Research on the Design of Color Emotional Preferences in Products for Children with Autism Spectrum Disorder Based on Physiological Measurements and Random Forests

This study explores whether professionals engaged in Autism Spectrum Disorder (ASD) rehabilitation work develop similar color preferences through prolonged interactions with ASD children. The aim is to utilize these preferences for conducting color emotion research with ASD children, thereby mitigating challenges and limitations posed by ASD children in experiments. Physiological measurements, including Electrodermal Activity (EDA) and Electrocardiogram (ECG), were obtained from ASD rehabilitation teachers during specific color stimuli. The study aimed to reveal whether there are emotional tendencies similar to ASD children in color emotions. Industrial design undergraduate students were invited to participate in the evaluation, utilizing their physiological signals and a random forest model to establish a dual-layer correlation structure of "physiological signals - wearable color distance perception" for emotional image evaluation. The model was compared with GBDT, Adaboost, Xgboost, and RF algorithm models. Experimental results indicate (1) ASD rehabilitation teachers exhibit color preferences similar to ASD children; (2) the dual-modal (EDA + ECG) product color emotion image evaluation model based on random forests achieves a prediction accuracy of 85.35%; (3) wearable products (bracelets) based on the yellow + green combination provide a deeper sense of affinity.