| Blood products(BPs)refer to protein and blood cells,which are purified from healthy human plasma or human plasma that has been specifically immunized or prepared by recombinant DNA technology.BPs are clinically irreplaceable and play an important role in coronary pneumonia,which belong to the national strategic reserve of materials.As a "Gatekeeper",the apheresis station plays an important role in controlling the quality of BPs for the reason of providing source plasma.How to control the quality of plasma is a major concern for quality control personnel.The protein content is the critical quality attribute for the quality evaluation of raw plasma.The Chinese Pharmacopoeia(2020)stipulates that the protein content of a single human plasma used for BPs production must be determined one by one,which should not be less than 50 g/L.This requires quality control personnel to adopt corresponding methods to evaluate the quality of plasma.The Kjeldahl method is a classic protein determination method with high accuracy,high stability,and small error.However,cumbersome,complicated,and time-consuming operation makes it not suitable for rapid determination of large quantities of samples.The refraction method with a refractometer is simple and fast to operate,but the accuracy is low and the error is large.Therefore,it is imperative to break through the limitations of conventional analysis methods and establish a new rapid analysis method.This is conducive to getting through the speed-limiting steps in BP's entire production process and realizing quality control throughout the entire process.Besides,this will lay the foundation for the development of our country's BP industry towards intelligent production.The preliminary research of our group shows that the laboratory benchtop near-infrared(NIR)spectroscopy instrument could be used for the quantitative analysis of protein content and NIR technology is a new and powerful tool for the determination of raw plasma.However,laboratory benchtop NIR analytical instruments have limitations such as bulkiness and high price.The main purpose of this research is to develop a special analyzer with strong stability,high accuracy,and cost-effectiveness based on the NIRONE miniature spectrometer.The specific chapters are as follows:(1)Analysis and correction of the influence of temperature on the near-infrared spectra of raw plasmaTemperature is one of the key factors affecting spectral stability.This chapter explores the hydrogen and other bond change pattern in the plasma solution at different temperatures,which could provide theoretical support for the temperature control of raw plasma using NIR technology at the molecular level.Besides,the effects of different temperature correction methods on the quantitative model were compared.The NIR spectra of raw plasma from 5? to 50? were collected and processed with continuous wavelet transform(CWT).Then,the changes of absorption peak intensity with temperature in the two spectral ranges of 7700-5500 cm-1 and 5000-4180 cm-1 were analyzed.Divide the sample into a calibration set and a validation set,and use external parameter orthogonalization(EPO),multilevel simultaneous component analysis(MSCA),and multilevel partial least square(MPLS)to correct the effects of temperature for quantitative analysis.The results show that there is no phase change in the plasma protein solution from 5 to 50?.Under the influence of temperature,the root means square error of prediction(RMSEP)of the quantitative analysis model of the protein content in the two ranges is 4.4536 g/L and 4.8439 g/L,respectively.Using EPO-PLS,MSCA,and MPLS methods to correct the influence of temperature,the RMSEP value after the MPLS method correction is reduced to 2.3691 g/L and 2.7887 g/L,which are reduced by 46.8%and 42.4%,respectively.This chapter clarified the effect of temperature 5-50? on protein structure at the molecular mechanism level and spectral characterization.MPLS is the optimal temperature correction method and has important guiding significance for improving the model effect under temperature disturbances.(2)Research on model transfer method between different types of instruments under the influence of temperatureTo improve the quantitative model accuracy of NIRONE spectrometers,model transfer research was carried out between spectrometers under different conditions.Taking raw plasma as the research object,improved principal component analysis(IPCA),which could ensure the spectral information is not lost as much as possible,was utilized for model transfer between the NIRONE(1.7 or 2.0)and the FT-NIR spectrometer.The model transfer includes three parts;(1)model transfer between different instruments at the same temperature;(2)model transfer at different temperatures within the instrument;(3)model transfer at different temperatures between instruments.Piecewise direct standardization(PDS)method was utilized to compare with the transfer effect of the IPCA.The results show that the RMSEP values before the model transfer of the NIRONE 1.7 and 2.0 models are 5.1307 g/L and 6.0918 g/L,respectively.After model transfer with PDS,the RMSEP values of the two instruments are 5.4440 g/L and 5.9532 g,respectively.It means that the NIRONE 1.7 model transfer results are poor,while the NIRONE 2.0 model results are slightly improved.The RMSEP values of the IPCA method for model transfer between different instruments at the same temperature are 4.9896 g/L and 5.4351 g/L and model prediction errors were reduced by 2.7%and 10.8%,respectively.Besides,the RMSEP values of the IPCA method for model transfer at different temperatures between instruments are 3.3403 g/L and 3.4732 g/L,which are lower than 5.0328 g/L and 5.6760 g/L of NIRONE and closer to 3.2837 g/L of FT-NIR.The research in this chapter proves that IPCA method can achieve excellent results for the model transfer from NIRONE to FT-NIR spectrometer at the same or different temperatures.It plays a role in improving the prediction accuracy of the low-resolution NIRONE spectrometer and is an effective chemometric method to improve the accuracy of the dedicated instrument.(3)Similarity methods based on vector inner product and Cosine 0 for NIR spectra discriminationAiming at the prediction accuracy of the qualitative model,similarity calculation methods S1 and S2 are proposed for the first time,which breaks through the bottleneck of low accuracy of extremely similar spectra.This method characterizes the spectral difference from the angle of the inner product of the two vectors and the cosine ?,then introduces the concept of sensitivity factors(u and v).The formula S1=(2|X'|Y|/|X|2+|Y|2)u(cos ?)v,and S2=(2|(X-X)'(Y-Y)|/|X-X|2+|Y-Y|2)u(corr(X,Y))v by centering each item.The experiments in this chapter are based on microcrystalline cellulose(MCC)PH101 spectrum and PH102 spectrum to investigate the discrimination ability of the sensitivity factors u and v in S1 and S2.By using the Gaussian fitting method to fit a series of small peaks on the PH101 spectrum,the ability of the S1 and S2 methods to recognize very small differences is verified.To investigate the discriminative effect of S1,S2,correlation coefficient method,Cosine 0,partial least squares discriminative analysis(PLS-DA),support vector machine(SVM),and soft independent modeling of class analysis(SIMCA)methods by using MCC PH101,MCC PH102,carboxymethyl Starch(CMS),hydroxypropyl cellulose(HPC)and starch.The results show that the discriminative ability of u in S1 and S2 is higher than that of v and S2 is higher than S1.The experimental results of the ability to recognize very small differences show that the method has different recognition capabilities for different peak widths,peak heights,and peak areas.The two methods are easier to identify the higher peak with fixed peak width,the wider peak with fixed peak height,and the sharper peak with constant peak area.The qualitative discrimination experiment of raw materials identification shows that the sensitivity of correlation coefficient and Cosine ? methods are 100%,while the specificity results are only 75%.The sensitivity and specificity results of S1,S2,PLS-DA,SVM,and SIMCA are all 100%.The research in this chapter proposes new qualitative discrimination methods S1 and S2,both of which can distinguish extremely similar spectra,realize the qualitative discrimination of different materials,and provide new methods and new ideas for the identification of subtle differences.(4)Establishment of qualitative discriminant model of total protein content in raw plasma The spectrum of raw plasma is extremely similar,and the bottleneck technology for evaluating the similarity of the spectrum is in urgent need of a breakthrough.This chapter establishes a new discriminant model for raw plasma based on S1 or S2.In this chapter,raw plasma samples with total protein higher than 50 g/L were defined as category H,and lower than 50 g/L were defined as category L.Correlation coefficient,Cosine ?,S1 and S2 were used for raw plasma NIR spectroscopy to determine whether the protein content in raw plasma was higher than 50 g/L.The results show that the correlation coefficient,Cosine ?,S1(u=v=1),and S2(u=v=1)can correctly identify the H samples and the sensitivity reaches 100%,but the L samples cannot be rejected,the specificity is 0%.Select range of(a)9000-8200 cm-1,(b)7800-6200 cm-1,(c)6100-4800 cm-1 to optimize the sensitivity factor.The results show that the S1(52?u/v?70)and S2(47?u/v?66)methods in the interval of 9000-8200 cm-1 can distinguish the H and L category in the calibration set with 100%sensitivity and specificity.The sensitivity and specificity in the validation set are 96%and 100%,respectively.The research in this chapter shows that S1 and S2 methods can amplify the difference between the highly similar plasma spectra by adjusting the sensitivity factors u and v,which can be used to quickly determine whether the protein content of the raw plasma is higher than 50 g/L.Therefore,a new method for the rapid release of plasma during new collection and re-examination at the plasma station was provided.(4)Design of portable analyzer for raw plasmaTemperature correction,model transfer,and qualitative discrimination methods of extremely similar spectra have solved the problems of the stability and accuracy of the spectrum of the NIRONE from the perspective of chemometrics.Then we design and develop an analyzer based on the NIRONE miniature NIR sensor.The overall structure design of the analyzer mainly includes the following sub-modules:power supply system,a lighting module,measuring platform,detector module,data communication and processing,human-computer interaction.The analyzer was designed with side-illuminated transmission mode from hardware,software,and mechanical structure subsystem.Then relevant tests were performed to verify analyzer performance.The hardware subsystem design includes the main control chip,power supply circuit,clock circuit,near-infrared spectrometer signal acquisition circuit,man-machine interface circuit,temperature acquisition circuit,light source,and fan control circuit.The software subsystem design includes upper computer software and lower computer software.The upper computer software is Windows 10 system software,which communicates with the analyzer through a USB data cable.The main interface of the software includes several functional modules such as model management,spectrum acquisition,data retrieval,workflow setting,instrument performance testing,etc.Patented methods S1,S2,MPLS,IPCA,and other raw plasma protein analysis chemometric methods were embedded in the software.Test results show that the performance parameters of the analyzer meet the requirements of the analysis instruments.The study in this chapter developed a special analyzer with low cost,stable performance,and accurate results,which could be a reliable tool for the quality control of raw plasma. |
PDF Full Text Request