Supervised Analysis

Supervised analysis is performed when there additional information or data available, such as reference spectra, calibration samples and concentrations. 

The simplest form of supervised analysis is to look for a chemical component using a reference spectrum. For this, the most widely used methods include correlation coefficients, Fuclidean distance and DCLS regression, in supervised model (using reference spectra). 

When a supervised analysis is mentioned in the pharmaceutical industry, it often refers to a concentration prediction using a chemometric model. By nature, the objective of the analysis is not to identify the ingredients of the sample, as they are all known. Rather, the aim is to predict their concentrations in the sample. 

The second step is to prepare a validation set of samples in the same way that the calibration set is prepared. The validation set is also measured in the same way as the calibration set. The model is then applied to the spectra of the validation 

---

Since 1992 a variety of related but much more powerful data-handling strategies have been applied to the supervised analysis of PyMS data. Such methods fall within the framework of chemometrics the discipline concerned with the application of statistical and mathematical methods to chemical data.81-85 These methods seek to relate known spectral inputs to known targets, and the resulting model is then used to predict the target of an unknown input.86... 

St/pen/7sed Data Mining. Searching large volumes of data for hidden predictive relationships. Supervised analysis requires one or more "dependent" or response variables, to be predicted from a set of "independent" or predictor variables. The techniques used include various classification methods (decision tree, support vector, Bayesian) and various estimation methods (regression, neural nets). 

Figure 10.5 Cluster analysis, (a) A combination of unsupervised clustering and heatmap visualization. The Euclidean distance measure and Ward linkage are used. Peptide intensities are log-transformed and normalized to zero mean unit variance (row by row). The profiles of 27 non-small-cell lung cancer patients are intermingled with those of 13 healthy controls (columns) (b) Supervised analysis using 11 peptides with Benjamini-Hochberg adjusted p-values <0.001 results in two distinctive branches at the root of the tree. Two cancer profiles are grouped with those of the healthy controls. All but one of the peptides are upregulated in cancer samples.

The main purposes of multivariate analysis are data reduction (unsupervised analysis) and data modeling like regression and/or classification models (supervised analysis). 

The variable of the data matrix yielded by supervised analysis are the real concentrations of the detected metabolites, and the appUcatimi of multivariate methods can directly lead to the understanding of the metabolic system under study. 

Multiple linear regression is strictly a parametric supervised learning technique. A parametric technique is one which assumes that the variables conform to some distribution (often the Gaussian distribution) the properties of the distribution are assumed in the underlying statistical method. A non-parametric technique does not rely upon the assumption of any particular distribution. A supervised learning method is one which uses information about the dependent variable to derive the model. An unsupervised learning method does not. Thus cluster analysis, principal components analysis and factor analysis are all examples of unsupervised learning techniques. 

Discriminant emalysis is a supervised learning technique which uses classified dependent data. Here, the dependent data (y values) are not on a continuous scale but are divided into distinct classes. There are often just two classes (e.g. active/inactive soluble/not soluble yes/no), but more than two is also possible (e.g. high/medium/low 1/2/3/4). The simplest situation involves two variables and two classes, and the aim is to find a straight line that best separates the data into its classes (Figure 12.37). With more than two variables, the line becomes a hyperplane in the multidimensional variable space. Discriminant analysis is characterised by a discriminant function, which in the particular case of hnear discriminant analysis (the most popular variant) is written as a linear combination of the independent variables ... 

Sample test data are either manually entered into the system or captured from analytical instmments coimected to the LIMS. The system performs any necessary calculations and compares the result to the appropriate specification stored in the database. If the comparison indicates the material is in conformance, the system can automatically provide an approval. Otherwise, the LIMS can alert lab supervision to the nonconforming sample analysis. 

In all of these cases, the extent and nature of supervision can vary greatly. Like management systems themselves, supervision can be results-oiiented or procedurally focused, depending on the preferences of the company and the individual. One boss may tell her staff, "Fix that problem we re having with the product specs in the ag-chem division," and expect to hear no more about it until the problem is resolved. Another supervisor with the same problem may expect a detailed situation analysis, formal recommendations, and con-... 

Most of the supervised pattern recognition procedures permit the carrying out of stepwise selection, i.e. the selection first of the most important feature, then, of the second most important, etc. One way to do this is by prediction using e.g. cross-validation (see next section), i.e. we first select the variable that best classifies objects of known classification but that are not part of the training set, then the variable that most improves the classification already obtained with the first selected variable, etc. The results for the linear discriminant analysis of the EU/HYPER classification of Section 33.2.1 is that with all 5 or 4 variables a selectivity of 91.4% is obtained and for 3 or 2 variables 88.6% [2] as a measure of classification success. Selectivity is used here. It is applied in the sense of Chapter... 

It is not necessary that participating laboratories be formally recognized, accredited or certified. Measurement of the property of interest should be completed by, or under the supervision of a technically competent manager qualified either in terms of suitable academic qualifications or relevant work experience. The participating laboratory should consider the analysis as a very special one, to be performed with special attention and all possible care, and not have it performed as part of its regular routine. 

Consequently, separate experiments for the determination of extraction efficiency are often not required. An expert statement based on the results of metabolism studies is sufficient in most cases. These statements should also refer to the extraction solvent used for the analysis of samples of supervised trials. Residue levels found in these trials are the criterion for GAP and the basis for the setting of MRLs. Even if a solvent with insufficient extraction efficiency is used for samples from supervised trials, the later choice of better solvents would not result in lower safety for the consumer. 

Goodacre, R. Trew, S. Wrigley-Jones, C. Saunders, G. Neal, M. J. Porter,N. Kell, D. B. Rapid and quantitative analysis of metabolites in fermentor broths using pyrolysis mass spectrometry with supervised learning Application to the screening of Penicillium chrysogenum fermentations for the overproduction of penicillins. Anal. Chim. Acta 1995,313, 25 43. 

---