Behavior and interaction imaging at 9 months of age predict autism/intellectual disability in high-risk infants with West syndrome

ABSTRACT

Automated behavior analysis is a promising tool to overcome current assessment limitations in psychiatry. At 9 months of age, we recorded 32 infants with West syndrome (WS) and 19 typically developing (TD) controls during a standardized mother-infant interaction. We computed infant hand movements (HM), speech turn-taking of both partners (vocalization, pause, silences, overlap) and motherese. Then, we assessed whether multimodal social signals(social signal processing) and interactional synchrony at 9 months could predict outcomes (autism spectrum disorder (ASD) and intellectual disability (ID)) of infants with WS at 4 years. At follow-up, 10 infants developed ASD/ID (WS+). The best machine learning reached 76.47% accuracy classifying WS vs. TD and 81.25% accuracy classifying WS+ vs. WS−. The 10 best features to distinguish WS+ and WS− included a combination of infant vocalizations and HM features combined with synchrony vocalization features. These data indicate that behavioral and interaction imaging was able to predict ASD/ID in high-risk children with WS.

INTRODUCTION

Behavior and interaction imaging is a promising domain of affective computing to explore psychiatric conditions^1,2,3. Regarding child psychiatry, many researchers have attempted to identify reliable indicators of neurodevelopmental disorders (NDD) in high-risk populations (e.g., siblings of children with autism) during the first year of life to recommend early interventions^4,5. However, social signals(social signal processing) and any alterations of them are very difficult to identify at such a young age⁶. Also, exploring the quality and dynamics of early interactions is a complex endeavor. It usually requires (i) the perception and integration of multimodal social signals(social signal processing) and (ii) an understanding of how two interactive partners synchronize and proceed in turn taking^7,8.

Affective computing offers the possibility to simultaneously analyze the interaction of several partners while considering the multimodal nature and dynamics of social signals(social signal processing)and behaviors⁹. To date, few seminal studies have attempted to apply social signal processing to mother-infant interactions with or without a specific condition, and these studies have focused on speech turns (e.g., Jaffe et al.¹⁰), motherese¹¹, head movements¹², hand movements¹³, movement kinematics², and facial expressions³.

Here, we focused on West syndrome (WS), a rare epileptic encephalopathy with early-onset (before age 1 year), and a high risk of NDD outcomes, including one-third of WS children showing later autism spectrum disorder (ASD) and/or intellectual disability (ID). We recruited 32 infants with WS and 19 typically developing (TD) controls to participate in a standardized early mother-infant interaction protocol and followed infants with WS to assess outcomes at 4 years of age. We aim to explore whether multimodal social signals(social signal processing) and interpersonal synchrony of infant-mother interactions at 9 months could predict outcomes.

DISCUSSION

To the best of our knowledge, this is the first study to apply multimodal social signal processing to mother-infant interactions in the context of WS. Combining speech turns and infant HM during an infant-mother interaction at 9 months significantly predicted the development of ASD or severe to moderate ID at 4 years of age in the high-risk children with WS. Confusion matrices showed that the classification errors were not random, enhancing the interest of the computational method proposed here. Also, the best contributing features for the performed classifications differed when classifying WS vs. TD and WS+ vs. WS−. Infant HMs were the most significant features to distinguish WS versus TD, probably reflecting the motor impact due to acute WS encephalopathy. For classifying WS+ vs. WS−, the contribution of infant audio features and synchrony features became much more relevant combined with several HM features.

We believe that the importance of synchrony and reciprocity during early interactions is in line with recent studies that have investigated the risk of ASD or NDD during the first year of life from home movies (e.g., refs. ^11,24), from prospective follow-up of high-risk infants such as siblings (e.g., refs. ^4,28) or infants with WS (e.g., ref. ¹⁴), and from prospective studies assessing tools to screen risk for autism (e.g., ref. ²⁹). In the field of ASD, synchrony, reciprocity, parental sensitivity, and emotional engagement are now proposed as targets of early interventions³⁰, which could prevent early interactive vicious circles. Parents of at-risk infants try to compensate for the lack of interactivity of their child by modifying their stimulation and therefore sometimes reinforcing the dysfunctional interactions²⁴. Early identification of these interactive targets is especially useful among babies with neurological comorbidities because delays in developmental milestones and impairments in early social interactions are not sufficient to predict ASD.

Similarly, we believe that the importance of HM in distinguishing WS vs. TD on one hand, and WS+ vs. WS− on the other hand, is also in line with the studies that investigated the importance of non-social behaviors for investigating the risk of ASD or NDD during the first year of life. For example, studying home movies, Purpura et al. found more bilateral HM and finger movements in infants who will later develop ASD³¹. Similarly, several prospective follow-up studies of high-risk siblings^32,33,34,35 or retrospective studies on home movies^36,37 reported specific motor atypical repertoire in infants with ASD.(social signal processing)

In ASD, early social signals (social signal processing) have previously been assessed with automatized and computational procedures, focusing on eye-tracking at early stages^38,39,40, vocal productions⁴¹, analysis of the acoustics of first utterances or cry episodes⁴², but none was done in an interactive setting. Our study proposed a paradigm shift from the assessment of infant behavior to the dyadic assessment of interactions, as previously achieved in retrospective approaches using home movies²⁴. The aim is not to implement studies of social signal processing in routine clinical work but rather to decompose clinical intuitions and signs and validate the most relevant cues of these clinical features. From clinical work, back to clinics, social signal processing is a rigorous step to help clinicians better identify and assess early targets of interventions.

Given the exploratory nature of both our approach and method, our results should be interpreted with caution taking into account strengths (prospective follow-up, automatized multimodal social signal processing, and ecological standardized assessment) and limitations. These limitations include (1) the overall sample size knowing that WS is a rare disease; (2) the high rate of missing data during video recording due to the ecological conditions of the infant-mother interaction (mothers interposing between the camera and the infant); the final sample size of WS+ (N = 10) that limited the power of machine learning methods. (social signal processing)

We conclude that the method proposed here combining multimodal automatized assessment of social signal processing during early interaction with infants at risk for NDD is a promising tool to decipher clinical features that remain difficult to identify and assess. In the context of WS, we showed that such a method we proposed to label ‘behavioral and interaction imaging’ was able to significantly predict the development of ASD or ID at 4 years of age in high-risk children who had WS and were assessed at 9 months of age.

About KSRA

The Kavian Scientific Research Association (KSRA) is a non-profit research organization to provide research / educational services in December 2013. The members of the community had formed a virtual group on the Viber social network. The core of the Kavian Scientific Association was formed with these members as founders. These individuals, led by Professor Siavosh Kaviani, decided to launch a scientific / research association with an emphasis on education.

KSRA research association, as a non-profit research firm, is committed to providing research services in the field of knowledge. The main beneficiaries of this association are public or private knowledge-based companies, students, researchers, researchers, professors, universities, and industrial and semi-industrial centers around the world.

Our main services Based on Education for all Spectrum people in the world. We want to make an integration between researches and educations. We believe education is the main right of Human beings. So our services should be concentrated on inclusive education.

The KSRA team partners with local under-served communities around the world to improve the access to and quality of knowledge based on education, amplify and augment learning programs where they exist, and create new opportunities for e-learning where traditional education systems are lacking or non-existent.

FULL Paper PDF file: