Acceptance areas for multivariate classification derived by projection methods

Pomerantsev, Alexey L.

doi:10.1002/cem.1147

Cited by 120 publications

(67 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A supervised extension of PCA was used to establish a statistical model of normality in a healthy population of Turkish neonates (Mohammadi Aygen et al 2011;Pomerantsev 2008;Vanderginste and Massart 1998). In this approach hierarchical multi-model/soft independent modeling for class analogy (HMM/SIMCA) was used.…”

Section: Discussionmentioning

confidence: 99%

NMR-Based Screening for Inborn Errors of Metabolism: Initial Results from a Study on Turkish Neonates

et al. 2014

View full text Add to dashboard Cite

Approximately 1 in 400 neonates in Turkey is affected by inherited metabolic diseases. This high prevalence is at least in part due to consanguineous marriages. Standard screening in Turkey now covers only three metabolic diseases (phenylketonuria, congenital hypothyroidism, and biotinidase deficiency). Once symptoms have developed, tandem-MS can be used, although this currently covers only up to 40 metabolites. NMR potentially offers a rapid and versatile alternative.We conducted a multi-center clinical study in 14 clinical centers in Turkey. Urine samples from 989 neonates were collected and investigated by using NMR spectroscopy in two different laboratories. The primary objective of the present study was to explore the range of variation of concentration and chemical shifts of specific metabolites without clinically relevant findings that can be detected in the urine of Turkish neonates. The secondary objective was the integration of the results from a healthy reference population of neonates into an NMR database, for routine and completely automatic screening of congenital metabolic diseases.Both targeted and untargeted analyses were performed on the data. Targeted analysis was aimed at 65 metabolites. Limits of detection and quantitation were determined by generating urine spectra, in which known concentrations of the analytes were added electronically as well as by real spiking. Untargeted analysis involved analysis of the whole spectrum for abnormal features, using statistical procedures, including principal component analysis. Outliers were eliminated by model building. Untargeted analysis was used to detect known and unknown compounds and jaundice, proteinuria, and acidemia. The results will be used to establish a database to detect pathological concentration ranges and for routine screening.

show abstract

Section: Discussionmentioning

confidence: 99%

NMR-Based Screening for Inborn Errors of Metabolism: Initial Results from a Study on Turkish Neonates

et al. 2014

View full text Add to dashboard Cite

show abstract

“…In particular, recently other criteria for the definition of the model space have been proposed by Pomerantsev [26]. …”

Section: Soft Independent Modeling Of Class Analogies (Simca)mentioning

confidence: 99%

Classification Methods in Chemometrics

Marini

2010

CAC

View full text Add to dashboard Cite

Pattern recognition methods, i.e. the methods concentrating on the possibility of assigning an object to a class based on the result of a set of measurements are ubiquitous in chemometrics. In this paper, the main chemometric classification methods are discussed in terms of their nature, behavior, advantages and drawbacks. Both parametric and non parametric or discriminant and modeling techniques are illustrated together with a discussion of some applications to real world problems

show abstract

“…On the other hand, the score distance gives an account of how far the observation is from the bulk of samples of the particular class, and, as the name suggests, is linked to the value of the scores. Starting from this general concept, several ways of defining the terms in Equation 12 have been proposed (33,(37)(38)(39). Among these, one of the most frequently used relies on statistics borrowed from multivariate statistical process control (MSPC; 40), and defines OD ig and SD ig using the two variables T 2 and Q, which represent the squared Mahalanobis distance of a sample to the center of the score space and the sum of the squared residuals, respectively.…”

Section: Class-modeling Techniquesmentioning

confidence: 99%

Chemometric Classification Techniques as a Tool for Solving Problems in Analytical Chemistry

Bevilacqua

Nescatelli

Bucci

et al. 2014

Journal of AOAC INTERNATIONAL

View full text Add to dashboard Cite

Supervised pattern recognition (classification) techniques, i.e., the family of chemometric methods whose aim is the prediction of a qualitative response on a set of samples, represent a very important assortment of tools for solving problems in several areas of applied analytical chemistry. This paper describes the theory behind the chemometric classification techniques most frequently used in analytical chemistry together with some examples of their application to real-world problems.

show abstract

Acceptance areas for multivariate classification derived by projection methods

Cited by 120 publications

References 31 publications

NMR-Based Screening for Inborn Errors of Metabolism: Initial Results from a Study on Turkish Neonates

NMR-Based Screening for Inborn Errors of Metabolism: Initial Results from a Study on Turkish Neonates

Classification Methods in Chemometrics

Chemometric Classification Techniques as a Tool for Solving Problems in Analytical Chemistry

Contact Info

Product

Resources

About