Study On Disease Complications Mining Method

With the application of information technology in the traditional medical industry, the storage method of medical clinical data has changed a lot, and a large amount of medical data is collected in real time or semi real time, forming the electronic medical record system. The electronic medical records system has integrated large amounts of patient information, medical history, medication use, inspection and diagnosis data, and shows the vast value of application and research in the field of high medical decision support and biological genetic research etc. Among them, exploring the disease complications from the electronic medical record in work has become an important way to find disease complications. Complication is one or several diseases developed with another one disease at the same time, and it always has high complexity and uncertainty. So the study of seeking associations between diseases and the complications of diseases is of significance.In this paper, a new method to find the complications of the disease is proposed, and a risk prediction system of the disease complications is established, aiming to prevent patient from disease complications. In order to get more and exacter association rules of diseases from clinical data, this paper proposes the concept of disease concurrent weight, and calculates disease concurrent weights with BP neural network model. Then the disease concurrent weights are combined with the weighted association rules mining algorithm to mine diseases association rules and successfully obtain diseases association rules. Based on these association rules, this paper constructs a risk prediction system for disease complications, which can predict the risk of complications for doctors and patients. Finally, this paper selects some complications data from clinical diagnostic data as the test data, verifies the satisfactory accuracy of the prediction results of the risk prediction system and proves the availability of the method proposed by this paper.