Abstract: PUB028
Artificial Intelligence-Assisted Development of Patient Education Materials to Improve Glomerular Disease Knowledge: A Pilot Implementation in Lupus Nephritis
Session Information
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Hasan, Irtiza, University of Florida College of Medicine - Jacksonville, Jacksonville, Florida, United States
- Holub, Olena, University of Cincinnati, Cincinnati, Ohio, United States
- Allen, Rebecca Jane, Mount Saint Joseph University, Cincinnati, Ohio, United States
Background
Glomerular diseases (GN), including lupus nephritis (LN), require complex, long-term management and high levels of patient engagement. However, patients frequently experience significant cognitive burden at diagnosis, compounded by time-limited clinical encounters and limited access to clear, personalized educational resources. Traditional patient education materials are time-intensive to develop and difficult to individualize. Artificial intelligence (AI) offers a scalable approach to generate tailored, patient-centered educational content.
Methods
We designed a three-phase pilot study to evaluate AI-assisted development of patient-facing lupus nephritis (LN) educational materials. Phase 1 includes a structured literature review and identification of key educational domains. Phase 2 uses Meta-Llama-3.1-8B-Instruct, hosted locally through LM Studio, to generate content with structured prompts and controlled parameters for reproducibility. Outputs are screened for readability using Flesch-Kincaid, SMOG, Gunning Fog, and Flesch Reading Ease, with an eligibility target of approximately a sixth-grade level. Eligible outputs are then assessed with PEMAT for understandability and actionability and reviewed by nephrologists for clinical accuracy, safety, and appropriateness. Phase 3 includes pilot deployment on a patient-accessible digital platform, followed by evaluation of usability, engagement, and acceptability.
Results
AI-enabled workflows allow rapid generation of standardized, customizable educational modules adaptable to patient literacy levels and preferences. Local deployment supports data privacy and HIPAA-compliant integration while enabling reproducible and auditable content generation. Preliminary implementation demonstrates feasibility of scalable, clinician-validated educational material development.
Conclusion
AI-assisted patient education represents a promising strategy to address critical gaps in GN education. This approach enables timely, personalized, and scalable content delivery while reducing clinician burden. Future work will focus on expanding this framework into a comprehensive GN digital education platform incorporating patient engagement tools and clinical trial integration, with potential applicability across a broad range of kidney diseases.
Acknowledgment
This project is being conducted as a capstone initiative within the American Society of Nephrology Fostering Innovative Leaders in Nephrology and Dialysis (FIND) Program.