Conceptualization of the latent structure of autism: further evidence and discussion of dimensional and hybrid models

Autism spectrum disorder (ASD) might be conceptualized as an essentially dimensional, categorical, or hybrid model. Yet, current empirical studies are inconclusive and the latent structure of ASD has explicitly been examined only in a few studies. The aim of our study was to identify and discuss the latent model structure of behavioral symptoms related to ASD and to address the question of whether categories and/or dimensions best represent ASD symptoms. We included data of 2920 participants (1-72 years of age), evaluated with the Autism Diagnostic Observation Schedule (Modules 1-4). We applied latent class analysis, confirmatory factor analysis, and factor mixture modeling and evaluated the model fit by a combination of criteria. Based on the model selection criteria, the model fits, the interpretability as well as the clinical utility we conclude that the hybrid model serves best for conceptualization and assessment of ASD symptoms. It is both grounded in empirical evidence and in clinical usefulness, is in line with the current classification system (DSM-5) and has the potential of being more specific than the dimensional approach (decreasing false positive diagnoses).

Automated ASD detection using hybrid deep lightweight features extracted from EEG signals

BACKGROUND

Autism spectrum disorder is a common group of conditions affecting about one in 54 children. Electroencephalogram (EEG) signals from children with autism have a common morphological pattern which makes them distinguishable from normal EEG. We have used this type of signal to design and implement an automated autism detection model.

MATERIALS AND METHOD

We propose a hybrid lightweight deep feature extractor to obtain high classification performance. The system was designed and tested with a big EEG dataset that contained signals from autism patients and normal controls. (i) A new signal to image conversion model is presented in this paper. In this work, features are extracted from EEG signal using one-dimensional local binary pattern (1D_LBP) and the generated features are utilized as input of the short time Fourier transform (STFT) to generate spectrogram images. (ii) The deep features of the generated spectrogram images are extracted using a combination of pre-trained MobileNetV2, ShuffleNet, and SqueezeNet models. This method is named hybrid deep lightweight feature generator. (iii) A two-layered ReliefF algorithm is used for feature ranking and feature selection. (iv) The most discriminative features are fed to various shallow classifiers, developed using a 10-fold cross-validation strategy for automated autism detection.

RESULTS

A support vector machine (SVM) classifier reached 96.44% accuracy based on features from the proposed model.

CONCLUSIONS

The results strongly indicate that the proposed hybrid deep lightweight feature extractor is suitable for autism detection using EEG signals. The model is ready to serve as part of an adjunct tool that aids neurologists during autism diagnosis in medical centers.

Autism Classification Using Topological Features and Deep Learning: A Cautionary Tale

The identification of autistic individuals using resting state functional connectivity networks can provide an objective diagnostic method for autism spectrum disorder (ASD). The present state-of-the-art machine learning model using deep learning has a classification accuracy of 70.2% on the ABIDE (Autism Brain Imaging Data Exchange) data set. In this paper, we explore the utility of topological features in the classification of ASD versus typically developing control subjects. These topological features have been shown to provide a complementary source of discriminative information in applications such as 2D object classification and social network analysis. We evaluate the performance of three different representations of topological features - persistence diagrams, persistence images, and persistence landscapes - for autism classification using neural networks, support vector machines and random forests. We also propose a hybrid approach of augmenting topological features with functional correlations, which typically outperforms the models that use functional correlations alone. With this approach, even with a simple 3-layer neural network, we are able to achieve a classification accuracy of 69.2% on the ABIDE data set. However, our experiments also show that the improvement due to topological features is not always statistically significant. Therefore, we offer a cautionary tale to the practitioners regarding the limited discriminative power of topological features derived from fMRI data for the classification of autism.