Voluntary ChatGPT Query Sharing With Nurse Feedback and Serum Phosphorus in Peritoneal Dialysis Patients

INTRODUCTION: This study examined whether voluntary ChatGPT query sharing with nurse feedback was associated with serum phosphorus in patients receiving peritoneal dialysis (PD). METHODS: In this prospective longitudinal study, eighty-eight patients completed a 3-month program (T1-T3) and 2-month follow-up. Twenty shared ChatGPT query screenshots with nurses via LINE and received feedback. Linear mixed-effects models examined serum phosphorus and the ratio relative to the 12-month pre-program mean (PR-P12M). RESULTS: Query sharing was associated with lower serum phosphorus across 5 months (p = 0.025), without a significant group × time interaction. Sidak-adjusted month-specific comparisons showed lower serum phosphorus at T2 (mean difference = 0.687 mg/dL; 95% CI, 0.199-1.174; p = 0.006). PR-P12M was similarly associated with query sharing (p = 0.008), without a significant group × time interaction. CONCLUSION: Voluntary ChatGPT query sharing with nurse feedback was associated with lower serum phosphorus, suggesting a potential role for AI-assisted dietary information seeking within nurse-guided phosphorus education in PD care.