Abstract: Objective To develop a prediction model of prolonged hospital stay (PLOS) in patients undergoing laparoscopic radical surgery for colon cancer. Method A total of 369 patients were randomly divided into a modeling group and a validation group at a ratio of 7:3. The modeling group was used to analyze risk factors of PLOS and establish a risk prediction model. The model performance was evaluated using ROC curve, followed by external validation in the validation group. Results The prediction model for PLOS in patients undergoing laparoscopic radical resection for colon cancer was as follows: Logit (P/1-P)=-0.155 + 0.321 × age +0.067 × BMI + 0.092 × preoperative intestinal obstruction + 0.128 × surgery time + 0.185 × postoperative hospital acquired pneumonia + -0.201 × perioperative use of nonsteroidal anti-inflammatory drugs. The predictive performance of the model constructed from the modeling group demonstrated a ROC-AUC of 0.871. Validation set analysis yielded a ROC-AUC of 0.863. DeLong’s test comparing ROC-AUC values between modeling and validation group showed no statistically significant difference (P=0.779). Conclusion The PLOS prediction model constructed in this study demonstrates high predictive efficiency for patients undergoing laparoscopic radical surgery for colon cancer.

Key words: laparoscopy, radical surgery for colon cancer, length of hospital stay, prediction model