Study On The Construction Of Prediction Model For Early Recurrence And Metastasis Of Colorectal Cancer Based On Machine Learning

Objectives1 To understand the current situation of early recurrence and metastasis of colorectal cancer after surgery,and analyze the influencing factors of early recurrence and metastasis.2 Based on the machine learning algorithm,a prediction model for early recurrence and metastasis of colorectal cancer after surgery is constructed,which provides clinical workers with a more scientific and accurate prediction tool for early recurrence and metastasis of colorectal cancer.MethodsThis study is a prospective cohort study design,and 498 postoperative patients with colorectal cancer who were treated in the Affiliated Hospital of Qingdao University were included as the research objects by using the convenience sampling method from June2021 to December 2021.The patients were enrolled 6 months after the operation,and were followed up to the 16 th month after the operation to determine the recurrence and metastasis of the patients.According to the five aspects of the health ecological model,personal traits,behavioral characteristics,interpersonal network,living and working conditions,and policy environment,we used general data questionnaire,Simplified Food Frequency Questionnaire,International Physical Activity Questionnaire,Hospital Anxiety and Depression Scale,Connor-Davidson Resilience Scale,and Perceived Social Support Scale to collected data during outpatient visits and hospitalizations.SPSS 26.0was used to statistically analyze the data,and the influencing factors of early recurrence and metastasis of colorectal cancer were screened by univariate analysis and multivariate Logistic regression analysis.Using software such as Python,the data set is randomly divided into 70% training set and 30% test set,the training set is used to train the model,and the test set is used to test the generalization ability of the model.Four machine learning algorithms,Logistic regression,support vector machine,XGBoost and LightGBM,were used to construct a prediction model for early recurrence and metastasis of colorectal cancer after surgery.Use accuracy,specificity,precision,recall,F1 value and area under the ROC curve(AUC)to evaluate the performance of the model and screen the optimal model.ResultsAmong the 498 patients,51 cases(10.24%)had early recurrence and metastasis,and447 cases(89.76%)had no early recurrence and metastasis.Univariate analysis and multivariate Logistic regression analysis showed that early postoperative recurrence and metastasis of colorectal cancer were affected by personal traits(T stage,degree of differentiation,histological type,number of positive lymph nodes,history of cancer in first-degree relatives),behavioral characteristics(psychological resilience level,intake of refined grains,whole grains,fish,shrimp,crab,and nuts)and interpersonal network(level of social support)(P<0.05).The prediction model based on the LightGBM algorithm showed the best prediction performance among the four models.The accuracy of this model was 0.891,the precision was 0.829,the recall was 0.860,the specificity was 0.839,the F1 value of the comprehensive performance evaluation index was 0.883,and the AUC was 0.901,all of which were better than the other three models.Conclusions1 The early recurrence and metastasis rate of colorectal cancer was 10.24%.Based on the analysis of the health ecological model,it was concluded that the early recurrence of colorectal cancer after surgery was affected by personal characteristics(T stage,degree of differentiation,histological type,number of positive lymph nodes,cancer history of first-degree relatives),behavioral characteristics(level of psychological resilience,refined Intake of cereals,whole grains,fish,shrimp,crab,and nuts)and interpersonal network(level of social support).2 The prediction model of early recurrence and metastasis after colorectal cancer surgery based on the LightGBM algorithm had the best prediction effect among the four prediction models,and can be applied to the prediction of early recurrence and metastasis after colorectal cancer surgery.