Application And Research Of Rough Set Theory On Diabetes Mellitus Diagnostic Analysis

Diabetes can cause Diabetic retinopathy, kidney diseases, diabetic neuropathy, cardiovascular disease and other complications, which are serious threats to human health and life.Currently, differences in the level of regional health care, in the skill of doctors, in the diagnosis of decision-making of the same doctor under different conditions, Will inevitably result in that there is no guarantee of the high level of medical expertise and good medical equipment with clinical stability, which could have been better made use of for large quantities of confirmed cases. In order to better carry out diagnosis and treatment of diabetes and its complications, an exploratory research is made in this passage in the application of the Rough Set Theory on the diagnostic analysis of diabetes, which has a positive significance on the promotion of human health career by assisting the diagnostic process of clinical doctors, raising the level of diagnosis, and diagnosing patients of diabetes as early and more as possible, in terms of confirmed cases.The main work and contribution in this passage are:Firstly, providing an effective data source, in the light of collection of confirmed diabetic cases and relevant pres-processing, for the application of the Rough Set Theory on the diagnostic analysis of diabetes.Secondly, based on rough set methods and in accordance with diagnosed mode-of-thinking of medical experts, proposing the method that the diagnostic rules for diabetic cases are extracted by rough inclusion, rough exclusion and rough contrast, the accuracy under which is 97.6% shown by laboratory data.Thirdly, according to the clinical practice including the false-positive and false-negative phenomena, different diagnostic requirements by different doctors, the homogeneity and heterogeneity of diabetic complications and their possibility of synchronization, the Error-tolerating Degree is proposed to match the rules, Reducing missed diagnosis and misdiagnosis in a certain sense.Fourthly, based on the extracting and analyzing rules above, and in terms of the need of clinical diagnosis, the computer-aided diagnosis system is developed, which has demonstrated by experiments the feasibility and effectiveness of the method proposed above, the testing result of which indicates that it has successfully helped clinical doctors diagnose patients with diabetic complications.