特发性突聋风险预测中关联规则的应用

目的 对特发性突聋的风险预测进行数据挖掘，并形成关联规则。方法 收集517例特发性突聋患者的临床资料，包括19项特征属性，分别为性别、年龄、季节、高血压、糖尿病、心脏病、高胆固醇血症、动脉粥样硬化、长期抽烟、酗酒、精神紧张、失眠、体质弱、长期卧床、感染、先天性畸形、创伤、肿瘤和自身免疫性疾病。将源数据库进行数据清洗后，映射为挖掘数据库；设置最小支持度为0.1，最小置信度为0.9，进行关联规则分析。结果 共形成106个强关联规则，这些强关联规则中蕴含特发性突聋与19项特征属性之间的关联关系。结论 本方法有利于将抽象的数理统计理论转变为实用的关联规则来指导疾病预防控制实践。

Application of association rules to risk prediction of sudden deafness

Objective To apply data mining to risk prediction of sudden deafness, and form the association rules.Methods The clinical data of 517 patients with sudden deafness was collected, including the characteristics of 19 attributes: sex, age, season, hypertension, diabetes, heart disease, hypercholesterolemia, atherosclerosis, long-term smoking, alcoholism, mental tension, insomnia, weakness, bedridden, infection, congenital malformation, trauma, tumour and autoimmune diseases. The source database were cleaned, then mapped for mining database. Minimum support to 0.1 and minimum confidence level to 0.9 were set for analysis of association rules. Results One hundred and six strong association rules were formed, and the rules contained the relation between the incidence of sudden deafness and the characteristics of 19 attributes. Conclusion This method is conducive to make the abstract theory of mathematical statistics into useful association rules to guide the practice of disease prevention and control.