Document resource
Background and Importance Medication errors represent a critical global healthcare challenge, ranking third in mortality causes with over 200,000 annual deaths in the United States. Literature demonstrates 39% of adverse drug events result from preventable medication errors, significantly impacting patient safety and healthcare costs. Existing clinical decision support systems show limited effectiveness due to poor integration, complex interfaces, and inadequate real-time capabilities. Community hospitals face additional challenges including limited resources, insufficient pharmaceutical expertise, and lack of accessible technology solutions. Despite advances in artificial intelligence and digital health platforms, no comprehensive AI-based Clinical Decision Support System integrated with social messaging platforms has been developed for resource-limited healthcare settings, representing a critical unmet need.Aim and Objectives To develop and evaluate an AI-powered clinical decision support system using LINE Official Account that reduces medication errors and improves decision-making efficiency in community hospitals. Secondary objectives included measuring system accuracy, response times, and user satisfaction.Material and Methods Quasi-experimental one-group pre-post design conducted over six months. Forty healthcare professionals (7 physicians, 7 pharmacists, 26 nurses) recruited using stratified random sampling. AI-based CDSS featured ten modules: five core clinical modules (drug safety screening, drug interaction checking, dose calculation, pregnancy/lactation safety, cross-allergy prevention) and five supportive modules (medication reminders, pharmacy locator, adverse drug reaction reporting, health product complaints, disease surveillance). Primary outcomes included medication error rates per 1000 patient-days, response time, and accuracy rates.Results AI-based CDSS significantly reduced medication error incidents from 26.17±2.46 to 8.75±1.82 events per month per 1000 patient-days, representing 66.6% reduction (p<0.001). System accuracy achieved 96.9% overall. Response time improved from 8.0±2.8 minutes to 0.4±0.167 minutes (86.5% improvement, p<0.001). Core modules responded within 24 seconds, supportive modules within 1.2 minutes. This resulted in 6.3 hours daily time savings. User satisfaction reached 4.8/5.0, exceeding target of 4.21.Conclusion and Relevance This study presents first successful AI-based CDSS integrated with social messaging platform, demonstrating remarkable 66.6% medication error reduction. High accuracy rates and efficiency improvements validate artificial intelligence potential in resource-limited settings. Findings have significant implications for global digital health transformation, creating scalable worldwide implementation model addressing medication safety challenges while improving healthcare quality and accessibility.Conflict of Interest No conflict of interest