AUTHOR=Alcañiz Raya Mariano , Chicchi Giglioli Irene Alice , Marín-Morales Javier , Higuera-Trujillo Juan L. , Olmos Elena , Minissi Maria E. , Teruel Garcia Gonzalo , Sirera Marian , Abad Luis 

TITLE=Application of Supervised Machine Learning for Behavioral Biomarkers of Autism Spectrum Disorder Based on Electrodermal Activity and Virtual Reality

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 14 - 2020

YEAR=2020

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2020.00090

DOI=10.3389/fnhum.2020.00090

ISSN=1662-5161

ABSTRACT=Objective. Sensory processing is the ability to capture, elaborate and integrate information through the five senses and is impaired in over 90% of children with autism spectrum disorder (ASD). ASD population shows a hyper-hypo sensitiveness to sensory stimuli that can generate an alteration in the information processing, affecting cognitive and social responses to daily life situations. Structure, semi-structure interviews are generally used for ASD assessment and the evaluation depends on the subjectivity and expertise of the examiner that can lead to misleading outcomes. Recently, there is a growing need for more objective, reliable and valid diagnostic measures, such as biomarkers, to distinguish normal and non-normal functioning, to track the progression of illness reliably, thus helping to diagnose ASD. Implicit measures and ecological valid settings are showing high accuracy on predicting outcomes, and classifying correctly populations in separate categories.  
Methods. Two experiments investigated whether sensory processing can discriminate between ASD and typical development populations using electro dermal activity (EDA) in two multimodal virtual environments (VE): a forest VE and a city VE. In the first experiment, 24 children with an ASD diagnosis and 30 with a typical development participated in virtual experiences and changes in EDA have been recorded before and during the presentation of visual, auditory and olfactory stimuli. 
Results. The first exploratory results on EDA comparison models showed that, the integration of visual, auditory, and olfactory stimuli in the forest environment provided a higher accuracy (90.3%) on sensory processing than the specific stimuli. 
In the second experiment, 92 subjects experienced the forest VE and the result on 82 subjects showed that the stimuli integration achieved the 86.59% of accuracy. The final confirmatory test set (n=10) achieved a 90% of accuracy, recognizing the 100% of subjects with sensory dysfunction. Another relevant result concerns the visual stimuli condition in the first experiment that achieved an 84.6% of accuracy in recognizing ASD sensory dysfunctions. 
Conclusions. According to our studies results, implicit measures, such as EDA, and ecological valid settings can represent valid quantitative methods, along with traditional assessment measures, to classify ASD population, enhancing knowledge on the development of relevant specific treatments.