With the popularity of medical information systems and the improvement of national health care system, it needs to deal with massive information and provide comprehensive, accurate diagnosis decision support system. Traditional data analysis tools and software is automatic data modeling and analysis. The entire process is difficult to understand and modify and waste the experts'knowledge. Besides, there are different task in the analysis process, which need medical workers to instruct them. The model and algorithm in the platform is finite relative to complex kinds of disease. Therefore, the system should provide interface for users to assist to build the platform.According to the above, MedicalSAS provides a visual tool for medical modelling and analyzing. In the workflow mechanism, it modified and expanded the process model based on activity network. It defined some terms such as logical task node, actions and states and used IPO structure to describe the operator. It constructs the platform by modularizing universal data-mining algorithms, structure of the model library, the KBM and its maintaining algorithms and encapsulates grid services which include some functions such as parameters setting, service executing and rules verifying. Meanwhile, the XML and PMML technical are applied to describe and restore the model.The system testing shows that, with the workflow mechanism, the medical subjective-oriented analysis platform is good for medical experts to provide a diagnosis and disease prediction. The platform has a good expansibility and flexibility. With the user's experience and knowledge to guide the subjective analysis complementally, the analysis process become interactive and understandable. |