Home About us Contact
|
Home About us Contact | ||
|
|
||
Classification Methods (classification + methods)
Selected AbstractsUsing unlabelled data to update classification rules with applications in food authenticity studiesJOURNAL OF THE ROYAL STATISTICAL SOCIETY: SERIES C (APPLIED STATISTICS), Issue 1 2006Nema Dean Summary., An authentic food is one that is what it purports to be. Food processors and consumers need to be assured that, when they pay for a specific product or ingredient, they are receiving exactly what they pay for. Classification methods are an important tool in food authenticity studies where they are used to assign food samples of unknown type to known types. A classification method is developed where the classification rule is estimated by using both the labelled and the unlabelled data, in contrast with many classical methods which use only the labelled data for estimation. This methodology models the data as arising from a Gaussian mixture model with parsimonious covariance structure, as is done in model-based clustering. A missing data formulation of the mixture model is used and the models are fitted by using the EM and classification EM algorithms. The methods are applied to the analysis of spectra of food-stuffs recorded over the visible and near infra-red wavelength range in food authenticity studies. A comparison of the performance of model-based discriminant analysis and the method of classification proposed is given. The classification method proposed is shown to yield very good misclassification rates. The correct classification rate was observed to be as much as 15% higher than the correct classification rate for model-based discriminant analysis. [source] Categorizing Urgency of Infant Emergency Department Visits: Agreement between CriteriaACADEMIC EMERGENCY MEDICINE, Issue 12 2006Rakesh D. Mistry MD Abstract Background The lack of valid classification methods for emergency department (ED) visit urgency has resulted in large variation in reported rates of nonurgent ED utilization. Objectives To compare four methods of defining ED visit urgency with the criterion standard, implicit criteria, for infant ED visits. Methods This was a secondary data analysis of a prospective birth cohort of Medicaid-enrolled infants who made at least one ED visit in the first six months of life. Complete ED visit data were reviewed to assess urgency via implicit criteria. The explicit criteria (adherence to prespecified criteria via complete ED charts), ED triage, diagnosis, and resources methods were also used to categorize visit urgency. Concordance and agreement (,) between the implicit criteria and alternative methods were measured. Results A total of 1,213 ED visits were assessed. Mean age was 2.8 (SD ± 1.78) months, and the most common diagnosis was upper respiratory infection (21.0%). Using implicit criteria, 52.3% of ED visits were deemed urgent. Urgent visits using other methods were as follows: explicit criteria, 51.8%; ED triage, 60.6%; diagnosis, 70.3%; and resources, 52.7%. Explicit criteria had the highest concordance (78.3%) and agreement (,= 0.57) with implicit criteria. Of limited data methods, resources demonstrated the best concordance (78.1%) and agreement (,= 0.56), while ED triage (67.9%) and diagnosis (71.6%) exhibited lower concordance and agreement (,= 0.35 and ,= 0.42, respectively). Explicit criteria and resources equally misclassified urgency for 11.1% of visits; ED triage and diagnosis tended to overclassify visits as urgent. Conclusions The explicit criteria and resources methods best approximate implicit criteria in classifying ED visit urgency in infants younger than six months of age. If confirmed in further studies, resources utilized has the potential to be an inexpensive, easily applicable method for urgency classification of infant ED visits when limited data are available. [source] Clustering technique for risk classification and prediction of claim costs in the automobile insurance industryINTELLIGENT SYSTEMS IN ACCOUNTING, FINANCE & MANAGEMENT, Issue 1 2001Ai Cheo Yeo This paper considers the problem of predicting claim costs in the automobile insurance industry. The first stage involves classifying policy holders according to their perceived risk, followed by modelling the claim costs within each risk group. Two methods are compared for the risk classification stage: a data-driven approach based on hierarchical clustering, and a previously published heuristic method that groups policy holders according to pre-defined factors. Regression is used to model the expected claim costs within a risk group. A case study is presented utilizing real data, and both risk classification methods are compared according to a variety of accuracy measures. The results of the case study show the benefits of employing a data-driven approach. © 2001 John Wiley & Sons, Ltd. [source] Assessment of sun reactive skin type with multiple correspondence analysis, hierarchical and tree-structured classification methodsINTERNATIONAL JOURNAL OF COSMETIC SCIENCE, Issue 4 2002C. Guinot Synopsis The sun reactive skin type classification is based on sunburn susceptibility, tanning ability and phenotypic information. As subjects rarely match all features of a given skin type, the attribution to a class is partially subjective. The aims of the study, were to analyse the contribution of each characteristic to the classification made by the expert, and to establish a classification based on a statistical approach conducted on 212 women living in the Ile-de-France area. Multiple regression was used to construct a formula for each phototype. The coefficients obtained demonstrated that the importance of each characteristic was extremely variable from one phototype to another, suggesting that the phototype determination could be facilitated by adding a weight for every characteristic in the decision. Then, multiple correspondence analysis and clustering analysis methods showed that one phototype could be divided into two more homogenous classes. Résumé La classification du phototype, reflet de la protection naturelle de la peau au soleil, repose sur la susceptibilitéà prendre des coups de soleil, la capacitéà bronzer et trois informations phénotypiques. En pratique, les individus présentant rarement l'ensemble des caractéristiques prévues pour un phototype donné, la décision de l'expert d'attribuer un sujet particulier à une classe repose sur une part de subjectivité. Le but de cette étude, conduite sur 212 femmes volontaires d'Ile-de-France, était d'analyser la contribution de chacune des caractéristiques dans le processus de classification effectué par l'expert pour déterminer le phototype, et de proposer une classification reposant sur une approche statistique. Une équation de régression multiple a été construite pour chacun des phototypes représentés dans l'échantillon. Les coefficients obtenus ont démontré que l'importance accordée en pratique à chaque caractéristique était très variable d'un phototype à l'autre, suggérant que la détermination du phototype pourrait être facilitée par la mise en place d'une pondération des caractéristiques dans la règle de décision. Les méthodes d'analyse factorielle des correspondances multiples et de classification ascendante hiérarchique ont ensuite montré que les sujets de l'un des phototypes pourraient être répartis en deux classes plus homogènes, les autres classes demeurant globalement inchangées par rapport à leur définition originelle. [source] Discrimination of cyanobacterial strains isolated from saline soils in Nakhon Ratchasima, Thailand using attenuated total reflectance FTIR spectroscopyJOURNAL OF BIOPHOTONICS, Issue 8-9 2010Somchanh Bounphanmy Abstract A method was developed whereby high quality FTIR spectra could be rapidly acquired from soil-borne filamentous cyanobacteria using ATR FTIR spectroscopy. Spectra of all strains displayed bands typical of those previously reported for microalgae and water-borne cyanobacteria, with each strain having a unique spectral profile. Most variation between strains occurred in the C,O stretching and the amide regions. Soft Independent Modelling by Class Analogy (SIMCA) was used to classify the strains with an accuracy of better than 93%, with best classification results using the spectral region from 1800,950 cm,1. Despite this spectral region undergoing substantial changes, particularly in amide and C,O stretching bands, as cultures progressed through the early-, mid- to late-exponential growth phases, classification accuracy was still good (,80%) with data from all growth phases combined. These results indicate that ATR/FTIR spectroscopy combined with chemometric classification methods constitute a rapid, reproducible, and potentially automated approach to classifying soil-borne filamentous cyanobacteria. (© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim) [source] Classification of GC-MS measurements of wines by combining data dimension reduction and variable selection techniquesJOURNAL OF CHEMOMETRICS, Issue 8 2008Davide Ballabio Abstract Different classification methods (Partial Least Squares Discriminant Analysis, Extended Canonical Variates Analysis and Linear Discriminant Analysis), in combination with variable selection approaches (Forward Selection and Genetic Algorithms), were compared, evaluating their capabilities in the geographical discrimination of wine samples. Sixty-two samples were analysed by means of dynamic headspace gas chromatography mass spectrometry (HS-GC-MS) and the entire chromatographic profile was considered to build the dataset. Since variable selection techniques pose a risk of overfitting when a large number of variables is used, a method for coupling data dimension reduction and variable selection was proposed. This approach compresses windows of the original data by retaining only significant components of local Principal Component Analysis models. The subsequent variable selection is then performed on these locally derived score variables. The results confirmed that the classification models achieved on the reduced data were better than those obtained on the entire chromatographic profile, with the exception of Extended Canonical Variates Analysis, which gave acceptable models in both cases. Copyright © 2008 John Wiley & Sons, Ltd. [source] Glass analysis for forensic purposes,a comparison of classification methodsJOURNAL OF CHEMOMETRICS, Issue 5-6 2007Grzegorz Zadora Abstract One of the purposes of the chemical analysis of glass fragments (pieces of glass of linear dimension ca. 0.5,mm) for forensic purposes is a classification of those fragments into use categories, for example windows, car headlights and containers. The object of this research was to check the efficiency of Naïve Bayes Classifiers (NBCs) and Support Vector Machines (SVMs) to the application of the classification of glass objects when those objects may be described by the major and minor elemental concentrations obtained by Scanning Electron Microscopy coupled with an Energy Dispersive X-ray spectrometer which is routinely used in many forensic institutes. Copyright © 2007 John Wiley & Sons, Ltd. [source] Hybrid Bayesian networks: making the hybrid Bayesian classifier robust to missing training dataJOURNAL OF CHEMOMETRICS, Issue 5 2003Nathaniel A. Woody Abstract Many standard classification methods are incapable of handling missing values in a sample. Instead, these methods must rely on external filling methods in order to estimate the missing values. The hybrid network proposed in this paper is an extension of the hybrid classifier that is robust to missing values. The hybrid network is produced by performing empirical Bayesian network structure learning to create a Bayesian network that retains its classification ability in the presence of missing data in both training and test cases. The performance of the hybrid network is measured by calculating a misclassification rate when data are removed from a dataset. These misclassification curves are then compared against similar curves produced from the hybrid classifier and from a classification tree. Copyright © 2003 John Wiley & Sons, Ltd. [source] Parental Bonding and Adult Attachment Styles in Different Types of Stalker,JOURNAL OF FORENSIC SCIENCES, Issue 6 2008Rachel D. MacKenzie D.Psych. Abstract:, Attachment theory is one of the earliest and most vigorously promoted explanations of the psychological processes that underlie stalking behavior. Insecure attachment has been proposed as impairing the management of relationships, thus increasing the propensity to stalk. The current study explored the parental bonding and adult attachment styles of 122 stalkers referred to a specialist forensic clinic. Stalkers were grouped according to two common classification methods: relationship and motivation. Compared to general community samples, stalkers were more likely to remember their parents as emotionally neglectful and have insecure adult attachment styles, with the degree of divergence varying according to stalker type and mode of classification. In offering support for the theoretical proposition that stalking evolves from pathological attachment, these findings highlight the need to consider attachment in the assessment and management of stalkers. Also emphasized is the importance of taking classification methods into account when interpreting and evaluating stalking research. [source] Computational methods in authorship attributionJOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY, Issue 1 2009Moshe Koppel Statistical authorship attribution has a long history, culminating in the use of modern machine learning classification methods. Nevertheless, most of this work suffers from the limitation of assuming a small closed set of candidate authors and essentially unlimited training text for each. Real-life authorship attribution problems, however, typically fall short of this ideal. Thus, following detailed discussion of previous work, three scenarios are considered here for which solutions to the basic attribution problem are inadequate. In the first variant, the profiling problem, there is no candidate set at all; in this case, the challenge is to provide as much demographic or psychological information as possible about the author. In the second variant, the needle-in-a-haystack problem, there are many thousands of candidates for each of whom we might have a very limited writing sample. In the third variant, the verification problem, there is no closed candidate set but there is one suspect; in this case, the challenge is to determine if the suspect is or is not the author. For each variant, it is shown how machine learning methods can be adapted to handle the special challenges of that variant. [source] Bayesian classification of tumours by using gene expression dataJOURNAL OF THE ROYAL STATISTICAL SOCIETY: SERIES B (STATISTICAL METHODOLOGY), Issue 2 2005Bani K. Mallick Summary., Precise classification of tumours is critical for the diagnosis and treatment of cancer. Diagnostic pathology has traditionally relied on macroscopic and microscopic histology and tumour morphology as the basis for the classification of tumours. Current classification frameworks, however, cannot discriminate between tumours with similar histopathologic features, which vary in clinical course and in response to treatment. In recent years, there has been a move towards the use of complementary deoxyribonucleic acid microarrays for the classi-fication of tumours. These high throughput assays provide relative messenger ribonucleic acid expression measurements simultaneously for thousands of genes. A key statistical task is to perform classification via different expression patterns. Gene expression profiles may offer more information than classical morphology and may provide an alternative to classical tumour diagnosis schemes. The paper considers several Bayesian classification methods based on reproducing kernel Hilbert spaces for the analysis of microarray data. We consider the logistic likelihood as well as likelihoods related to support vector machine models. It is shown through simulation and examples that support vector machine models with multiple shrinkage parameters produce fewer misclassification errors than several existing classical methods as well as Bayesian methods based on the logistic likelihood or those involving only one shrinkage parameter. [source] Unsupervised classification methods in food sciences: discussion and outlookJOURNAL OF THE SCIENCE OF FOOD AND AGRICULTURE, Issue 7 2008Marcin Kozak Abstract This paper reviews three unsupervised multivariate classification methods: principal component analysis, principal component similarity analysis and heuristic cluster analysis. The theoretical basis of each method is presented in brief, and assumptions inherent to the methods are highlighted. A literature review shows that these methods have sometimes been used inappropriately or without referencing all essential parameters. The paper also brings to the attention of the reader a relatively unknown method: probabilistic or model-based cluster analysis. The goal of this method is to uncover the true classification of objects rather than a convenient classification provided by the other methods. For this reason it is felt that model-based cluster analysis will have broad application in the future. Copyright © 2008 Society of Chemical Industry [source] Reproducibility of the Italian ISQ method for quality classification of bread wheats: An evaluation by expert assessorsJOURNAL OF THE SCIENCE OF FOOD AND AGRICULTURE, Issue 5 2007Giorgia Foca Abstract The great variety of different bakery products in Italy has led to the development of a method, the Synthetic Index of Quality (Indice Sintetico di Qualità, ISQ), for the classification of bread wheats in different quality categories. Based on chemical and rheological properties, each wheat sample is assigned to the most suitable class by an expert assessor. In many cases this procedure is not straightforward, making the class assignation uncertain, thus leading to the possibility of controversies during the trading phase. In the present study, in order to have a quantitative estimate of the validity and reliability of this procedure, a panel composed of nine expert assessors was utilised for the repeated evaluation of 100 samples of bread wheats of various qualities. The results suggest that the proposed approach can be used both to monitor the reliability of the single assessors, and to identify samples whose class assignation is reasonably indubitable, e.g. to be used for the development of automated classification methods. Moreover, the analysis of the most uncertain assignation cases can be useful in order to enhance the ISQ classification method itself. Copyright © 2007 Society of Chemical Industry [source] Multivariate exploratory analysis of ordinal data in ecology: Pitfalls, problems and solutionsJOURNAL OF VEGETATION SCIENCE, Issue 5 2005János Podani Abstract Questions: Are ordinal data appropriately treated by multivariate methods in numerical ecology? If not, what are the most common mistakes? Which dissimilarity coefficients, ordination and classification methods are best suited to ordinal data? Should we worry about such problems at all? Methods: A new classification model family, OrdClAn (Ordinal Cluster Analysis), is suggested for hierarchical and non-hierarchical classifications from ordinal ecological data, e.g. the abundance/dominance scores that are commonly recorded in relevés. During the clustering process, the objects are grouped so as to minimize a measure calculated from the ranks of within-cluster and between-cluster distances or dissimilarities. Results and Conclusions: Evaluation of the various steps of exploratory data analysis of ordinal ecological data shows that consistency of methodology throughout the study is of primary importance. In an optimal situation, each methodological step is order invariant. This property ensures that the results are independent of changes not affecting ordinal relationships, and guarantees that no illusory precision is introduced into the analysis. However, the multivariate procedures that are most commonly applied in numerical ecology do not satisfy these requirements and are therefore not recommended. For example, it is inappropriate to analyse Braun-Blanquet abudance/dominance data by methods assuming that Euclidean distance is meaningful. The solution of all problems is that the dissimilarity coefficient should be compatible with ordinal variables and the subsequent ordination or clustering method should consider only the rank order of dissimilarities. A range of artificial data sets exemplifying different subtypes of ordinal variables, e.g. indicator values or species scores from relevés, illustrate the advocated approach. Detailed analyses of an actual phytosociological data set demonstrate the classification by OrdClAn of relevés and species and the subsequent tabular rearrangement, in a numerical study remaining within the ordinal domain from the first step to the last. [source] Advancing the diagnosis and treatment of hepatocellular carcinomaLIVER TRANSPLANTATION, Issue 4 2005J. Wallis Marsh MD We analyzed global gene expression patterns of 91 human hepatocellular carcinomas (HCCs) to define the molecular characteristics of the tumors and to test the prognostic value of the expression profiles. Unsupervised classification methods revealed two distinctive subclasses of HCC that are highly associated with patient survival. This association was validated via 5 independent supervised learning methods. We also identified the genes most strongly associated with survival by using the Cox proportional hazards survival analysis. This approach identified a limited number of genes that accurately predicted the length of survival and provides new molecular insight into the pathogenesis of HCC. Tumors from the low survival subclass have strong cell proliferation and antiapoptosis gene expression signatures. In addition, the low survival subclass displayed higher expression of genes involved in ubiquitination and histone modification, suggesting an etiological involvement of these processes in accelerating the progression of HCC. In conclusion, the biological differences identified in the HCC subclasses should provide an attractive source for the development of therapeutic targets (e.g., HIF1a) for selective treatment of HCC patients. Supplementary material for this article can be found on the HEPATOLOGY Web site (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html) Copyright 2004 American Association for the Study of Liver Diseases. Hepatology. 2004 Sep;40(3):667,76. [source] Multiple approaches to data-mining of proteomic data based on statistical and pattern classification methodsPROTEINS: STRUCTURE, FUNCTION AND BIOINFORMATICS, Issue 9 2003Jacob W. Tatay Abstract The data-mining challenge presented is composed of two fundamental problems. Problem one is the separation of forty-one subjects into two classifications based on the data produced by the mass spectrometry of protein samples from each subject. Problem two is to find the specific differences between protein expression data of two sets of subjects. In each problem, one group of subjects has a disease, while the other group is nondiseased. Each problem was approached with the intent to introduce a new and potentially useful tool to analyze protein expression from mass spectrometry data. A variety of methodologies, both conventional and nonconventional were used in the analysis of these problems. The results presented show both overlap and discrepancies. What is important is the breadth of the techniques and the future direction this analysis will create. [source] Prevalence of rheumatoid arthritis in persons 60 years of age and older in the United States: Effect of different methods of case classificationARTHRITIS & RHEUMATISM, Issue 4 2003Elizabeth K. Rasch Objective To determine prevalence estimates for rheumatoid arthritis (RA) in noninstitutionalized older adults in the US. Prevalence estimates were compared using 3 different classification methods based on current classification criteria for RA. Methods Data from the Third National Health and Nutrition Examination Survey (NHANES-III) were used to generate prevalence estimates by 3 classification methods in persons 60 years of age and older (n = 5,302). Method 1 applied the "n of k" rule, such that subjects who met 3 of 6 of the American College of Rheumatology (ACR) 1987 criteria were classified as having RA (data from hand radiographs were not available). In method 2, the ACR classification tree algorithm was applied. For method 3, medication data were used to augment case identification via method 2. Population prevalence estimates and 95% confidence intervals (95% CIs) were determined using the 3 methods on data stratified by sex, race/ethnicity, age, and education. Results Overall prevalence estimates using the 3 classification methods were 2.03% (95% CI 1.30,2.76), 2.15% (95% CI 1.43,2.87), and 2.34% (95% CI 1.66,3.02), respectively. The prevalence of RA was generally greater in the following groups: women, Mexican Americans, respondents with less education, and respondents who were 70 years of age and older. Conclusion The prevalence of RA in persons 60 years of age and older is ,2%, representing the proportion of the US elderly population who will most likely require medical intervention because of disease activity. Different classification methods yielded similar prevalence estimates, although detection of RA was enhanced by incorporation of data on use of prescription medications, an important consideration in large population surveys. [source] Trends and determinants of caesarean sections births in Queensland, 1997,2006AUSTRALIAN AND NEW ZEALAND JOURNAL OF OBSTETRICS AND GYNAECOLOGY, Issue 6 2009Stuart HOWELL Background:, The determinants of Queensland's rising caesarean section (CS) rate remain poorly understood because of the historical absence of standard classification methods. Aims:, We applied the Robson Ten Group Classification System (RTGCS) to population-based data to identify the main contributors to Queensland's rising CS rate. Method:, The RTGCS was applied retrospectively to the Queensland Perinatal Data Collection. CS rates were described for all ten RTGCS groups using data from 2006. Trends were evaluated using data for the years 1997,2006. Public and private sector patients were evaluated separately. Results:, In Queensland, in 2006, CS rates were 26.9 and 48.0% among public and private sector patients, respectively. Multiparous women with a previous caesarean birth (Group 5) made the greatest contribution to the CS rate in both sectors, followed by nulliparous women who had labour induced or were delivered by CS prior to the onset of labour (Group 2) and nulliparous women in spontaneous labour (Group 1). CS rates have risen in all RTGCS groups between 1997 and 2006. The trend was pronounced among multiparous women with a previous caesarean delivery (Group 5), among women with multiple pregnancies (Group 8) and among nulliparous women who had labour induced or were delivered by CS prior to the onset of labour (Group 2). Conclusions:, The CS rate in Queensland in 2006 was higher than in any other Australian state. The increase in Queensland's CS rates can be attributed to both the rising number of primary caesarean births and the rising number of repeat caesareans. [source] Fetal alcohol syndrome (FAS) in C57BL/6 mice detected through proteomics screening of the amniotic fluid,BIRTH DEFECTS RESEARCH, Issue 4 2008Susmita Datta Abstract BACKGROUND: Fetal Alcohol Syndrome (FAS), a severe consequence of the Fetal Alcohol Spectrum Disorders, is associated with craniofacial defects, mental retardation, and stunted growth. Previous studies in C57BL/6J and C57BL/6N mice provide evidence that alcohol-induced pathogenesis follows early changes in gene expression within specific molecular pathways in the embryonic headfold. Whereas the former (B6J) pregnancies carry a high-risk for dysmorphogenesis following maternal exposure to 2.9 g/kg alcohol (two injections spaced 4.0 h apart on gestation day 8), the latter (B6N) pregnancies carry a low-risk for malformations. The present study used this murine model to screen amniotic fluid for biomarkers that could potentially discriminate between FAS-positive and FAS-negative pregnancies. METHODS: B6J and B6N litters were treated with alcohol (exposed) or saline (control) on day 8 of gestation. Amniotic fluid aspirated on day 17 (n = 6 replicate litters per group) was subjected to trypsin digestion for analysis by matrix-assisted laser desorption,time of flight mass spectrometry with the aid of denoising algorithms, statistical testing, and classification methods. RESULTS: We identified several peaks in the proteomics screen that were reduced consistently and specifically in exposed B6J litters. Preliminary characterization by liquid chromatography tandem mass spectrometry and multidimensional protein identification mapped the reduced peaks to alpha fetoprotein (AFP). The predictive strength of AFP deficiency as a biomarker for FAS-positive litters was confirmed by area under the receiver operating characteristic curve. CONCLUSIONS: These findings in genetically susceptible mice support clinical observations in maternal serum that implicate a decrease in AFP levels following prenatal alcohol damage. Birth Defects Research (Part A), 2008. © 2008 Wiley-Liss, Inc. [source] Measurement of peripheral B cell subpopulations in common variable immunodeficiency (CVID) using a whole blood methodCLINICAL & EXPERIMENTAL IMMUNOLOGY, Issue 3 2005B. L. Ferry Summary Recent reports have described reduced populations of CD27+ memory B cells and increased percentages of undifferentiated B cells in peripheral blood of patients with common variable immunodeficiency (CVID). This work has prompted two attempts to classify CVID based on rapid flow cytometric quantification of peripheral blood memory B cells and immature B cells. Evidence to support the hypothesis that such in vitro B cell classification systems correlate with clinical subtypes of CVID is being sought. For the classification to be useful in routine diagnosis, it is important that the flow cytometric method can be used without prior separation of peripheral blood mononuclear cells (PBMC). We have examined 23 CVID patients and 24 controls, using both PBMC and whole blood, and find an excellent correlation between these methods. The reproducibility of the method was excellent. We classified the CVID patients by all three of the existing classifications, including secretion of immunoglobulin by B cells in vitro as described by Bryant, as well as the more recent flow cytometric classification methods. Only one patient changed classification as a result of using whole blood. [source] | |||||||||||||||||||