in Toddlers. J Autism Dev Disord. 2021;51(2):476–86.
Wang Q, Han Z, Hu X, Feng S, Wang H, Liu T, Yi L. Autism Symptoms Modulate Interpersonal Neural Synchronization in Child ren with Autism Spectrum Disorder in Cooperative Interactions. Brain Topogr, 33(1):112–22.
Bhat AN, McDonald NM, Eilbott JE, Pelphrey KA. Exploring cortical activation and connectivity in infants with and wit hout familial risk for autism during naturalistic social interactions: A preliminary study. Infant Behav Dev, 57:101337.
Bell MA, Cuevas K. Using EEG to Study Cognitive Development: Issues and Practices. J Cogn Dev. 2012;13(3):281–94.
Swingler MM, Perry NB, Calkins SD, Bell MA. Maternal behavior predicts infant neurophysiological and behavioral attention processes in the first year. Dev Psychol. 2017;53(1):13–27.
Kraybill JH, Bell MA. Infancy predictors of preschool and post-kindergarten executive function. Dev Psychobiol. 2013;55(5):530–8.
Bernier A, Calkins SD, Bell MA. Longitudinal Associations Between the Quality of Mother-Infant Interactions and Brain Development Across Infancy. Child Dev. 2016;87(4):1159–74.
Dawson G, Jones EJ, Merkle K, Venema K, Lowy R, Faja S, Kamara D, Murias M, Greenson J, Winter J, et al. Early behavioral intervention is associated with normalized brain activity in young children with autism. J Am Acad Child Adolesc Psychiatry. 2012;51(11):1150–9.
Jones EJ, Venema K, Lowy R, Earl RK, Webb SJ. Developmental changes in infant brain activity during naturalistic social experiences. Dev Psychobiol. 2015;57(7):842–53.
Lord C, Elsabbagh M, Baird G, Veenstra-Vanderweele J. Autism spectrum disorder. The Lancet. 2018;392(10146):508–20.
Barnhill GP. Asperger syndrome: A guide for parents and educators. National Association of School Psychologists.
Marchese F. High Functioning Autism: Relevant Variables to Consider from the Rorsc hach Interpersonal Domain: Immaculata University.
Pizer J, ElBassiouny A. Wechsler Preschool and Primary Scale of Intelligence (WPPSI). In: The Wiley Encyclopedia of Personality and Individual Differences. edn.; 2020: 473–475.
Kasari C, Gulsrud AC, Wong C, Kwon S, Locke J. Randomized controlled caregiver mediated joint engagement intervention for toddlers with autism. J Autism Dev Disord. 2010;40(9):1045–56.
Kim SH, Grzadzinski R, Martinez K, Lord C. Measuring treatment response in children with autism spectrum disorder: Applications of the Brief Observation of Social Communication Change to the Autism Diagnostic Observation Schedule. Autism. 2019;23(5):1176–85.
Akshoomoff N, Corsello C, Schmidt H. The Role of the Autism Diagnostic Observation Schedule in the Assessment of Autism Spectrum Disorders in School and Community Settings. Calif School Psychol. 2006;11:7–19.
Gotham K, Risi S, Pickles A, Lord C. The Autism Diagnostic Observation Schedule: revised algorithms for improved diagnostic validity. J Autism Dev Disord. 2007;37(4):613–27.
Schopler E, Reichler RJ, DeVellis RF, Daly K. Toward objective classification of childhood autism: Childhood Autism Rating Scale (CARS). J Autism Dev Disord. 1980;10(1):91–103.
DiLalla DL, Rogers SJ. Domains of the Childhood Autism Rating Scale: relevance for diagnosis and treatment. J Autism Dev Disord. 1994;24(2):115–28.
Moon SJ, Hwang JS, Shin AL, Kim JY, Bae SM, Sheehy-Knight J, Kim JW. Accuracy of the Childhood Autism Rating Scale: a systematic review and meta-analysis. Dev Med Child Neurol. 2019;61(9):1030–8.
Kamio Y, Inada N, Moriwaki A, Kuroda M, Koyama T, Tsujii H, Kawakubo Y, Kuwabara H, Tsuchiya KJ, Uno Y, et al. Quantitative autistic traits ascertained in a national survey of 22 529 Japanese schoolchildren. Acta Psychiatr Scand. 2013;128(1):45–53.
Charman T, Baird G, Simonoff E, Loucas T, Chandler S, Meldrum D, Pickles A. Efficacy of three screening instruments in the identification of autistic-spectrum disorders. Br J Psychiatry. 2007;191:554–9.
Nguyen PH, Ocansey ME, Miller M, Le DTK, Schmidt RJ, Prado EL. The reliability and validity of the social responsiveness scale to measure autism symptomology in Vietnamese children. Autism Res. 2019;12(11):1706–18.
Wigham S, McConachie H, Tandos J, Le Couteur AS. The reliability and validity of the Social Responsiveness Scale in a UK general child population. Res Dev Disabil. 2012;33(3):944–50.
Cen CQ, Liang YY, Chen QR, Chen KY, Deng HZ, Chen BY, Zou XB. Investigating the validation of the Chinese Mandarin version of the Social Responsiveness Scale in a Mainland China child population. BMC Psychiatry. 2017;17(1):51.
Gabard-Durnam LJ, Wilkinson C, Kapur K, Tager-Flusberg H, Levin AR, Nelson CA. Longitudinal EEG power in the first postnatal year differentiates autism outcomes. Nat Commun. 2019;10(1):4188.
Arns M, Bruder G, Hegerl U, Spooner C, Palmer DM, Etkin A, Fallahpour K, Gatt JM, Hirshberg L, Gordon E. EEG alpha asymmetry as a gender-specific predictor of outcome to acute treatment with different antidepressant medications in the randomized iSPOT-D study. Clin Neurophysiology: Official J Int Federation Clin Neurophysiol. 2016;127(1):509–19.
Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9–21.
Frohlich J, Reiter LT, Saravanapandian V, DiStefano C, Huberty S, Hyde C, Chamberlain S, Bearden CE, Golshani P, Irimia A, et al. Mechanisms underlying the EEG biomarker in Dup15q syndrome. Mol Autism. 2019;10:29.
Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9–21.
Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem. 1993;39(4):561–77.
Metz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283–98.
Kellerman AM, Schwichtenberg AJ, Abu-Zhaya R, Miller M, Young GS, Ozonoff S. Dyadic Synchrony and Responsiveness in the First Year: Associations with Autism Risk. Autism Res. 2020;13(12):2190–201.
Wan MW, Green J, Elsabbagh M, Johnson M, Charman T, Plummer F. Quality of interaction between at-risk infants and caregiver at 12–15 months is associated with 3-year autism outcome. J Child Psychol Psychiatry. 2013;54(7):763–71.
Wang Z, Goerlich KS, Luo Y-j, Xu P, Aleman A. Social-specific impairment of negative emotion perception in alexithymia. Soc Cognit Affect Neurosci. 2022;17(4):387–97.
Jones EJ, Venema K, Earl RK, Lowy R, Webb SJ. Infant social attention: an endophenotype of ASD-related traits? J Child Psychol Psychiatry. 2017;58(3):270–81.
Edgar JC, Heiken K, Chen YH, Herrington JD, Chow V, Liu S, Bloy L, Huang M, Pandey J, Cannon KM, et al. Resting-state alpha in autism spectrum disorder and alpha associations with thalamic volume. J Autism Dev Disord. 2015;45(3):795–804.
Cantonas L-M, Seeber M, Mancini V, Bochet A, Kojovic N, Rihs TA, Schaer M. Impaired alpha and beta modulation in response to social stimuli in children with autism spectrum disorder. medRxiv 2022:2022.2001.2025.22269816..
Mathewson KJ, Jetha MK, Drmic IE, Bryson SE, Goldberg JO, Schmidt LA. Regional EEG alpha power, coherence, and behavioral symptomatology in autism spectrum disorder. Clin Neurophysiol. 2012;123(9):1798–809.
Carter Leno V, Tomlinson SB, Chang SA, Naples AJ, McPartland JC. Resting-state alpha power is selectively associated with autistic traits reflecting behavioral rigidity. Sci Rep. 2018;8(1):11982.
Hirsch J, Zhang X, Noah JA, Dravida S, Naples A, Tiede M, Wolf JM, McPartland JC. Neural correlates of eye contact and social function in autism spectrum disorder. PLoS ONE. 2022;17(11):e0265798.
Orekhova EV, Stroganova TA, Posikera IN, Elam M. EEG theta rhythm in infants and preschool children. Clin Neurophysiol. 2006;117(5):1047–62.
Coben R, Clarke AR, Hudspeth W, Barry RJ. EEG power and coherence in autistic spectrum disorder. Clin Neurophysiol. 2008;119(5):1002–9.
Pop-Jordanova N, Zorcec T, Demerdzieva A, Gucev Z. QEEG characteristics and spectrum weighted frequency for children diagnosed as autistic spectrum disorder. Nonlinear Biomedical Physics. 2010;4(1):4.
Kana RK, Libero LE, Moore MS. Disrupted cortical connectivity theory as an explanatory model for autism spectrum disorders. Phys Life Rev. 2011;8(4):410–37.
Schmid RG, Tirsch WS, Scherb H. Correlation between spectral EEG parameters and intelligence test vari ables in school-age children. Clin neurophysiology: official J Int Feder ation Clin Neurophysiol, 113(10):1647–56.
Mostile G, Giuliano L, Monastero R, Luca A, Cicero CE, Donzuso G, Dibilio V, Baschi R, Terranova R, Restivo V, et al. Electrocortical networks in Parkinson's disease patients with Mild Cognitive Impairment. The PaCoS study. Parkinsonism Relat Disord. 2019;64:156–62.
Doppelmayr M, Klimesch W, Stadler W, Pöllhuber D, Heine C. EEG alpha power and intelligence. Intelligence. 2002;30(3):289–302.

No competing interests reported.

0804Supplementary.docx

Download PDF

Journal Publication

published 14 Dec, 2023

Read the published version in Child and Adolescent Psychiatry and Mental Health →

Editorial decision: Major revision
06 Oct, 2023
Reviews received at journal
06 Oct, 2023
Reviewers agreed at journal
26 Sep, 2023
Reviewers invited by journal
20 Aug, 2023
Editor assigned by journal
12 Aug, 2023
Submission checks completed at journal
07 Aug, 2023
First submitted to journal
07 Aug, 2023

You are reading this latest preprint version

Caregiver-Child Interaction as an Effective Tool for Identifying Autism Spectrum Disorder: Evidence from EEG Analysis

Status:

Journal Publication

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

1. Introduction

2. Method

2.1 Participants

2.2 Measures

2.2.1 Caregiver-child dyads coding

2.2.2 Behavior assessment

2.2.3 EEG recording and pre-processing

2.2.4 Alpha and Theta Power Spectral Density (PSD)

2.3 Statistical Analysis

3. Results

3.1 Participant Information

3.2 Caregiver-child dyads coding

3.3 EEG Power and its correlations with behavioral indicators and IQ level

4.Discussion

5. Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1