KidsAbility and University of Waterloo Break New Ground in AI for Pediatric Speech Therapy

KidsAbility and the University of Waterloo are collaborating to develop AI tools tailored for pediatric speech-language pathologists (SLPs). Traditional AI scribes often miss the nuances of children’s speech and play-based therapy, leading to heavy documentation burdens for clinicians. To address this, Waterloo researcher Tia Tuinstra, supervised by Dr. Bryan Tripp, worked with KidsAbility clinicians to design a custom AI pipeline. The system features SLP-specific language models, speaker-differentiated transcription, and phoneme-level analysis to track articulation accuracy. Early testing has shown improved note accuracy, better performance in structured sessions, and potential for detailed speech analysis. Future development will focus on refining prompts, enhancing transcription, expanding applications like referrals and progress reports, and integrating the tool with health record systems. The ultimate goal is to reduce paperwork and free clinicians to spend more time helping children and families thrive.