D2.537 - Advances in Drug Allergy & Penicillin Testing Phase 2 (ADAPT-2): an AI-powered drug allergy educational initiative spanning across 4 localities

Mislabelled drug allergy (DA) remains a global public health challenge. A prior randomized trial (ADAPT; NCT06399601) demonstrated that an intensive educational course improved DA knowledge and confidence among non-specialists. However, ADAPT was restricted to English-speaking participants and its generalizability remains unknown. To address this, a multinational implementation study expanding ADAPT (ADAPT-2) was performed.

Non-allergist physicians from Colombo (Sri Lanka), Guangzhou and Shenzhen (Mainland China), Hong Kong (Special Administrative Region of China), and Perth (Australia) completed a standardized DA educational course. In Mainland China, training was delivered via AI-assisted video localization (converted into Mandarin while preserving speaker’s voice with lip-synced adaptation). DA knowledge, confidence, and practice were assessed before and after completion. Subgroup analyses compared pre–post changes between Advanced Economies (AE: Australia, Hong Kong) and Emerging Economies (EE: Mainland China, Sri Lanka).

Of 181 participants, overall baseline knowledge (53.5±17.2%) and confidence (47.5±22.7%) scores were suboptimal. EE participants had lower knowledge level than AE (49.1±15.5% vs. 70.1±12.7%; p<0.001). Following ADAPT-2, both knowledge (72.5±16.0%, p<0.001) and confidence (71.3±17.5%, p<0.001) scores significantly improved across all groups. ADAPT-2 delivered by AI-assisted video localization was non-inferior to the English course in effectiveness (p>0.05) and achieved high participant satisfaction (98.9% as “somewhat clear” or better in clarity)

Deficits in DA knowledge persist widely among non-specialists, with marked disparities between AE and EE. ADAPT-2 bridged these gaps by universal improvements in both DA knowledge and confidence. AI-assisted training represents a scalable, equitable strategy for global implementation of standardized and evidence-based DA education.