Large Language Model-Based Simplification of Digital Therapeutics Explanations for Insomnia and Nicotine Dependence:...

Area of Science:

• Digital Health
• Health Informatics
• Medical Education

Background:

• Digital therapeutics (DTx) are software interventions for health conditions.
• Limited public awareness and complex educational materials hinder DTx adoption.
• The impact of large language models (LLMs) on patient understanding of DTx materials is not well-established.

Purpose of the Study:

• To evaluate if LLM-simplified DTx materials improve perceived understanding and readability compared to original manufacturer documents.
• To assess user perceptions of clarity and comprehensibility of simplified versus original DTx explanations.

Main Methods:

• A GPT-4o-based simplification tool was developed with structured readability instructions.
• Original DTx materials for insomnia and nicotine dependence were simplified using the tool.
• Two online randomized experiments (n=1000) compared LLM-simplified materials with original versions, assessing perceived understanding and subjective evaluations.

Main Results:

• LLM-simplified materials significantly improved perceived understanding in both insomnia and nicotine dependence experiments.
• Participants rated simplified explanations as easier, clearer, and more comprehensible than original versions.
• Effect sizes for subjective evaluations ranged from small to moderate.

Conclusions:

• LLM-based simplification enhances perceived understanding and readability of DTx materials for lay audiences after single exposure.
• This approach offers a scalable strategy to improve health information accessibility.
• Further research with objective measures is needed to confirm comprehension benefits.