D. Coomans

Alternative k-nearest neighbour rules in supervised pattern recognition : Part 1. k-Nearest neighbour classification by using alternative voting rules

This paper discusses an extension of the well known k-nearest neighbour method. The majority voting procedure is replaced by an alternative voting method. This alternative kNN method is approached from both probabilistic and non-probabilistic points of view. On the basis of an example of differentiation between EU thyroid function and HYPER thyroid function, it is shown that alternative votes can give rise to better classification results than the majority vote.

Use of a microcomputer for the definition of multivariate confidence regions in medical diagnosis based on clinical laboratory profiles

The use of multivariate confidence regions is proposed for the classification of clinical laboratory profiles into diagnostic classes. For this purpose, a multivariate confidence region is developed for each diagnostic class. Three methods (UNEQ, EQ, and SIMCA) are evaluated and compared with classical linear discriminant analysis. As an example, a small data set concerning the differentiation of the thyroid functional states on the basis of five laboratory tests is used. It is shown that related procedures can produce results of very different quality and that the multivariate region approach is attractive for the clinicians daily practice since the methods are easily implemented on a microcomputer.

High-throughput 16S rRNA gene sequencing reveals alterations of intestinal microbiota in myalgic encephalomyelitis/chronic fatigue syndrome patients.

Human intestinal microbiota plays an important role in the maintenance of host health by providing energy, nutrients, and immunological protection. Intestinal dysfunction is a frequent complaint in myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) patients, and previous reports suggest that dysbiosis, i.e. the overgrowth of abnormal populations of bacteria in the gut, is linked to the pathogenesis of the disease. We used high-throughput 16S rRNA gene sequencing to investigate the presence of specific alterations in the gut microbiota of ME/CFS patients from Belgium and Norway. 43 ME/CFS patients and 36 healthy controls were included in the study. Bacterial DNA was extracted from stool samples, PCR amplification was performed on 16S rRNA gene regions, and PCR amplicons were sequenced using Roche FLX 454 sequencer. The composition of the gut microbiota was found to differ between Belgian controls and Norwegian controls: Norwegians showed higher percentages of specific Firmicutes populations (Roseburia, Holdemania) and lower proportions of most Bacteroidetes genera. A highly significant separation could be achieved between Norwegian controls and Norwegian patients: patients presented increased proportions of Lactonifactor and Alistipes, as well as a decrease in several Firmicutes populations. In Belgian subjects the patient/control separation was less pronounced, however some abnormalities observed in Norwegian patients were also found in Belgian patients. These results show that intestinal microbiota is altered in ME/CFS. High-throughput sequencing is a useful tool to diagnose dysbiosis in patients and could help designing treatments based on gut microbiota modulation (antibiotics, prexa0and probiotics supplementation).

Alternative k-nearest neighbour rules in supervised pattern recognition : Part 2. Probabilistic classification on the basis of the kNN method modified for direct density estimation

Abstract An implementation of the k NN classification method of Loftsgaarden and Quesenberry is discussed. It is a theoretically valid alternative for probabilistic pattern recognition by means of the Bayes equation. The method is illustrated on the basis of a data set containing laboratory tests for three categories of functional states of the thyroid gland and the performance is compared with those of other probabilistic pattern recognition techniques. Although the results are acceptable, the k NN method did not perform as well as some other probabilistic techniques.

Optimization by statistical linear discriminant analysis in analytical chemistry

Abstract The application of statistical linear discriminant analysis in analytical chemistry is discussed. In addition to a general discussion of the theory of the method, which is illustrated by some examples, its suitability for problem solving in analytical chemistry is demonstrated by a review of published applications. A more mathematical point of view is added as an appendix.

The application of linear discriminant analysis in the diagnosis of thyroid diseases

Abstract The effectiveness of five in-vitro laboratory tests for differentiation between three thyroid functional states (EU, HYPO and HYPER thyroidism) has been determined by using statistical linear discriminant analysis. The optimal linear combination of laboratory tests obtained by means of linear discriminant analysis results in a better use of the information present in each test, so that the possible redundancy of tests can be assessed. In this context, some feature selection criteria were evaluated. It is shown that in this application only two laboratory tests are necessary to obtain a sufficiently high diagnostic effectiveness when linear discriminant analysis is applied.

The use of principal components analysis for the investigation of an organic air pollutants data set

Abstract Principal components analysis (PCA) was used for the investigation of an air pollutants data base. The data set consists of nearly 400 measurements of 26 gaseous organic compounds and meteorological data. The measurements were carried out at four different places in The Netherlands. PCA is a simple way to display visually most of the total variation in a few dimensions. It is helpful in the identification of outliers, the recognition of sources and the investigation of meteorological effects.

Effect of scaling on class modeling with the simca method

Abstract The first step in multivariate analysis is almost always the scaling of the variables. The pattern recognition technique SIMCA provides the possibility of scaling the variables over all the objects of the training set (classical scaling), or only over the objects belonging to the same group (separate scaling). The former method of scaling is the more used. The effect of separate scaling on the classification of objects with SIMCA is investigated for a data set consisting of the percentage distribution of fatty acids in olive oils originating from two neighbouring regions in Italy. It is shown that separate scaling has a beneficial effect on the classification.

Potential methods in pattern recognition : Part 1. Classification aspects of the supervised method ALLOC

Abstract This paper on the application of potential functions in pattern recognition introduces the software package ALLOC to analytical chemistry, emphasizing the methodology of classifying objects. ALLOC is compared with other classification techniques on the basis of two data sets and is shown to perform very well.

Potential methods in pattern recognition : Part 2. CLUPOT —an unsupervised pattern recognition technique

Abstract The applicability of potential functions in unsupervised pattern recognition is demonstrated on the basis of a new clustering technique called CLUPOT. CLUPOT is a centrotype sorting technique which means that for each detected cluster of objects a representative object can be selected. CLUPOT uses a reliability curve which permits the detection of significant clusters. Applications to four data sets (Kowalskis archeological artefact data, Ruspinis fuzzy set data. Fishers Iris data and a part of Esbensens meteorite data) show that CLUPOT yields significant clusterings.