Sepsis is a life-threatening immune response caused by infection, and its early detection is critical due to its high mortality. In this study, a Transformer encoder–based early prediction model for sepsis was developed using the 2019 PhysioNet Challenge dataset, which contains multivariate time-series data from approximately 40,000 patients. The model was trained to perform a classification task using hourly labeled data according to the presence or absence of sepsis occurring within six hours prior to onset. The proposed model achieved an area under the receiver operating characteristic curve of 0.924 and an area under the precision–recall curve of 0.182.