A Risk Prediction Model for Hypothermia After Laparoscopic Gastrointestinal Tumor Surgery

Construction and Validation of a Risk Prediction Model for Hypothermia After Laparoscopic Gastrointestinal Tumor Surgery Based on Machine Learning

The incidence of postoperative hypothermia in patients with laparoscopic gastrointestinal tumors is high. Hypothermia increases the risk of postoperative complications and medical costs. Early warning can effectively reduce the incidence of postoperative hypothermia in patients. Multivariate prediction models help identify high-risk patients and reversible factors. At present, there are few reports on the risk factors and prediction models of postoperative hypothermia in patients with laparoscopic gastrointestinal tumors. Therefore, this study aims to clarify the risk factors of postoperative hypothermia in patients with laparoscopic gastrointestinal tumors. Four machine learning algorithms, traditional Logistic regression analysis, decision tree, random forest and naive Bayes, were used to establish risk prediction models. According to the TRIPOD statement, C-index, Hosmer-Lemeshow ( H-L ) test and decision curve analysis ( DCA ) were used to evaluate the prediction and fitting effects of the models in all aspects, and the optimal model was selected and verified. Provide reference for subsequent research.

Inclusion Criteria:

• The types of surgery included laparoscopic radical gastrectomy for gastric cancer, duodenal cancer, colon cancer, rectal cancer and small bowel cancer. Age ≥ 18 years ; the anesthesia method was general anesthesia ; elective surgery.

Exclusion Criteria:

• Patients with abnormal body temperature such as fever and hypothermia before operation ; patients who need low temperature to protect organs during operation ; patients who were converted to laparotomy during the operation ; a clear diagnosis of sweat gland dysfunction ; patients with external auditory canal active disease ; reject participants.