| abstract |
The invention relates to the technical field of liver cancer risk prediction, in particular to a method for establishing a liver cancer risk prediction model for hepatitis B cirrhosis, which comprises the following steps: s1, constructing a hepatitis B cirrhosis and liver cancer related factor table; s2, crawling historical clinical data based on the relevant factor table, acquiring the incidence relation between each relevant factor and liver cancer, and dividing each relevant factor into a main factor, a secondary factor and a secondary factor; s3, configuring corresponding weight and score evaluation standard for each major factor, minor factor and minor factor; s4, training and constructing the Bi-LSTM + Attention model based on the major factors, the minor factors and the corresponding weight and score evaluation criteria. The prediction model obtained by the invention can better predict the occurrence probability of the hepatitis B liver cirrhosis liver cancer, and the accuracy rate is about 91.76%. |