Data-Driven Digital PECS System for Supporting Communication Development in Non-Verbal Children with Autism Spectrum Disorder

Abstract

Non-verbal children with Autism Spectrum Disorder (ASD) have obstacles to the expression of needs, emotions, and intentions. Picture Exchange Communication System (PECS) works with functional communication based on visual symbols, although most of the existing implementations are either paper-based or partially digitized. This constrains individualization and does not offer much longitudinal evidence regarding communication in the long run. This results in a situation where clinical progress assessment is typically based on manual documentation and subjective inspection instead of regular and objective data-driven pointers. We create EZPECS, a multi-purpose platform that unites parents and specialists, as well as organizations. Children compose sentences using ordered picture cards on personalized boards and read them using text-to-speech in Kazakh/Russian/English. The system keeps a record of sentence history (card sequences, board context, time stamps, language), the session activity (time spent in boards), and usage and coverage on the categories level. Based on such logs, we create an interpretable analytics framework, which has three layers: descriptive analytics to summarize engagement and vocabulary use; diagnostic analytics to surface patterns, e.g., drop-off in engagement and category dominance; and prescriptive analytics to convert indicators into rule-based recommendations (e.g., expanding underused categories and aligning sessions with high-engagement windows). The framework is demonstrated for an iPad deployment at the rehabilitation center in Astana. The thesis provides a useful data model and a logging strategy of digital PECS, role-oriented dashboards, and a basis of future subgroup-specific personalization, which relies on the characteristics of diagnoses and therapeutic objectives.