Transmissible diseases pose significant risks within healthcare facilities, which can lead to accelerated viral spread and exacerbate healthcare burdens, particularly for patients with severe conditions. During the COVID-19 pandemic, numerous cases of nosocomial infections were reported, underscoring the shortcomings of current in-hospital management strategies for controlling such outbreaks. This study aims to provide insights for developing improved in-hospital management strategies for future infectious disease outbreaks.
Comprehensive data on the infection status and clinical profiles of all patients and healthcare staff, gathered during the COVID-19 outbreak at a Shanghai brain hospital, were analyzed to evaluate transmission dynamics and identify risk factors. Machine learning techniques were utilized to recognize infection factors, while time series analyses—including cross-correlation and Granger causality tests—were performed to explore temporal relationships.
The infection rate within the hospital significantly exceeded that of the surrounding community, confirming a heightened risk of nosocomial transmission. Notably, peak infections among healthcare workers lagged behind those among patients, with time series analysis revealing a clear patient-to-worker transmission pathway within hospital wards. Furthermore, a machine learning prediction model trained on all patient data demonstrated high accuracy in predicting (Area under curve = 1.00 for training set and 0.85 for testing set) susceptible individuals and successfully identified risk factors such as age, doctor number in ward, red blood cell count, etc
The study elucidated a distinct patient-to-worker transmission pattern of respiratory infectious diseases in healthcare settings and identified critical susceptibility risk factors. These findings provide valuable strategic guidance for managing and mitigating future nosocomial outbreaks, ultimately enhancing infection control and preparedness in hospital environments.