Abstract: OBJECTIVE To construct multiple prediction models for adverse prognosis of the patients with Enterococcus bloodstream infection based on machine learning and evaluate its predictive efficiency. METHODS The clinical data of 128 patients with Enterococcus bloodstream infection who were treated in Jiangning Hospital Affiliated to Nanjing Medical University from Jan. 1, 2021 to Dec. 31, 2024 were retrospectively analyzed. The significant variables associated with incidence of the infection were screened out by Lasso regression and multivariate logistic regression and were brought into the machine learning model. The prediction models were constructed by respectively adopting 7 machine learning methods including logistic regression, decision trees, random forests, extreme gradient boosting, lightweight gradient boosting machines, support vector machines and artificial neural networks. The precision, accuracy, sensitivity and F1 score were observed and compared among the 7 models so as to evaluate the predictive efficiencies of the models. RESULTS The accurate rates of logistic regression, decision trees, random forests, extreme gradient boosting, lightweight gradient boosting machines, support vector machines and artificial neural networks were respectively 83.33, 84.44, 87.78, 86.67, 82.22, 86.67 and 86.67 in the test set; the precise rates were 88.24, 78.72, 85.71, 83.72, 77.78, 83.72 and 83.72, respectively; the F1 scores were 0.800, 0.841, 0.867, 0.857, 0.814, 0.857 and 0.857, respectively; the AUCs were 0.922, 0.922, 0.952, 0.933, 0.878, 0.916 and 0.942, respectively. The random forest model showed that the hypoproteinemia was the most influential factor. CONCLUSIONS The models that can predict the adverse prognosis of the patients with Enterococcus bloodstream infection are successfully constructed, among which the random forest model shows the optimal predictive efficiency, and it can serve as an effective tool for early prediction, prevention and treatment of adverse prognosis of such group of patients during clinical nursing practice.