A machine learning approach to predict healthcare-associated infections at intensive care unit admission: findings from the SPIN-UTI project

Background Identifying patients at higher risk of healthcare-associated infections (HAIs) in intensive care unit (ICU) represents a major challenge for public health. Machine learning could improve patient risk stratification and lead to targeted infection prevention and control interventions. Aim To evaluate the performance of the Simplified Acute Physiology Score (SAPS) II for HAIs risk prediction in ICUs, using both traditional statistical and machine learning approaches. Methods We used data of 7827 patients from the “Italian Nosocomial Infections Surveillance in Intensive Care Units” project. The Support Vector Machines (SVM) algorithm was applied to classify patients according to sex, patient origin, non-surgical treatment for acute coronary disease, surgical intervention, SAPS II at admission, presence of invasive devices, trauma, impaired immunity, antibiotic therapy in 48 hours before ICU admission. Findings The performance of SAPS II for predicting the risk of HAIs provides a ROC (Receiver Operating Characteristics) curve with an AUC (Area Under the Curve) of 0.612 (p<0.001) and an accuracy of 56%. Considering SAPS II along with other characteristics at ICU admission, we found an accuracy of the SVM classifier of 88% and an AUC of 0.90 (p<0.001) for the test set. In line, the predictive ability was lower when considering the same SVM model but removing the SAPS II variable (accuracy= 78% and AUC= 0.66). Conclusions Our study suggested the SVM model as a tool to early predict patients at higher risk of HAI at ICU admission.