Cox's proportional hazards regression model for prognosis of hepatocellular carcinoma based on factor analysis with SPSS

Objective To explore the feasibility of Cox's proportional hazards regression model for the prognosis of hepatocellular carcinoma based on the factor analysis with SPSS. Methods One hundred and forty cases of hepatocellular carcinomas and 44 related prognosis factors were analyzed with SPSS13. 0 and Microsoft Office Excel 2007. The comparative analysis of traditional Cox's proportional hazards regression model and Cox's proportional hazards regression model based on the factor analysis was conducted to determine which one was more feasible. Results In traditional Cox's proportional hazards regression model,25 of 44 factors entered the multivariate analysis,of them 5 were reserved in the final equation. Remarkably,the results of the univariate analysis and the multivariate analysis for the factor X5 were contradictory,which implied the multicollinearity among the prognosis factors. Thirteen common factors were extracted by the factor analysis and then they all entered Cox's proportional hazards regression model,and of them 7 were reserved in the final equation. They were professional meaning and reasonable after being renamed. Conclusion It's feasible to solve the multicollinearity by Cox's proportional hazards regression model for the prognosis of hepatocellular carcinoma based on the factor analysis with SPSS.