Assisting Hearing Impaired Children by Personalized Gamification of Auditory Verbal Therapy

Children with hearing impairments often need to adapt to their hearing aids or cochlear implants through various techniques, with Auditory Verbal Therapy (AVT) being a prominent long-term method. AVT, led by auditory and speech clinical experts, involves four key stages: awareness, discrimination, identification, and comprehension, as outlined in the auditory skills checklist. To enhance the effectiveness of AVT and maintain the children's engagement, this study integrates personalized gamification into the therapy process. Focusing primarily on the comprehension stage, which aligns with daily human interactions, the study utilizes patient profiles and clinical suggestions to generate quiz-like activities. Prompt engineering and Large Language Model (LLM) evaluation were conducted to ensure gamification efficacy. To develop the most effective prompts to generate assistive activities for AVT, prompt engineering techniques were utilized. Additionally, the study introduced three evaluation metrics: Tri-Aspect Benchmark (TAB), Token-Based Cost Effectiveness Metric (TBCEM), and Fernando's Cost-Performance Index (FCPI); to assess the performance and cost-effectiveness of the corresponding LLMs, ensuring producing content is therapeutically valuable and cost-efficient. GPT -4o Mini emerged as the optimal model, balancing therapeutic value and cost-efficiency for generating AVT content.