An Interlocutor-Modulated Attentional LSTM for Differentiating between Subgroups of Autism Spectrum Disorder

Abstract

Recalling and discussing personal emotional experiences is one of the key procedures in assessing complex affect processing of individuals with Autism Spectrum Disorder (ASD). This procedure is a standard subpart of a diagnostic interview to assess ASD-the Autism Diagnostic Observation Schedule (ADOS). Previous work has demonstrated that the behavior features computed from this procedure in ADOS possess discriminative information between the three distinct ASD subgroups: Autistic Disorder (AD), High Functioning Autism (HFA), and Asperger Syndrome (AS). In this work, we propose an interlocutor-modulated attentional long short term memory network (IM-aLSTM) that models the ASD individual's acoustic features with a novel interlocutor-modulated attention mechanism. Our IM-aLSTM achieves ASD subgroup categorization accuracy of 66.5%, which is a 14% absolute improvement over baseline method on the same database. Our analyses further indicate that the attention weights are concentrated more on interaction segments where the ASD individual is being asked to recall and discuss his/her own negative emotional experiences.

PRAUTOCAL corpus: a corpus for the study of Down syndrome prosodic aspects

“…Other corpora of aphasic speech are designed for automatic speech intelligibility assessment (Le et al, 2016), and contain interactions with a tablet application designed for therapeutic purposes. As far as the autism spectrum disorder is concerned, speech corpora based on recording the sessions of the Autism Diagnostic Observation Schedule (ADOS) have been collected (Lahiri et al, 2020;Li et al, 2019;Lin et al, 2018), an instrument for the diagnosis and assessment of autism. There are also corpora recorded with the aim of identifying and assessing the severity of the speech disorder in patients with Parkinson's disease (Hauptman et al, 2019;Khan et al, 2020).…”

Region-Wise Brain Response Classification of ASD Children Using EEG and BiLSTM RNN

“…Interocular-modulated attentional LSTM network distinguished the subgroups of ASD at an accuracy of 66.5%. 40 Identification of ASD using an eye-tracking technique with specific LSTM network at an accuracy of 83%. 41 Our present study consists of 3 main modules: Preprocessing of EEG signal.…”