多属性皮肤指标的中医体质模糊优化分类模型

摘要

This paper explores the invisible incidence relation between facial skin indexes and traditional Chinese med-icine (TCM) constitution to determine constitution types, which can provide new ways for skin diagnosis and treat-ment. Multi-attribute skin background sample values tested by China Cosmetics Research Center and the constitution classification results determined through constitution questionnaire are selected as model matching data. Firstly, princi-pal component analysis based on Spearman correlation is adopted to reduce the dimension of indexes. Then, fuzzy membership degrees of indexes to constitution are gotten by statistical experiment to realize intelligent classification of non-standardized data. On this basis, this paper proposes a fuzzy optimization combination weighting method fused by subjective and objective information to reduce classification error, and establishes TCM constitution fuzzy optimization classification model of multi-attribute skin indexes. Testing results of new samples show that classifica-tion accuracy of model reaches 80%. Thus the fuzzy classification model in this paper has good practicality and va-lidity, which can provide new ideas for further research on fuzzy multi-attribute classification and evaluation of oth-er non-standardized and personalized data.%探索面部皮肤指标与人体内在的中医体质类型间的隐形关联关系并确定体质类型，可为皮肤养护与诊疗提供新途径。选取中国化妆品研究中心实测的具有多属性特性的皮肤本底样本值及由中医体质问卷确定的体质分类结果作为建模匹配数据对，依据Spearman相关性进行主成分分析，实现了皮肤指标降维，而后依据模糊统计方法得到皮肤指标对体质的模糊隶属度，实现了非标准化数据的智能化划分。在此基础上，提出了一种融合主客观信息的模糊优化组合赋权法，用以降低分类误差，建立多属性皮肤指标的中医体质模糊优化分类模型。针对新样本的测试结果表明，模型分类正确率达到80%，从而证明了模糊分类模型具有较好的实用性及有效性，并可为其他非标准化与个性化数据的模糊多属性分类与评价问题的深入研究提供新思路。