多模型共识偏最小二乘法用于近红外光谱定量分析

将多模型共识偏最小二乘法用于近红外光谱定量分析。利用随机抽取的训练子集建立一系列偏最小二乘模型,选取其中性能较好的部分模型作为成员模型,用这些成员模型来预测未知样品。将该方法用于一组生物样本的近红外光谱与样品中人血清白蛋白、γ-球蛋白以及葡萄糖含量之间的建模研究,并与单模型偏最小二乘法了进行比较。结果 PLS对独立测试集中三种组分进行50次重复预测的平均RMSEP分别为0.1066,0.0853和0.1338,RMSEP的标准偏差分别为0.0174,0.0144和0.0416;而本方法重复预测的平均RMSEP分别为0.0715,0.0750和0.0781,RMSEP的标准偏差分别为0.0033,0.2729×10-4和0.0025。     

Infrared spectroscopic analysis of human interstitial fluid in vitro and in vivo using FT-IR spectroscopy and pulsed quantum cascade lasers (QCL): Establishing a new approach to non invasive glucose measurement

Interstitial fluid, i.e. the liquid present in the outermost layer of living cells of the skin between the Stratum corneum and the Stratum spinosum, was analyzed by Fourier transform infrared spectroscopy and by infrared spectroscopy using pulsed quantum cascade infrared lasers with photoacoustic detection. IR spectra of simulated interstitial fluid samples and of real samples from volunteers in the 850-1800cm(-1) range revealed that the major components of interstitial fluid are albumin and glucose within the physiological range, with only traces of sodium lactate if at all. The IR absorbance of glucose in interstitial fluid in vivo was probed in healthy volunteers using a setup with quantum cascade lasers and photoacoustic detection previously described. A variation of blood glucose between approx. 80mg/dl and 250mg/dl in the volunteers was obtained using the standard oral glucose tolerance test (OGT). At two IR wavelengths, 1054cm(-1) and 1084cm(-1), a reasonable correlation between the photoacoustic signal from the skin and the blood glucose value as determined by conventional glucose test sticks using blood from the finger tip was obtained. The infrared photoacoustic glucose signal (PAGS) may serve as the key for a non-invasive glucose measurement, since the glucose content in interstitial fluid closely follows blood glucose in the time course and in the level (a delay of some minutes and a level of approx. 80-90% of the glucose level in blood). Interstitial fluid is present in skin layers at a depth of only 15-50μm and is thus within the reach of mid-IR energy in an absorbance measurement. A non-invasive glucose measurement for diabetes patients based on mid-infrared quantum cascade lasers and photoacoustic detection could replace the conventional measurement using enzymatic test stripes and a drop of blood from the finger tip, thus reducing pain and being a cost-efficient alternative for millions of diabetes patients.     

中红外光谱无创伤检测血糖新方法

糖尿病是一种常见的代谢内分泌疾病,是由于人体内缺乏胰岛素或其受体异常所致,以高血糖为主要特征,为一种世界范围内的流行疾病,近年来其发病呈显著上升趋势,目前全世界约有10％的成年人身患此病,在我国,糖尿病患者约有四千万人,血糖的测量对于临床诊断和糖尿病人血糖的控制很重要,常规的方法是静脉取血测量血糖,该方法会给病人带来一定的痛苦,同时增加了感染机会,     

基于导数光谱融合建模的红外光谱定量分析方法

设计了基于奇摄动技术的导数光谱估计器并提出基于不同阶次导数光谱空间的融合建模定量分析方法。方法充分利用导数光谱信息空间、区间最小二乘法和融合建模的优点,挖掘光谱深层次信息进行融合建模。分别利用麦汁浓度范围4.23~18.76° P (柏拉图度)的啤酒红外光谱公共数据集和配制的浓度为0.04%~5%范围的葡萄糖溶液实测光谱数据集进行定量分析方法的对比实验。实验结果表明,融合建模定量分析方法能获得最小的预测均方根误差(RMSEP),其值分别为0.121和0.087,能够准确地进行定量分析。与其它建模方法相比较,基于导数光谱的融合建模方法所建立的预测模型具有明显优越的性能。     

Improvement of the partial least squares model performance for oral glucose intake experiments by inside mean centering and inside multiplicative signal correction.

Near infrared (NIR) spectroscopy has become a promising technique for the in vivo monitoring of glucose. Several capillary-rich locations in the body, such as the tongue, forearm, and finger, have been used to collect the in vivo spectra of blood glucose. For such an in vivo determination of blood glucose, collected NIR spectra often show some dependence on the measurement conditions and human body features at the location on which a probe touches. If NIR spectra collected for different oral glucose intake experiments, in which the skin of different patients and the measurement conditions may be quite different, are directly used, partial least squares (PLS) models built by using them would often show a large prediction error because of the differences in the skin of patients and the measurement conditions. In the present study, the NIR spectra in the range of 1300-1900 nm were measured by conveniently touching an optical fiber probe on the forearm skin with a system that was developed for in vivo measurements in our previous work. The spectra were calibrated to resolve the problem derived from the difference of patient skin and the measurement conditions by two proposed methods, inside mean centering and inside multiplicative signal correction (MSC). These two methods are different from the normal mean centering and normal multiplicative signal correction (MSC) that are usually performed to spectra in the calibration set, while inside mean centering and inside MSC are performed to the spectra in every oral glucose intake experiment. With this procedure, spectral variations resulted from the measurement conditions, and human body features will be reduced significantly. More than 3000 NIR spectra were collected during 68 oral glucose intake experiments, and calibrated. The development of PLS calibration models using the spectra show that the prediction errors can be greatly reduced. This is a potential chemometric technique with simplicity, rapidity and efficiency in the pretreatment of NIR spectra collected during oral glucose intake experiments.     

Prediction of CO2 Solubility in Polymers by Radial Basis Function Artificial Neural Network Based on Chaotic Self-adaptive Particle Swarm Optimization and Fuzzy Clustering Method

To replace costly and time-consuming experimentation in laboratory, a novel solubility prediction model based on chaos theory, self-adaptive particle swarm optimization （PSO）, fuzzy c-means clustering method, and radial ba- sis function artificial neural network （RBF ANN） is proposed to predict CO2 solubility in polymers, hereafter called CSPSO-FC RBF ANN. The premature convergence problem is overcome by modifying the conventional PSO using chaos theory and self-adaptive inertia weight factor. Fuzzy c-means clustering method is used to tune the hidden centers and radial basis function spreads. The modified PSO algorithm is employed to optimize the RBF ANN connection weights. Then, the proposed CSPSO-FC RBF ANN is used to investigate solubility of CO2 in polystyrene （PS）, polypropylene （PP）, poly（butylene succinate） （PBS） and poly（butylene succinate-co-adipate） （PBSA）, respec- tively. Results indicate that CSPSO-FC RBF ANN is an effective method for gas solubility in polymers. In addition, compared with conventional RBF ANN and PSO ANN, CSPSO-FC RBF ANN shows better performance. The values of average relative deviation （ARD）, squared correlation coefficient （R2） and standard deviation （SD） are 0.1071, 0.9973 and 0.0108, respectively. Statistical data demonstrate that CSPSO-FC RBF ANN has excellent prediction capability and high-accuracy, and the correlation between prediction values and experimental data is good.     

Nondestructive determination of compound amoxicillin powder by NIR spectroscopy with the aid of chemometrics

Near-infrared (NIR) spectroscopy, in combination with chemometrics, enables nondestructive analysis of solid samples without time-consuming sample preparation methods. A new method for the nondestructive determination of compound amoxicillin powder drug via NIR spectroscopy combined with an improved neural network model based on principal component analysis (PCA) and radial basis function (RBF) neural networks is investigated. The PCA technique is applied to extraction relevant features from lots of spectra data in order to reduce the input variables of the RBF neural networks. Various optimum principal component analysis-radial basis function (PCA-RBF) network models based on conventional spectra and preprocessing spectra (standard normal variate (SNV) and multiplicative scatter correction (MSC)) have been established and compared. Principal component regression (PCR) and partial least squares (PLS) multivariate calibrations are also used, which are compared with PCA-RBF neural networks. Experiment results show that the proposed PCA-RBF method is more efficient than PCR and PLS multivariate calibrations. And the PCA-RBF approach with SNV preprocessing spectra is found to provide the best performance.     

Frequency self deconvolution in the quantitative analysis of near infrared spectra

In this paper a new model based on frequency self deconvolution (FSD) is proposed for the quantitative analysis of a near infrared (NIR) spectrum. The model couples FSD and partial least square regression (PLS). The grid search optimization method is used to select the optimal values of the full width at half height (FWHH) and the truncation point of the apodization function. The proposed FSD-PLS provides a significant improvement in the prediction ability of the PLS model. Furthermore, a modification of the new FSD-PLS method is introduced to enable the removal of the baseline variations from the NIR spectra. The proposed models were validated using absorbance spectra of mixtures composed from glucose, urea and triacetin in a phosphate buffer solution where the concentrations of the components are selected to be within their physiological range in blood. The whole experiments were carried out in a non-controlled environment to show that the model can suppress effectively most of the experimental variations. The results show that the standard error of prediction (SEP) decreases from 35.58 mg dL(-1) using 8 factors for the PLS model to 15.53 mg dL(-1) by using 12 factors for the modified FSD-PLS model. The proposed models are also shown to yield a slightly improved performance than a newly developed second derivative-PLS model without incurring the shortcoming associated with the derivative approach in not providing interpretable results and in degrading the SNR of the spectra at a faster rate.     

径向基函数网络在近红外人体无创伤血糖浓度检测基础研究中的应用

在近红外无创伤血糖浓度检测的基础研究中，对于多组分的混合物的分析，常因光谱与样品浓度之间呈现非线性响应，使得基于线性模型的校正方法失效。本文讨论了非线性校正方法径向基函数神经网络( RBFN )的有效性，并与线性校正方法中的主成分分析和偏最小二乘法作了对比研究。验证实验所用样品为①葡萄糖水溶液②包含牛血红蛋白和白蛋白的葡萄糖水溶液，结果表明：在①实验中PLS模型和RBFN预测标准偏差分别为8．2、8．9；在②实验中分别为15．6、8．8。可见在样品组分增多时，RBFN算法较线性PLS方法建立的模型预测能力强。     

Comparison of multivariate calibration models for glucose,urea, and lactate from near-infrared and Raman spectra

Partial least-squares (PLS) calibration models have been generated from a series of near-infrared (near-IR) and Raman spectra acquired separately from sixty different mixed solutions of glucose, lactate, and urea in aqueous phosphate buffer. Independent PLS models were prepared and compared for glucose, lactate, and urea. Near-IR and Raman spectral features differed substantially for these solutes, with Raman spectra enabling greater distinction with less spectral overlap than features in the near-IR spectra. Despite this, PLS models derived from near-IR spectra outperformed those from Raman spectra. Standard errors of prediction were 0.24, 0.11, and 0.14 mmol L⁻¹ for glucose, lactate, and urea, respectively, from near-IR spectra and 0.40, 0.42, and 0.36 mmol L⁻¹ for glucose, lactate, and urea, respectively, from Raman spectra. Differences between instrumental signal-to-noise ratios were responsible for the better performance of the near-IR models. The chemical basis of model selectivity was examined for each model by using a pure component selectivity analysis combined with analysis of the net analyte signal for each solute. This selectivity analysis showed that models based on either near-IR or Raman spectra had excellent selectivity for the targeted analyte. The net analyte signal analysis also revealed that analytical sensitivity was higher for the models generated from near-IR spectra. This is consistent with the lower standard errors of prediction.     

Determination of sugars in depilatory formulations: a green analytical method employing infrared detection and partial least squares regression

A green analytical method was developed for the analysis of sugar-based depilatories. Three independent partial least squares (PLS) regression models were built for the direct determination of glucose, fructose and maltose without any sample pretreatment based on their attenuated total reflectance - Fourier transform infrared (ATR-FTIR) spectra. The models showed adequate prediction capabilities with root-mean-square-errors of prediction ranging from 7.04 to 12.55 mg sugar g⁻¹ sample. As a reference procedure, gradient liquid chromatography with on-line infrared detection, employing background correction based on cubic smoothing splines, was used. The analysis revealed changes in the sugar concentration due to the formulation process as compared to information on the ingredients provided by the manufacturers. Although fructose, glucose and sucrose were declared to be used for the production of depilatories, in the final products only fructose, glucose and maltose were determined. This fact was attributed to pH and temperature conditions employed during the production process as well as to the use of glucose syrup instead of crystalline glucose. The present ATR-FTIR-PLS method enables an accurate, cheap and fast determination without solvent consumption or toxic waste generation and offers therefore a green screening alternative to methods employing chromatographic techniques.     

Multivariate calibration of glucose in blood by PLS using spectral and fourier-domain data

Infrared attenuated total reflection spectra of 133 whole EDTA blood samples, from patients of a general hospital population, in the range from 1500 to 750 cm^–1 were used for the calibration of glucose. Reference concentration values were provided by the enzymatic glucose dehydrogenase method. The partial-least squares (PLS) algorithm was used to solve the inverse regression problem. The prediction results from, calculations using spectral and Fourier-transformed data were compared, and in the latter case, the data reduction yielded no advantage. The spectral range optimization for calibration can be carried out more flexibly in the spectral domain which is more readily interpreted by the spectroscopist.     

Quantitative analysis of serum and serum ultrafiltrate by means of Raman spectroscopy

The fast and reliable determination of concentrations of blood, plasma or serum constituents is a major requirement in clinical chemistry. We explored Raman spectroscopy as a reagent-free tool for predicting the concentrations of different parameters in serum and serum ultrafiltrate. In an investigation using samples from 247 blood donors (which we believe to be the largest study on Raman spectroscopy of serum so far) the concentrations of glucose, triglycerides, urea, total protein, cholesterol, high density lipoprotein, low density lipoprotein and uric acid were determined with an accuracy within the clinically interesting range. After training a multivariate algorithm for data analysis, using 148 samples, concentrations were predicted blindly for the remaining 99 serum samples based solely on the Raman spectra. Relative errors of prediction around 12% were obtained. Moreover, to the best of our knowledge, differentiation between HDL and LDL cholesterol as well as the quantification of uric acid was for the first time successfully accomplished for serum-based Raman spectroscopy. Finally, we showed that ultrafiltration can efficiently reduce fluorescent light background to improve prediction accuracy such that the relative coefficient of variation was reduced for glucose and urea in ultrafiltrate by more than a factor of 2 when compared to serum.     

Current development in wearable glucose meters

Diabetes is one of the most disturbing chronic diseases in the world. The improvement of treatment efficiency brought by self-monitoring of blood glucose can relieve symptoms and reduce complications, which is considered as the gold standard of diabetes diagnosis and nursing. Compared to the traditional finger pricking measurement with painful and discontinuous processes, continuous blood glucose monitoring (CGM) presents superior advantages in wearable and continuous assessment of blood glucose levels. However, widely used implantable CGM systems at present require implantation operation and are highly invasive, so it is hard to be accepted by users. Except for the blood, available fluids in humans, such as interstitial fluid (ISF), sweat, tears and saliva, also contain glucose associated with blood sugar and can be extracted more easily. Therefore, these more accessible fluids are expected to realize minimized traumatic blood glucose monitoring. This review introduces the latest development of wearable minimally-/non-invasive CGM device, focusing on the types of blood substitute biological fluid and suitable monitoring approaches. We also analysis the merits and drawbacks of each method, and discuss the properties such as sensitivity, stability and convenience of each meter. Beyond highlighting recent key work in this field, we discuss the future development trend of wearable minimally-/non-invasive glucose meters.     

Comparison of performance of partial least squares regression,secured principal component regression,and modified secured principal component regression for determination of human serum albumin, γ-globulin and glucose in buffer solutions and in vivo blood glucose quantification by near-infrared spectroscopy

The performances of three multivariate analysis methods—partial least squares (PLS) regression, secured principal component regression (sPCR) and modified secured principal component regression (msPCR)—are compared and tested for the determination of human serum albumin (HSA), γ-globulin, and glucose in phosphate buffer solutions and blood glucose quantification by near-infrared (NIR) spectroscopy. Results from the application of PLS, sPCR and msPCR are presented, showing that the three methods can determine the concentrations of HSA, γ-globulin and glucose in phosphate buffer solutions almost equally well provided that the prediction samples contain the same spectral information as the calibration samples. On the other hand, when some potential spectral features appear in new measurements, sPCR and msPCR outperform PLS significantly. The reason for this is that such spectral features are not included during calibration, which leads to a degradation in PLS prediction performance, while sPCR and msPCR can improve their predictions for the concentrations of the analytes by removing the uncalibrated features from the original spectra. This point is demonstrated by successfully applying sPCR and msPCR to in vivo blood glucose measurements. This work therefore shows that sPCR and msPCR may provide possible alternatives to PLS in cases where some uncalibrated spectral features are present in measurements used for concentration prediction.     

Multivariate calibration for near-infrared spectroscopic assays of blood substrates in human plasma based on variable selection using PLS-regression vector choices

A multicomponent assay for the blood substrates of total protein, glucose, total cholesterol, triglycerides and urea in human EDTA-plasma by FT-IR spectroscopy is described based on near-infrared spectra of human plasma recorded in a 1 mm quartz transmission cell. Partial least-squares was applied for multivariate calibration taking into account absorbance or logarithmized single beam spectra. Further data reduction was applied using the pairwise selection of spectral variables suggested by the weights of the optimum full spectrum PLS-regression vector. The standard errors of prediction for protein, cholesterol, triglycerides, glucose and urea are calculated by cross-validation for the population of 124 plasma samples of different patients. These values are compared for full spectrum and reduced spectral variable set regression. Received: 16 March 1998 / Accepted: 29 May 1998     

Multivariate calibration for near-infrared spectroscopic assays of blood substrates in human plasma based on variable selection using PLS-regression vector choices

A multicomponent assay for the blood substrates of total protein, glucose, total cholesterol, triglycerides and urea in human EDTA-plasma by FT-IR spectroscopy is described based on near-infrared spectra of human plasma recorded in a 1 mm quartz transmission cell. Partial least-squares was applied for multivariate calibration taking into account absorbance or logarithmized single beam spectra. Further data reduction was applied using the pairwise selection of spectral variables suggested by the weights of the optimum full spectrum PLS-regression vector. The standard errors of prediction for protein, cholesterol, triglycerides, glucose and urea are calculated by cross-validation for the population of 124 plasma samples of different patients. These values are compared for full spectrum and reduced spectral variable set regression. Received: 16 March 1998 / Accepted: 29 May 1998     

Removal of interference signals due to water from in vivo near-infrared (NIR) spectra of blood glucose by region orthogonal signal correction (ROSC).

A novel chemometric method, region orthogonal signal correction (ROSC), is proposed and applied to pretreat near-infrared (NIR) spectra of blood glucose measured in vivo. Water is the most serious interference component in such kinds of noninvasive measurements, because it shows very high absorbance in the spectra. In the present study, the spectra of blood glucose in the range of 1212 - 1889 nm are used, in which the absorption of water around 1440 nm is very high. ROSC aims at removing the interference signal due to water from the spectra by selecting a set of spectra with a special region of 1404 - 1454 nm that mainly contain information about the variation of the interference component, water, and calculating the orthogonal components to the concentrations of glucose that will be removed. The difference between ROSC and orthogonal signal correction (OSC) is that ROSC uses a special region of spectra for the estimation of scores and loading weights of orthogonal components to pretreat the spectra in other regions, while OSC only uses one fixed region of spectra to calculate loadings, scores and weights of OSC components and removes the OSC components in the same region. A clear advantage of ROSC is that it is more interpretable than OSC, because one can select a spectral region to remove the variation of a special component such as water. Another chemometric method, moving window partial least squares (MWPLSR), is also used to select informative regions of glucose from the NIR spectra of blood glucose measured in vivo, leading to improved PLS models. Results of the application of ROSC demonstrate that ROSC-pretreated spectra including the whole spectral region of 1212 - 1889 nm or an informative region of 1600- 1730 nm selected by MWPLSR provide very good performance of the PLS models. Especially, the later region yields a model with RMSECV of 15.8911 mg/dL for four PLS components. ROSC is a potential chemometric technique in the pretreatment of various spectra.     

Investigation of the specificity of Raman spectroscopy in non-invasive blood glucose measurements

Although several in vivo blood glucose measurement studies have been performed by different research groups using near-infrared (NIR) absorption and Raman spectroscopic techniques, prospective prediction has proven to be a challenging problem. An important issue in this case is the demonstration of causality of glucose concentration to the spectral information, especially as the intrinsic glucose signal is smaller compared with that of the other analytes in the blood–tissue matrix. Furthermore, time-dependent physiological processes make the relation between glucose concentration and spectral data more complex. In this article, chance correlations in Raman spectroscopy-based calibration model for glucose measurements are investigated for both in vitro (physical tissue models) and in vivo (animal model and human subject) cases. Different spurious glucose concentration profiles are assigned to the Raman spectra acquired from physical tissue models, where the glucose concentration is intentionally held constant. Analogous concentration profiles, in addition to the true concentration profile, are also assigned to the datasets acquired from an animal model during a glucose clamping study as well as a human subject during an oral glucose tolerance test. We demonstrate that the spurious concentration profile-based calibration models are unable to provide prospective predictions, in contrast to those based on actual concentration profiles, especially for the physical tissue models. We also show that chance correlations incorporated by the calibration models are significantly less in Raman as compared to NIR absorption spectroscopy, even for the in vivo studies. Finally, our results suggest that the incorporation of chance correlations for in vivo cases can be largely attributed to the uncontrolled physiological sources of variations. Such uncontrolled physiological variations could either be intrinsic to the subject or stem from changes in the measurement conditions.     

A sampling-based method for ranking protein structural models by integrating multiple scores and features

One of the major challenges in protein tertiary structure prediction is structure quality assessment. In many cases, protein structure prediction tools generate good structural models, but fail to select the best models from a huge number of candidates as the final output. In this study, we developed a sampling-based machine-learning method to rank protein structural models by integrating multiple scores and features. First, features such as predicted secondary structure, solvent accessibility and residue-residue contact information are integrated by two Radial Basis Function (RBF) models trained from different datasets. Then, the two RBF scores and five selected scoring functions developed by others, i.e., Opus-CA, Opus-PSP, DFIRE, RAPDF, and Cheng Score are synthesized by a sampling method. At last, another integrated RBF model ranks the structural models according to the features of sampling distribution. We tested the proposed method by using two different datasets, including the CASP server prediction models of all CASP8 targets and a set of models generated by our in-house software MUFOLD. The test result shows that our method outperforms any individual scoring function on both best model selection, and overall correlation between the predicted ranking and the actual ranking of structural quality.