Screening of colorectal cancer risk factors based on Lasso regression and construction of nomogram prediction model

Abstract

Abstract Purpose Colorectal cancer has become the number one cancer of the digestive system and a serious risk to human health. This study explores the risk factors of colorectal cancer and provides a scientific basis for developing primary prevention measures for colorectal cancer. Methods Case and control groups were matched according to age (± 2) and gender (1∶1) factors, and risk factors for colorectal cancer were screened according to lasso regression and logistic regression analysis, and nomogram models were established, and subject working characteristic curves (ROC curves), calibration curves and decision curves were drawn for validation. Results A total of 175 cases in the case group and 175 cases in the control group were included. Lasso regression screened 23 significant variables, and logistic regression analysis showed that age, intestinal adenoma, eating meat, lamb, smoking, alcohol consumption and frequency of alcohol consumption were independent risk factors for colorectal cancer, and eating fruits, vegetables, chicken and exercise intensity were protective factors for colorectal cancer (P value < 0.05).The ROC curve analyzed the predictive value of the nomogram model with an AUC of 0.945 (95% CI: 92.2%-96.8%), with a sensitivity of 88.571 and specificity of 92.000. calibration curves and decision curves showed fair agreement and benefit of the nomogram model. Conclusion Colorectal cancer occurs as a result of multiple factors, and constructing a prediction model that includes relevant factors can effectively predict the risk of development and achieve primary prevention.