부산대학교 도서관

Introduction: Coronavirus vaccines are effectively to protect against severe COVID-19 and reduce the length of stay in intensive care unit (ICU) during the pandemic.(1,2) Awareness of outcome differences between vaccinated and unvaccinated patients is helpful for effective treatment planning and resource allocation. This study examined the effect of vaccination on length of stay (LOS) at Hospital Israelita Albert Einstein (HIAE). Findings were then used as input for development of a machine learning model to predict LOS in critically ill patients. Objective: We evaluated the effect of vaccination on the length of stay at Hospital Israelita Albert Einstein and applied it as input for a machine learning model developed to predict lenght of stay in critically ill patients. Method: This study comprised 1,361 patients admitted to the HIAE ICU in 2021. Patients who had received at least a dose of the COVID-19 vaccine (n=873) were defined as vaccinated. Patients who had not been vaccinated were defined as unvaccinated (n=488). The Kruskal-Wallis test (p<0.05) was used to investigate significant LOS differences between vaccinated and unvaccinated patients. Vaccination status was used as a variable in a model for LOS prediction at HIAE. The coefficient of determination (R²-Score) and the root-mean-square error (RMSE) were measured to assess model performance after introduction of this new variable. This study was approved by the local Research Ethics Committee. Results: The median length of stay differed significantly between unvaccinated and vaccinated patients (7 and 5 days, respectively; p<0.05, Kruskal-Wallis test) (Figure 1). Vaccination status was also used as a new variable in a regression model constructed to predict LOS in the critically ill department of HIAE. The primary objective of this model is to predict patient length of stay based on diagnosis at admission, demographic, and anthropometric data, near real time vital signs, laboratory tests, risk scores, prior care, medications, comorbidities, and outcomes (discharge or death). Model performance before and after introduction of this new variable was measured and compared (Table 1). Conclusion: Vaccinated patients with COVID-19 hospitalized at Hospital Israelita Albert Einstein had a shorter length of stay than unvaccinated patients. Vaccination status was not a significant variable for prediction of length of stay in this cohort of patients using this model. More research into how to integrate machine learning algorithms into the healthcare routine is needed. [ABSTRACT FROM AUTHOR]