As a cancer with high incidence rate, Prostate cancer (PCa) endangers men’s health worldwide. In developing countries, medical staff are overloaded because of the lack of medical resources. Medical support system is a good technique to ease contradiction between the large number of patients and small number of doctors. In this paper, we have collected 1,933,535 patient information items from three hospitals, constructed a medical support system for PCa. It uses six relevant tumor markers as the input features and output a quantitative indicator EM value for staging and recommending treatment method. Classical machine learning techniques, data fusion and ensemble method are employed in the system to make the results more correct. In terms of staging PCa, it reaches an accuracy of 83$$\%$$%. Further research based on the system and collected data are carried out. It is found that the incidence of prostate cancer has been rising in the past five years and diet habit and genetic inheritance have a great impact on it.