Abstract: In recent years, due to the development of supercomputers and chemical statistical software, especially the in-depth study and extensive application of chemometrics, it has become the fastest-growing and most striking spectroscopic technology. And because the technology is convenient and fast, there is no need to pre-treat samples, and it is suitable for online analysis and other characteristics. It is constantly being paid attention to and applied in the field of pharmaceutical analysis. The near infrared (NearInfrared, NIR) spectrum has a wavelength range of 780 ~ 2526nm (12820 ~ 3959cm-1), and this wavelength range is usually divided into a near infrared short wave region (780 ~ 1100nm) and a near infrared long wave region (1100 ~ 2526nm) . Since this region is mainly frequency-doubled and combined-frequency absorption of OH, NH, CH, SH and other hydrogen-containing groups, the spectral bandwidth and overlap are serious, and the absorption signal is weak and the information analysis is complicated. Early, but the analytical value has not been given enough attention. 1 Measurement of near infrared spectroscopy According to the way of obtaining NIR spectrum, there are usually two types of transmission (Transmittance) and diffuse reflection (DiffuseReflectance) [2]. The quantitative relationship of the transmission measurement method complies with Lambert-Beer's law and is mainly applicable to liquid samples. Its normal operating wavelength range is 850 ~ 1050nm [3]. Shi Yuehua and others of Zhejiang University used this principle to establish a regression equation between the integrated value of the near-infrared absorption peak area of ​​vitamin E at 6061 and 5246 cm-1 and its concentration in a concentration range of 93% to 97.4%. The samples are predicted, and the error and relative error are within 0.79% to 0.9% [4, 5]. Diffuse reflectometry is a commonly used method for near-infrared measurements of solid samples. When the light source is perpendicular to the surface of the sample, part of the diffuse reflected light will be scattered in all directions. The method of measuring the scattered light intensity by placing the detector at a 45o angle to the vertical light is called the diffuse reflection method. The relationship between the diffuse reflected light intensity A and the reflectivity R is in the formula, R1 is the reflected light intensity, and R0 is the surface reflected light intensity that is not absorbed at all. Domestic people have successively measured the contents of arginine aspirin [6], analgin [7], rutin and vitamin E [8] using diffuse reflectance technique, and used reflection spectrometry to measure the quality of sulfathiazole [9] Evaluation. Based on the test of transmission and diffuse reflection, in order to adapt to the direct measurement of the near infrared spectrum of different substances in different states, fiber optic technology has been widely used in NIR since the 1990s. Optical fiber can not only conveniently transmit spectral signals, but various optical fiber probes also greatly facilitate NIR for various fast online analysis. 2 Application of near infrared spectroscopy in drug analysis 2.1 Scope of application Near-infrared spectroscopy has a wide range of applications in the field of drug analysis. It is not only suitable for many different states of drugs such as raw materials [10], complete tablets, capsules and liquid preparations [11], but also for different types Analysis of drugs such as protein [12], Chinese herbal medicine [13], antibiotics [14], etc. NIR is more suitable for the analysis and detection of the purity of raw materials, packaging materials, etc. and the monitoring of production processes [15, 16]; the use of different optical fiber probes can achieve online continuous analysis and monitoring of production processes [17, 18, 19, 20, twenty one]. 2.2 Qualitative and quantitative analysis Modern near-infrared spectroscopy technology does not directly perform qualitative or quantitative analysis by observing the characteristics of the test sample spectrum or measuring the test sample spectrum parameters, but first by determining the spectrum, composition or property data of the sample calibration set (composition or property data needs (Measured by other recognized standard methods), using appropriate chemometrics methods to establish a calibration model, and then through the established calibration model to compare with unknown samples to achieve qualitative or quantitative analysis. 2.2.1 Qualitative analysis The near-infrared spectrum has a wide band and is not very characteristic, so it is rarely used for the identification of compound groups and the identification of structures like other spectra (such as ultraviolet spectrum and infrared spectrum). Qualitative analysis of near infrared spectroscopy is generally used to determine the location of the analyzed sample in the known sample concentration [22]. Common methods include: (1) Discriminant analysis method: Discriminant analysis is a classic qualitative identification method. The basic idea is that the same sample has similar spectral absorption at different wavelengths. The comparison between such spectra can be the original spectrum or the processed spectrum. . (2) Principal Component Analysis (PrincipalComponentAnalysisPCA) method: the PCA method is used to compress the spectral data at multiple wavelengths into a limited number of factor spaces, and then determine the attribution category by the sample's score in each factor space, but PCA compares the sample with The exact position between the calibration sets lacks a quantitative explanation. Ren Yulin and others used this method to study the near-infrared diffuse reflectance spectrum of Qutong Tablet [23], and concluded that performing principal component analysis on the standardized data can reduce the scattering effect caused by the change in particle size, and used the second The principal component score plotted against the first principal component can distinguish the qualified sample from the unqualified sample. The disadvantage is that this method is easy to make mistakes when the content of real medicine and inferior medicine is quite close [24]. (3) Mahalanobis Distance MD: The core of this method is to quantitatively describe the position of the measurement sample from the calibration set sample through the spectral distance at multiple wavelengths, so it is very useful in spectral matching abnormal point detection and model extrapolation. . However, when applying this method, the choice of the wavelength position is very important. Too few wavelength points, the spectrum can not be described reasonably; too many wavelength points and large amount of calculation. Therefore, Xu Guangtong proposed to combine PCA and Mahalanobis distance to solve the model Judgment of the applicability of PCA can make full use of PCA to reduce the dimensionality of a large amount of spectral data, and also solve the problem of selecting the wavelength point when calculating the Mahalanobis distance, avoiding the collinearity that occurs when a large amount of spectral data is directly calculated by the Mahalanobis distance Or a large amount of calculation and other problems, and overcome the problem of using PCA itself to judge the boundary is not easy to quantify [25]. 2.2.2 Quantitative analysis Generally, there is no need to pre-treat the sample when measuring near infrared spectroscopy, but the measured spectrum may be affected by various interference factors. It is difficult to obtain accurate quantitative analysis results using the spectral data obtained at a single wavelength. The NIR spectrum structure is complex and the spectrum overlaps more. Therefore, when performing quantitative analysis, the data obtained at multiple wavelengths is generally used and certain data processing is performed to obtain accurate and reliable analysis results. Common methods are as follows: (1) Principal Component Regression (PCR): The principle is the same as PCA. Ren Yulin of Jilin University has conducted in-depth research in this regard [26]. PCR plays an important role in interpreting spectral data. From the weight map of principal components, it is possible to determine which component the principal component is related to, but it is still the most difficult problem to explain exactly and comprehensively what each principal component represents. (2) Partial Least Squares (PartialLeastSquarePLS): This method is a full-spectrum analysis method that makes full use of useful information at multiple wavelengths without the need to deliberately select wavelengths, and can filter out raw data noise and improve the signal-to-noise ratio. It is suitable to use in NIR to solve the non-linear problem of interaction effects [27]. Experiments have proved that the combination of PLS ​​and NIR diffuse reflectance spectroscopy can directly analyze solid powder drugs sulfamethoxazole [28], spironolactone [29], analgin [30], and sulfamidine [31]. Compared with other methods, it has the advantages of speed, simplicity and no damage to the sample. (3) Artificial Neural Networks (ArtificialNeuralNetworksANN): ANN method research has emerged in recent years. Based on the spectral data of each component of the sample, an artificial neural network model is established to predict unknown samples and discuss various parameters that affect the network. Quantitative analysis of aspirin [32], paracetamol [33], Midea Kang [34] and other drugs using ANN method shows that the biggest advantage of ANN method is its anti-interference, anti-noise and strong nonlinear conversion ability. For some In special cases, ANN will get smaller correction errors and prediction errors, and its prediction results are slightly better than PLS (t test has no significant difference). This may be due to the stronger nonlinear processing ability of the ANN method. In addition, multiple linear regression (MultipleLinearRegressionMLR), topology (TopologyTP) and other methods are also used in near infrared spectroscopy analysis. 3 problems and prospects Although NIR has shown great vitality in the field of pharmaceutical analysis, it currently has certain weaknesses. First, it is an indirect relative analysis technique. It collects a large number of representative standard samples, measures the necessary data through strict and detailed chemical analysis, and then establishes a mathematical model through the computer to predict the results of unknown samples. The establishment of the model requires a lot of manpower, material resources and financial resources; secondly, because the NIR spectral region is a molecular frequency doubling and combined frequency vibration spectrum, the signal is weak, and the spectral peaks overlap seriously, so it can only be used for constant analysis. The amount of the measured component should generally be greater than 0.1% of the sample weight; in addition, when performing near infrared spectroscopy, the characteristics of the sample, the design of the analysis experiment, and data processing should be considered to obtain the correct analysis results. Therefore, the establishment of a reliable calibration model is the key to the success of near-infrared spectroscopy, and the rational experimental design and appropriate analysis model are the key to the establishment of a calibration model [35]. The biggest feature of NIR spectroscopy is that it is simple and fast to operate, and it can be measured in situ and online without damaging the sample; the measurement signal can be transmitted and analyzed at a long distance; especially in combination with computer technology and optical fiber technology, NIR transmission, scattering, Diffuse reflectance spectroscopy detection method, without the use of chemical reagents, without pretreatment, can directly analyze granular, solid, paste, opaque samples. These characteristics are gradually recognized by the pharmaceutical industry and show great potential, and have broad application prospects in pharmaceutical work and quality control analysis. In addition, NIR used in the analysis of content and moisture in the production process also showed a unique charm [36]. At present, NIR has become a standard analysis method of AOAC (Association of Official Analytical Chemists) used in drug testing [37]. The cooperative development of instrument manufacturers and drug analysis experts has enabled the FDA, European and Canadian Drug Administrations to formally study the feasibility of replacing the cumbersome and time-consuming conventional analysis methods with near infrared spectroscopy, and some test items have been approved by the FDA as standard methods. USP (United States Pharmacopia 25th Edition) has recently added a near infrared analysis method in the appendix [38]. Domestically, with the support of SDA (State Drug Administration), our institute is exploring the feasibility of adopting NIR for rapid identification and quantitative analysis in the process of drug supervision, inspection and law enforcement. Combined with the national random inspection work, the accuracy of the NIR model and the error of the model transmission are systematically evaluated. The development of this work is of great significance to combat counterfeit and inferior drugs. Glitter Russian Lashes,Russian Sparkly Eyelashes,Russian Sparkly Lashes,Gold Glitter Russian Lashes Zhengzhou Cuka Electronic Commerce Co., Ltd. , https://www.cukalashes.com
Application of near infrared spectroscopy in drug analysis