In recent years, with the development of ultrasound digital level, storage technology, computer technology and other correlation technology, many hospitals begin to build ultrasound diagnosis digital management system in order to assist disease management, the thyroid disease is the common disease in clinic, and the ultrasound is the important technology of screening, diagnosis and curative effect evaluation of thyroid disease, and it can obtain rich information, and the data are of bigger mining value and are of higher value in intelligent diagnosis of thyroid ultrasound, analysis of epidemiology and re-search of pathogeny and pathogenesis, and the statistical analysis is vital to the thyroid ultrasound diagnosis digital manage-ment system, the available data mining methods include correlation rule, setpair analysis, rough set theory and cluster analy-sis, the available statistical methods include Logistic regression analysis, factor analysis, principal component analysis, canonical correlation analysis, structural equation model, linear mixed model and maximum likelihood discriminant. At pre-sent, the medical statistics of our country is still at a lower level, the data mining use level remains far from adequate, and the health workers should make an effort to master more specified statistical theory and method and improve their own math application level.%近年来，随着超声数字化水平、存储技术、计算机技术等关联技术的发展，许多医院开始着手构建超声诊断数字化管理系统，以辅助疾病管理。甲状腺疾病是临床常见病，超声是甲状腺疾病筛查、诊断、疗效评估等重要技术，获取信息十分丰富，这些数据具有较大挖掘价值，在甲状腺疾病超声智能诊断、流行病学分析、病因与发病机制等研究中具有较高的价值。统计分析在甲状腺超声诊断数字化管理系统中发挥关键作用，可供选择的数据挖掘方法包括关联规则、集对分析、粗糙集理论、聚类分析等，可供选择的统计学方法包括 Logistic回归分析、因子分析与主成分分析、典型相关分析、结构方程模型、线性混合模型、最大似然判别，目前我国医学统计学仍处于较低水平，数据挖掘利用水平远远不足，卫生工作者应努力掌握更加专业的统计学理论、方法，提高自身统计应用水平。
展开▼
