Study On Diagnosis And Decision Model For Amblyopia Based On Amblyopic Inspection Data

Objective:According to clinical and visual scientific researches, amblyopia is a neurodevelopmental disorder in the visual system that is associated with disrupted binocular vision during early childhood. With further studies of the mechanism of information processing and cortical plasticity in the visual system, the clinical diagnosis of amblyopia being questioned due to its empirical and subjective judgment. There are more and more researches reveal that the binocular perception plasticity model for restoration and reconstruction of amblyopia has significant effects.Research purposes:(1) Amblyopia diagnosis clustering model is used to provide a more comprehensive and objective information for the diagnosis of clinical, including clinical and training model amblyopia defect-related data.(2) Through the study of different visual perception training options, we used training data set to build a decision tree classifier for amblyopia treatments based on binocular perception plasticity biological model. Based on the above background and the providing accurate scientific references for the subsequent clinical diagnosis and treatment, in this paper, clustering and decision tree classification algorithm are innovatively used in amblyopia defect inspection data, which achieved better results.The main process are:(1) Data preprocess with background knowledge for the collected data. It provided a guarantee for subsequent data mining and the precision of the results due to the clean, tidy, efficient data.(2) Unsupervised learning clustering algorithm is used to support the diagnosis of amblyopia to precipitate the feature attributes associated with the type of amblyopia, including traditional clinical experience and new inspection from new model.(3) Through the study of different visual perception training options, we used training data set to build a decision tree classifier for amblyopia treatments. Methods:At present, the diagnostic data leads to inconsistency since it contains the examination of clinical ophthalmologists and the inspection results come from the new model. It became new areas of diagnosis and treatment of amblyopia for how to combine traditional experience and the data from new inspection.Results:In the diagnosis of amblyopia cluster model, the optimum number of categories matched with the clusters of known clinical amblyopia type through the test of statistic R2, which could efficiently and accurately cluster the corresponding type. In the clustering results of selected attributes compared with derived attributes, the derived attribute clustering accuracy rate(91.45%) was higher than the selected attributes(86.32%) In the decision tree model of amblyopia treatment, the accuracy of the training data set and test data set were 88% and 91% respectively, and the law of the decision were consistent with the actual situation. Conclusions:The results show that the clustering model can effectively improve the integration and the usage of information, and can efficiently and accurately judge different types of amblyopia. Under the premise of keeping clinical examination data, newly added model checking data can also be fully digging, at the same time of auxiliary clinical disease diagnosis, reduce the probability of misdiagnosis, enabling amblyopia clinical examination in the diagnosis of a more rigorous and objective standard.By correcting learning to achieve intelligent and automatic recognition, reducing the requirements of Ophthalmology professional background for the operator, improving efficiency and optimization of the training.Based on data characteristics and data mining approach, discover the laws of knowledge, promotion of the development of amblyopia visual perception deficits mechanism, not only to select the appropriate treatment for improving amblyopia patients with reasonable, scientific and effective basis; At the same time, the development of personalized medicine program that depends on network and data mining technology, which has a strong role in reference and promotion.