Application of artificial intelligence to predict the sepsis in neonates admitted in neonatal intensive care unit

Introduction: Sepsis is a major cause of morbidity and mortality in neonates admitted to neonatal intensive care units (NICUs) around the world. The importance of early discovery and appropriate management of these vulnerable patients is very important in improving the outcomes. This study includes a wide range of machine learning methods, such as support vector machines, Random forests, and other models, that have been used to construct predictive models for neonatal sepsis (NS). Methodology: The medical records of newborns diagnosed with bacterial sepsis between January 2017 and November 2021 were retrospectively analyzed. The WEKA v-3.8.6 tool was used for feature selection and developing prediction models. After substantial risk factors for NS were found, numerous machine learning models, including Naive Bayes, Random Forest, Bagging, Logistic Regression, and J48 models, were developed. We used these models to predict sepsis in the NICU based on their reliability. Result: The records of 388 sepsis patients were used to build the model using training and test data sets. For the prediction of culture-negative sepsis, using correlation attribute evaluation + Ranker feature selection method, Bagging (A = 98.4%; ROC = 0.984) performed well with 66% spilt. Random Forest and Bagging outperformed in predicting culture-positive sepsis. Conclusion: Healthcare professionals might potentially reduce the burden of sepsis-related problems by early detection and greatly enhance neonatal care outcomes by leveraging the power of artificial intelligence (AI). However, more study, validation, and clinical integration are required before AI's full potential in this vital field of neonatal medicine can be realized.