부산대학교 도서관

In order to make a best treatment plan for different patients of lung cancers, the prognosis of disease's development and its influence factors should be evaluated and analyzed accurately. In the traditional clinical pathway, we can only consider the linear classification of those factors or divide them into different stages. This paper proposed an effective machine learning method based on the combination of Cox regressive model and BP-GA neural network to predict the patient's expectation of survival rate, and so as to find the best treatment regime for each different patient. After the above intelligent analysis, the treatment decision procedures were presented for improving the current clinical pathway of lung cancer, and had been designed into the decision support system based on Hadoop system and Spring cloud framework for distributed applications of doctor's workbench system or mobile terminals from hospitals.