Clinical Application of Large Language Models for Intervention Plan Development in Speech-Language Pathology

This study investigates the speech and language intervention plan outputs generated by six different artificial intelligence (AI) tools powered by large language models (LLMs), currently available for clinical writing in the field of speech-language pathology. This study aims to evaluate the potential applications and limitations of these AI tools, as well as their ability to provide relevant and reliable information for developing intervention plans.

Using a mixed design including both quantitative and qualitative analyses, this study compared the performance outputs of the six AI tools across three fictional clinical cases, each involving different types of speech and language disorders in 5-year-old children. Two types of command prompts, each with three levels of input specificity, were used to generate AI outputs.

Results revealed that the intervention plans generated by these AI tools were rated between Needs Improvement and Meets Expectations in terms of clinical knowledge and competency. Detailed and structured command prompts than general prompts yielded outputs with higher ratings, while the specificity of case information did not consistently influence the outputs. Each AI tool demonstrated unique strengths and limitations in supporting the development of intervention plans.

The results of this study may serve as foundational data to provide insights into how clinicians, educators, and students in the field of speech-language pathology can appropriately and responsibly utilize existing AI resources when implementing these technologies into the development of intervention plans.