Purpose/Significance To construct an explainable prognostic prediction model for elderly patients with sepsis，and to sup-port clinical decision-making and improve patient outcomes. Method/Process Elderly patients(≥60 years)meeting the Sepsis-3.0 cri-teria are retrospectively analyzed from the eICU-CRD and MIMIC-IV databases. Thirteen core inflammatory indicators are extracted， and an explainable sepsis prognosis transformer(ESPT)model is constructed based on the Transformer architecture. The SHAP frame-work is employed for global feature importance ranking and visual interpretation of individual predictions. Result/Conclusion The pro-posed model outperforms traditional machine learning models such as random forest and XGBoost，demonstrating its potential as an ef-fective auxiliary tool for risk stratification and clinical decision-making in elderly patients with sepsis.