基于iPLS的血清胆固醇、甘油三酯近红外定量分析  

Determination of Cholesterol and Triglyceride in Serum Based on Interval Partial Least Squares by Near-Infrared Spectrum

作　　者：黄富荣[1] 李仕萍[1] 余健辉[1] 郑仕富[2] 陈哲[1] 

机构地区：[1]光电信息与传感技术广东普通高校重点实验室(暨南大学),广州市黄埔大道西601号510632 [2]暨南大学第一附属医院临床检验中心,广州市黄埔大道西601号510632

出　　处：《光谱实验室》2011年第6期2774-2778,共5页Chinese Journal of Spectroscopy Laboratory

基　　金：国家青年基金(50903038);广东省科技计划项目(2007B030501008);广东省自然科学基金(32208005)

摘　　要：为了建立血清样品胆固醇、甘油三酯近红外分析最优模型,利用近红外透射光谱技术结合间隔偏最小二乘法(iPLS)建立预测模型。结果表明,胆固醇最优建模波段是1700—1798nm,最优预测模型的相关系数Rp、预测均方差RMSEP分别为0.984、0.198mmol/L;甘油三酯最优建模波段是1654-1746nm,最优预测模型的Rp、RMSEP分别为0.967、0.157mmol/L。采用iPLS建立血清胆固醇、甘油三酯定量分析模型,不仅可以提高模型的预测精度,而且模型更加简洁、数据运算量也更少,优选出的特征谱区还可为设计小型专用近红外分析仪器提供依据。The prediction model was established by near infrared transmission spectrum combined with partial least squares interval（iPLS） to establish the near infrared analytical optimized model for cholesterol and triglyceride in serum.The optimal band for prediction the cholesterol was 1700—1798nm,and the prediction coefficient（Rp）and the root mean square error of prediction（RMSEP） of the optimized prediction model were 0.984 and 0.198mmol/L,respectively,and the optimal band for prediction the triglyceride was 1654—1746nm,and the prediction coefficient（Rp） of 0.967 with the root mean square error of prediction（RMSEP）of 0.157mmol/L.The iPLS model for establishment of analysis model of cholesterol and triglyceride in serum could not only improve precision,but also simplify the model.Optimized wave band selection will provide base for designing minitype special near infrared content analyzer.

关 键 词：近红外光谱 胆固醇 甘油三酯 间隔偏最小二乘法 波段优选 

分 类 号：O657.33[理学—分析化学]