Construction of a prediction model of survival rate of pancreatic cancer patients based on clinical and imaging data

Objective: To explore the independent prognostic factors of pancreatic cancer patients and establish a survival prediction model. Methods: The data of 141 patients with pancreatic cancer diagnosed in the First Affiliated Hospital of Guangxi Medical University from January 2020 to December 2021 were analyzed retrospectively. The clinical data of 44 patients with pancreatic cancer diagnosed in Guangxi Medical University Cancer Hospital from June 2022 to December 2023 were collected as external validation. Log-rank test and Cox proportional hazards model were used to conduct single factor and multiple factor analysis to obtain independent prognostic factors of pancreatic cancer. A nomogram prediction model was constructed based on the results of multiple factor analysis. The consistency index (C-index), receiver operating characteristic (ROC) curve, calibrate curve, and decision curve analysis (DCA) were calculated to validate and evaluate the prediction model. Results: The median survival time of pancreatic cancer patients was 7 months (95% CI: 5.5-8.5), the half-year survival rate was 55.2%, and the one-year survival rate was 29.5%. Cox regression analysis model of single factor and multiple factor showed that the age of onset, tumor stage, treatment mode, carbohydrate antigen 125 (CA 125), carbohydrate antigen 19-9 (CA 19-9) values of patients with pancreatic cancer were independent influencing factors of longterm survival time (P＜0.05). Conclusion: The prognosis of pancreatic cancer patients is poor. The age of onset, tumor stage, treatment mode, CA125, and CA19-9 of ancreatic cancer are related to the survival prognosis. The construction of a nomogram for predicting half-year and one-year survival rates has good predictive ability, calibration and clinical practicability.