JMIR Publications

ABSTRACT

Background:

Children with autism spectrum disorder (ASD) frequently experience heightened anxiety, sensory sensitivity, and cognitive challenges during oral health routines and dental visits. These difficulties often result in poor oral hygiene and reliance on sedation or general anesthesia. Conventional oral health education is poorly adapted to the cognitive, sensory, and emotional needs of these children, which creates a need for accessible and contextually appropriate digital interventions.

Objective:

This study aimed to design, develop, and formatively evaluate a bilingual, web-based gamified prototype that supports oral health routines and dental visit preparation for children with ASD within the Malaysian public dental clinic context.

Methods:

A Design Science Research methodology was adopted, comprising problem identification and requirement derivation followed by artifact design, development, and formative evaluation. Requirements were triangulated from a literature review, clinical practice guidelines, and a purposive survey of 7 dental practitioners, and were prioritized using predefined consensus thresholds. The prototype was built in React 18 with MediaPipe for real-time facial and motion interaction and a transparent, rule-based adaptive difficulty engine. A bilingual (English and Bahasa Malaysia) and sensory-friendly interface was incorporated, together with a parent and clinician dashboard and a cloud backend with row-level data isolation. Formative evaluation combined expert content validation, a synthetic-data technical validation suite, and a heuristic usability evaluation by 5 dental practitioners. No children or real user data were involved.

Results:

Triangulation produced a consolidated framework of 20 design requirements, of which 11 met the core consensus threshold. The survey confirmed near-universal sensory triggers (the sound of the dental drill or suction was reported by 7/7, 100%, of respondents), behavioral resistance (5/7, 71%), and the importance of parental involvement (mean 4.86 out of 5) and bilingual support (6/7, 86%). Expert content validation initially rated only 14 of 32 items (44%) as fully valid and identified gaps in reward systems, user control, error recovery, and help features (Nielsen heuristics 3, 9, and 10). These findings were used to refine the artifact. The synthetic-data suite confirmed that the rule-based adaptive engine promoted, maintained, and demoted difficulty as designed across 1000 stress-test sessions. In the heuristic evaluation of the completed system, practitioners rated overall usability at a mean of 4.15 out of 5 with 84% agreement, with technical robustness rated highest (mean 4.28) and the emotion recognition game rated lowest (mean 3.90).

Conclusions:

The artifact demonstrated technical feasibility and usability adequacy within the limits of a pre-clinical, proof-of-concept evaluation, supporting the study hypothesis. The work provides a validated foundation and a documented, reusable technical framework for a future ethically approved clinical evaluation with children.