Introduction: Sepsis is a severe medical concern, and timely and accurate risk prediction is crucial for improving patient safety and treatment outcomes. Traditional methods of sepsis risk assessment have their limitations. This study aims to explore the use of advanced machine learning techniques, particularly the Light Gradient Boosting Machine (LGBM), to enhance the accuracy of sepsis risk prediction and its potential application in the digital transformation of healthcare.

Aim: To enhance the accuracy of sepsis risk prediction by utilizing the LGBM model, and to explore its potential role in the digital transformation of healthcare.

Methods: This study employed the LGBM model to analyze a vast amount of clinical data to assess the risk of patients developing sepsis. Various risk factors were considered, and evaluation was performed using metrics including AUC, accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score.

Results: The LGBM model achieved a perfect AUC value of 0.99 during the training phase and an AUC of 0.89 during the testing phase. The model's accuracy was 83.04%, sensitivity was 85.16%, specificity was 81.58%, PPV was 75.98%, NPV was 88.93%, and the F1 score was 0.81. These results demonstrate the high accuracy and reliability of the LGBM model in predicting sepsis risk. 

Conclusion: This study validates the high performance of the LGBM machine learning model in sepsis risk assessment, showcasing the significant potential of artificial intelligence technology in enhancing healthcare quality and efficiency.