Abstract: Objective To establish and verify the risk prediction model of diabetic ketoacidosis, to provide reference for the early identification, prevention and control of the disease. Methods A total of 943 patients with type 2 diabetes hospitalized in Dali People's Hospital from January 2019 to December 2020 were selected by convenience sampling method. Logistic regression analysis was used to screen out independent risk factors to establish a risk prediction model. A total of 393 patients with type 2 diabetes hospitalized in the First Affiliated Hospital of Dali University from March 2020 to December 2020 were selected to verify the model. Results Four factors of acute infection（OR=8.210, stress events（OR=129.267）, random venous blood glucose（OR=1.986）, and HbA1c（OR=1.421）were included to construct a risk prediction model. The area under the ROC curve of the modeling group was 0.994; the maximum value of Youden index 0.929; the sensitivity 0.954; the specificity 0.975, and Hosmer-Lemeshow test P=0.975. The test results of the validation group indicated that the area under the ROC curve was 0.901 and the maximum value of Youden index, the sensitivity and the specificity was 0.754, 0.889, and 0.865 respectively. Conclusion Acute infection, stress event, random venous blood glucose, and glycosylatedhemoglobin are independent risk factors for type 2 diabetic ketoacid, and the prediction model constructed in this study shows good prediction efficiency.

Key words: Type 2 diabetes mellitus, ketoacidosis, risk factor, prediction model