Presentation (CTI)

OBJECTIVES: China’s rare disease patients face limited patient advocacy group (PAG) support, low health literacy, and specialist access barriers. This study evaluates an AI chatbot in WeChat communities, assessing impacts on scalable support, information quality, medication adherence, and real-world insights for healthcare decisions.
METHODS: HEOR methods evaluated an AI chatbot deployed in CORD-run rare disease WeChat groups. The system architecture utilized a disease-specific LLM fine-tuned on approved clinical content, PAG resources, and validated patient education materials, with a human volunteer expert escalation protocol for complex cases. Performance metrics captured included daily question-and-answer volume, response accuracy, user reach, escalation rates, and downstream patient actions including specialist referrals and treatment adherence improvements. Contextual benchmarking was conducted against 2025 DLP physician survey data (n=4,500) on patient AI usage patterns.
RESULTS: The deployed system processed approximately 280 patient queries per day at no extra staffing cost, offering 24/7. Response accuracy was maintained through mandatory source citation and real-time expert escalation for complex or clinically ambiguous cases. Patient-reported outcomes included reduced emotional burden at diagnosis and recurrence, improved medication adherence through integrated dosing reminders and side-effect tracking, and accelerated specialist referral via AI-powered hospital and clinic navigation. The structured interaction data generated a real-world evidence dataset of patient-reported disease concerns, unmet needs, and treatment barriers — available months before conventional clinical data signals.
CONCLUSIONS: AI chatbot deployment within rare disease patient communities demonstrates scalable, cost-effective patient support with measurable outcomes across emotional, adherence, and access dimensions. The real-world evidence generated by patient interactions provides a novel data source for HEOR analysis, NRDL access argumentation, and patient-centered outcome measure development. These findings suggest that AI-enabled PAG platforms should be incorporated into rare disease value frameworks as both a care delivery innovation and an RWD generation mechanism with direct policy implications.