Construction Of A Recurrence Risk Of Machine Learning-Based Model After Radical Resection Of Pancreatic Cancer

Objective:Pancreatic cancer(PC)is a highly deadly malignant tumor with a very high mortality rate,known as the "king of cancer".Surgical resection is currently the only treatment that can be used to cure pancreatic cancer,but the risk of recurrence after radical resection of pancreatic cancer is still high.Therefore,studying the prognosis model after radical resection of pancreatic cancer to improve the treatment effect and reduce the risk of recurrence is important to improve the prognosis of pancreatic cancer patients.This paper aims to explore the application of the rando.Methods:Clinical data of 372 patients with pancreatic ductal adenocarcinoma who underwent radical resection at Zhujiang Hospital of Southern Medical University from January 2012 to December 2017 were collected and analyzed.After screening,a total of 282 patients met the inclusion and exclusion criteria.The study subjects were randomly divided into training set and test set,and multifactor Cox regression analysis and random forest algorithm(RF)were used to construct the prediction model,and the test set and area(AUC)were used to verify the identification ability and clinical practicability of the two models.Results:The variables finally included in the model were age,primary site of tumor,degree of tumor differentiation,CA19-9,T stage,N stage,TNM stage,and tumor size.The results show that the predictive efficacy of Cox nomogram model is better than that of machine learning algorithm model in the prognosis of pancreatic cancer in predicting the prognosis after surgery.Moreover,the Cox nomogram model predicted results matched well with the correction curve for predicting the postoperative prognosis of pancreatic cancer patients.Conclusion:The postoperative prediction model of pancreatic cancer based on Cox nomogram is better than machine learning method for 1 year after surgery,which can provide clinicians to improve the prognosis and survival status of patients with pancreatic cancer.It can also provide guidance for the development of an individualized postoperative surveillance program for patients with pancreatic ductal adenocarcinoma.