Authors : Krishnapriya Ramanathan, Bhawini Vijayvergia, Piyush Kumar Rai, Mohd Aadam Bin Najeeb, Narayan Kamath
DOI : 10.5455/im.9874127
Volume : 7
Issue : 10
Year : 2025
Page No : 46-50
Background: Healthcare-associated infections (HAIs) pose a major global health challenge, increasing morbidity, mortality, and costs, especially in resource-limited settings like India. Traditional infection prevention and control (IPC) monitoring via manual audits is often intermittent and inefficient, prompting the need for innovative solutions. This study assesses the SmartIPC system, an artificial intelligence (AI)-based platform for continuous IPC monitoring and optimization in a Western Indian tertiary care hospital. Methods: Conducted from January 2024 to December 2024, this pre-post interventional study included 320 healthcare workers (HCWs) and 28,500 patients across high-risk units (intensive care unit, surgical wards, emergency department). The SmartIPC system integrated data from electronic health records (EHRs), wearable sensors, and IoT environmental monitors, utilizing convolutional neural networks (CNNs) and random forest models to classify compliance and predict HAI risks. Primary outcomes were IPC compliance rates (hand hygiene, PPE usage, surface disinfection) and HAI incidence, with secondary outcomes including system usability (System Usability Scale [SUS]) and HCW satisfaction. Results: The intervention period showed a 32% improvement in IPC compliance (hand hygiene: 65% to 86%, p < 0.001; PPE usage: 72% to 90%, p < 0.001; surface disinfection: 68% to 88%, p < 0.001) and a 15% reduction in HAI incidence (from 8.2 to 7.0 per 1,000 patient-days, p = 0.002). The system achieved an SUS score of 82 and 78% HCW satisfaction, though 15% raised privacy concerns. Post-hoc power analysis confirmed high statistical power (≥0.95). Conclusion: The Smart IPC system effectively enhances IPC practices and reduces HAIs via AI-driven monitoring. Its scalability and HCW acceptance indicate a promising model for improving patient safety in high-burden settings, with future research needed to address privacy and cost-effectiveness.