Objective To explore the influencing factors of moderate and severe cancer fatigue after colorectal cancer surgery and build a prediction model to provide guidance for early clinical prevention and intervention.

Methods A total of 320 patients with colorectal cancer were selected as the study objects, and divided into the moderate to severe group (n = 156) and non-moderate to severe group (n = 164) according to whether moderate to severe cancer-related fatigue occurred. Univariate analysis was used to analyze the factors affecting the fatigue caused by moderate to severe cancer in patients after colorectal cancer surgery. Two different prediction models, logistic regression and decision tree, were constructed with statistically significant variables (P ＜ 0.05), and the advantages and disadvantages of the two models were compared.

Results The results of logistic regression model analysis showed that the BMI, sleep quality, avoidance coping style and social support level were the influencing factors of moderate to severe cancer fatigue after colorectal cancer surgery. The analysis of decision tree model showed that the level of mental resilience was the main factor of moderate to severe cancer-related fatigue, followed by the level of social support, sleep quality and avoidance coping style. The AUC of logistic regression model and decision tree model were 0.977 and 0.965, sensitivity 89.0% and 85.4%, specificity 94.2% and 96.2%, positive predictive value 95.6% and 95.5%, negative predictive value 90.0% and 87.4%, respectively. The prediction accuracy was 91.3% and 90.6% respectively. There was no statistical significance in the AUC values between two models (Z = 1.11, P ＞ 0.05).

Conclusions Both logistic regression model and decision tree model have certain application value in predicting the risk of cancer-induced fatigue. It is suggested to combine the two models and guide clinical practice better.