Abstract:

Due to the atypical symptoms and lack of specificity, the detection rate of early gastric cancer is relatively low, leading to many patients already in advanced stage since first diagnosis. However, the prognosis of patients with advanced gastric cancer is generally poor and varies greatly, with the 5-year survival rate hovering between 20% and 60%. Therefore, it is necessary to develop more personalized treatment strategies to improve the survival rate of patients and reduce the difference in prognosis of different patients. The establishment of reliable models for gastric cancer is one of the important problems to be solved urgently in the diagnosis and treatment of gastric cancer. Cox regression analysis model can analyze the impact of multiple factors on the prognosis of patients; nomogram model can directly display the contribution of different prognostic factors to the prognosis of patients by calculating the scores of different prognostic factors; the artificial neural network model can be automatically learned, and the deep learning model can automatically extract features to establish an efficient prediction model, which are applicable to the prognosis prediction of gastric cancer with large-scale data. The above predictive models may be used to differentiate the risk of gastric cancer, which is expected to provide personalized reference for follow-up treatment for patients, and improve the survival of patients to some extent.

Key words: Gastric neoplasms, Prognosis, Prediction model