Application of AI-facilitated handwashing monitoring system on food handlers

Food handlers are a major route of foodborne disease transmission, yet handwashing in real food-service settings is often rushed and inconsistently performed. In a two-visit field study, 46 university canteen food handlers performed habitual (ad-hoc) handwashing during one visit and a standardized procedure guided by an AI-based computer-vision system during the other, with hand swabs collected before and after washing and enumerated on Plate Count Agar (PCA), Violet Red Bile Glucose Agar (VRBGA), and Baird-Parker Agar (BPA). AI-facilitated handwashing achieved significantly greater reductions than ad-hoc washing specifically for VRBGA-detected Enterobacteriaceae ( p < 0.05), revealing an indicator-specific effect on transient, hygiene-relevant microorganisms. An ex vivo porcine-skin model and 16S rRNA gene sequencing converged to explain this pattern: handwashing preferentially removes transient microorganisms while exposing resident skin-associated bacteria, demonstrating that the choice of microbial indicator critically shapes how hand-hygiene efficacy is judged. Food handlers found the system highly acceptable, valuing it as a real-time reminder that reinforces correct technique and duration. Collectively, this work provides the first field-based microbiological evidence that AI-facilitated monitoring efficiently supports and improves hand-hygiene compliance, while reframing how such interventions should be evaluated in food-service settings.