Abstract: In view of the shortcomings of too many stroke precursory diagnostic indicators and the fact that the diagnosis process completely relies on the doctor’s clinical experience, a new method of measurement diagnosis is proposed. Using Aprori algorithm, 16 diagnostic indicators that are highly correlated with stroke are mined, and their weights are determined based on the importance and support of these indicators. On this basis, using the extracted indicators, a risk prediction model of cerebral stroke based on fuzzy theory is constructed, and computer prediction of cerebral stroke is realized. Using the model to predict 100 outpatients, the sensitivity and specificity are 94.7% and 88.7%, respectively. Compared with the BP neural network discriminant model, the model established in this paper can more objectively evaluate the risk level of stroke in the examined population. The simulation results show that the model can objectively evaluate the stroke risk level of the examined population and provide a new method of measurement diagnosis for the early detection of stroke. At the same time, the diagnosis recommendations given by the model can provide scientific prevention and treatment measures for the examined population.

Key words: stroke, fuzzy theory, membership degree

摘要： 针对传统脑中风先兆诊断指标多，且诊断过程完全依赖医生临床经验的弊端，提出一种新的计量诊断方法。利用关联规则Apriori算法，挖掘与脑中风高度相关的16项诊断指标，并依据这些指标的重要性和支持度，确定其权重。在此基础上，利用提取的指标，构建了基于模糊理论的脑中风风险预报模型。将该模型应用于100例门诊患者的脑中风预报，其灵敏度和特异度分别为94.7%和88.7%。与BP神经网络判别结果的比较进一步表明，建立的模型能更客观地评价受检人群的脑中风风险等级，为脑中风的早期发现提供了一种新的计量诊断方法。同时，借助模型给出的诊断建议，可为受检人群提供科学的防治措施。

关键词: 脑中风, 模糊理论, 隶属度