Individuals with autism spectrum disorder (ASD) demonstrate impairments with pragmatic (social) language, including narrative skills and conversational abilities. We aimed to quantitatively characterize narrative performance in ASD using natural language processing techniques: sentiment and language abstraction analyses based on the Linguistic Category Model. Individuals with ASD and with typical development matched for age, gender, ethnicity, and verbal and nonverbal intelligence quotients produced language samples during two standardized tasks from the Autism Diagnostic Observation Schedule, Second Edition assessment: Telling a Story from a Book and Description of a Picture. Only the narratives produced during the Book Task differed between ASD and control groups in terms of emotional polarity and language abstraction. Participants with typical development used words with positive sentiment more often in comparison to individuals with ASD. In the case of words with negative sentiment, the differences were marginally significant (participants with typical development used words with negative sentiment more often). The Book Task narratives of individuals with ASD were also characterized by a lower level of language abstraction than narratives of peers with typical development. Linguistic abstraction was strongly positively correlated with a higher number of words with emotional polarity. Neither linguistic abstraction nor emotional polarity correlated with participants’ age or verbal and nonverbal IQ. The results support the promise of sentiment and language abstraction analyses as a useful tool for the quantitative, fully automated assessment of narrative abilities among individuals with ASD.

Detection of mental disorders from textual input is an emerging field for applied machine and deep learning methods. Here, we explore the limits of automated detection of autism spectrum disorder (ASD) and schizophrenia (SCZ). We compared the performance of: (1) dedicated diagnostic tools that involve collecting textual data, (2) automated methods applied to the data gathered by these tools, and (3) psychiatrists. Our article tests the effectiveness of several baseline approaches, such as bag of words and dictionary-based vectors, followed by a machine learning model. We employed two more refined Sentic text representations using affective features and concept-level analysis on texts. Further, we applied selected state-of-the-art deep learning methods for text representation and inference, as well as experimented with transfer and zero-shot learning. Finally, we also explored few-shot methods dedicated to low data size scenarios, which is a typical problem for the clinical setting. The best breed of automated methods outperformed human raters (psychiatrists). Cross-dataset approaches turned out to be useful (only from SCZ to ASD) despite different data types. The few-shot learning methods revealed promising results on the SCZ dataset. However, more effort is needed to explore the approaches to efficiently training models, given the very limited amounts of labeled clinical data. Psychiatry is one of the few medical fields in which the diagnosis of most disorders is based on the subjective assessment of a psychiatrist. Therefore, the introduction of objective tools supporting diagnostics seems to be pivotal. This paper is a step in this direction.

Background: Deficits in the ability to use language in social contexts, including storytelling skills, are observed across the autism spectrum. Development in machine‐learning approaches may contribute to clinical psychology and psychiatry, given its potential to support decisions concerning the diagnosis and treatment of psychiatric conditions and disorders.Aims: To evaluate the usefulness of deep neural networks for detecting autism spectrum disorder (ASD) from textual utterances, specifically from narrations produced by individuals with ASD.Methods & Procedures: We examined two text encoders: Embeddings from Language Models (ELMo) and Universal Sentence Encoder (USE), and three classification algorithms: XGBoost, support vector machines, and dense neural network layer. We aimed to classify 25 participants with ASD and 25 participants with typical development (TD) based on their narrations produced during the Autism Diagnostic Observation Schedule, Second Edition (ADOS‐2) picture book task. The results of computational approaches were compared with the results of standardized testing and classifications made by two psychiatrists (raters). The raters were asked to read utterances produced by a participant (without an examiner's statements and additional information) and assign a participant to one of the two groups: ASD or with typical development (TD).Outcomes & Results: The computer‐based models had higher sensitivity, specificity, positive predictive values and negative predictive values than the raters, and lower than the two standardized instruments: ADOS‐2 and Social Communication Questionnaire (SCQ).Conclusions & Implications: Our findings lay the groundwork for future studies involving deep neural network‐based text representation models as tools for augmenting the ASD diagnosis or screening. Both ELMo and USE text encoders provided promising specificities, sensitivities, positive predictive values and negative predictive values. Our results indicate the usefulness of page‐level embeddings for utterance representation in ADOS‐2 picture book task.WHAT THIS PAPER ADDS: What is already known on this subject: Deficits in the use of language in social contexts, and narrative ability in particular, are observed across the autism spectrum. Most research on narrative skills has applied hand‐coding methods. Hitherto, machine‐learning methods were used mostly for image recognition problems and data from screening questionnaires for ASD classification. Detection of mental and developmental disorders from textual input is an emerging field for machine and deep‐learning methods.What this paper adds to existing knowledge: This study explored the ability of several types of deep neural network‐based text representation models to detect ASD. Both ELMo and USE provided the most promising values of specificity, sensitivity, positive predictive values and negative predictive values.What are the potential or actual clinical implications of this work?: Competitive accuracy, repeatability, speed and ease of operation are all advantages of computerized methods. They allow for objective and quantitative assessment of narrative ability and complex language skills. Deep neural network‐based text representation models could in the future support clinicians and augment the decision‐making process related to ASD diagnosis, screening and intervention planning.

In this paper, we explore the approaches to the problem of cross-domain few-shot classification of sentiment aspects. By cross-domain few-shot, we mean a setting where the model is trained on large data in one domain (for example, hotel reviews) and is intended to perform on another (for example, restaurant reviews) with only a few labelled examples in the target domain. We start with pre-trained monolingual language models. Using the Polish language dataset AspectEmo, we compare model training using standard gradient-based learning to a zero-shot approach and two dedicated few-shot methods: ProtoNet and NNShot. We find both dedicated methods much superior to both gradient learning and zero-shot setup, with a small advantage held by NNShot. Overall, we find few-shot to be a compelling alternative, achieving a surprising amount of performance compared to gradient training on full-size data.