A deep learning latent variable model to identify children with autism through motor abnormalities

Abstract

IntroductionAutism Spectrum Disorder (ASD) is a by-birth neurodevelopmental disorder difficult to diagnose owing to the lack of clinical objective and quantitative measures. Classical diagnostic processes are time-consuming and require many specialists’ collaborative efforts to be properly accomplished. Most recent research has been conducted on automated ASD detection using advanced technologies. The proposed model automates ASD detection and provides a new quantitative method to assess ASD.MethodsThe theoretical framework of our study assumes that motor abnormalities can be a potential hallmark of ASD, and Machine Learning may represent the method of choice to analyse them. In this study, a variational autoencoder, a particular type of Artificial Neural Network, is used to improve ASD detection by analysing the latent distribution description of motion features detected by a tablet-based psychometric scale.ResultsThe proposed ASD detection model revealed that the motion features of children with autism consistently differ from those of children with typical development.DiscussionOur results suggested that it could be possible to identify potential motion hallmarks typical for autism and support clinicians in their diagnostic process. Potentially, these measures could be used as additional indicators of disorder or suspected diagnosis.

Enhancing early autism diagnosis through machine learning: Exploring raw motion data for classification

“…However, most of these studies require the analysis of a great amount of features to obtain a complete picture of the target behavior. In line with the suggestions of Zhao et al (2021) [ 45 ] and Milano et al (2023) [ 31 ], that highlighted the growing need to identify globally optimal features for the diagnosis of ASD, to reduce data processing, high consumption of computational resources and, most importantly, to avoid the inclusion of features that are not essential for diagnosis. Thus, the goal of researchers in this field is to identify relevant features for ASD classification and find the most ecological model to compute them.…”

“…Another important goal of our study was to investigate the influence of acceleration on the classification process between groups of ASD and TD, as previous research suggested that acceleration might be a significant factor in this classification [ 31 ]. So, after implementing and training the models, we tested them on simulated trajectories and the result confirmed that acceleration represents a core component of the prediction of autism, confirming the results from existing literature [ 30 , 31 ]. By the means of simulated trajectories, we observed that acceleration was positively correlated with autism diagnosis in both models.…”

Improved motor skills in autistic children after three weeks of neurologic music therapy via telehealth: a pilot study

“…Though movement difficulties are not an official part of the primary autism diagnostic criteria, researchers have increasingly recognized what Kanner and Lesser (1958) observed, that autistic persons also display difficulties with motor functioning ( Fournier et al, 2010 ; Bhat et al, 2011 ; Colombo-Dougovito and Block, 2019 ). In fact, technology that measures movement on a precise level can detect an autism diagnosis with extremely high reliability using movement differences alone ( Torres et al, 2013 ; Milano et al, 2023 ). It is estimated that up to 90% of autistic children may experience motor difficulties such that they can receive a co-occurring diagnosis of developmental coordination disorder ( Miller et al, 2021 ).…”

A Prediction Model of Autism Spectrum Diagnosis from Well-Baby Electronic Data Using Machine Learning

“…In the ASD literature, ML has been applied for various purposes, including differentiating subgroups [33] and, specifically, parsing the behavioral phenotype variance [34]. Deep learning methods have been applied for ASD detection based on motor abnormalities [35] and the prediction of behavioral intervention efficacy from patient data [36]. Previous electronic health record (EHR) data analysis by ML algorithms differentiated clusters of ASD patients with common co-morbidities (e.g., [37][38][39]).…”