A constrained-syntax genetic programming system for discovering classification rules: application to medical data sets

This paper proposes a new constrained-syntax genetic programming (GP) algorithm for discovering classification rules in medical data sets. The proposed GP contains several syntactic constraints to be enforced by the system using a disjunctive normal form representation, so that individuals represent valid rule sets that are easy to interpret. The GP is compared with C4.5, a well-known decision-tree-building algorithm, and with another GP that uses Boolean inputs (BGP), in five medical data sets: chest pain, Ljubljana breast cancer, dermatology, Wisconsin breast cancer, and pediatric adrenocortical tumor. For this last data set a new preprocessing step was devised for survival prediction. Computational experiments show that, overall, the GP algorithm obtained good results with respect to predictive accuracy and rule comprehensibility, by comparison with C4.5 and BGP.

Comparison of Different Classification Algorithms for Underwater Target Discrimination

Classification of underwater targets from the acoustic backscattered signals is considered here. Several different classification algorithms are tested and benchmarked not only for their performance but also to gain insight to the properties of the feature space. Results on a wideband 80-kHz acoustic backscattered data set collected for six different objects are presented in terms of the receiver operating characteristic (ROC) and robustness of the classifiers wrt reverberation.

Comparison of multiexcitation fluorescence and diffuse reflectance spectroscopy for the diagnosis of breast cancer (march 2003)

Nonmalignant (n = 36) and malignant (n = 20) tissue samples were obtained from breast cancer and breast reduction surgeries. These tissues were characterized using multiple excitation wavelength fluorescence spectroscopy and diffuse reflectance spectroscopy in the ultraviolet-visible wavelength range, immediately after excision. Spectra were then analyzed using principal component analysis (PCA) as a data reduction technique. PCA was performed on each fluorescence spectrum, as well as on the diffuse reflectance spectrum individually, to establish a set of principal components for each spectrum. A Wilcoxon rank-sum test was used to determine which principal components show statistically significant differences between malignant and nonmalignant tissues. Finally, a support vector machine (SVM) algorithm was utilized to classify the samples based on the diagnostically useful principal components. Cross-validation of this nonparametric algorithm was carried out to determine its classification accuracy in an unbiased manner. Multiexcitation fluorescence spectroscopy was successful in discriminating malignant and nonmalignant tissues, with a sensitivity and specificity of 70% and 92%, respectively. The sensitivity (30%) and specificity (78%) of diffuse reflectance spectroscopy alone was significantly lower. Combining fluorescence and diffuse reflectance spectra did not improve the classification accuracy of an algorithm based on fluorescence spectra alone. The fluorescence excitation-emission wavelengths identified as being diagnostic from the PCA-SVM algorithm suggest that the important fluorophores for breast cancer diagnosis are most likely tryptophan, NAD(P)H and flavoproteins.

Comparison of Linear and Nonlinear Classification Algorithms for the Prediction of Drug and Chemical Metabolism by Human UDP-Glucuronosyltransferase Isoforms

Partial least squares discriminant analysis (PLSDA), Bayesian regularized artificial neural network (BRANN), and support vector machine (SVM) methodologies were compared by their ability to classify substrates and nonsubstrates of 12 isoforms of human UDP-glucuronosyltransferase (UGT), an enzyme "superfamily" involved in the metabolism of drugs, nondrug xenobiotics, and endogenous compounds. Simple two-dimensional descriptors were used to capture chemical information. For each data set, 70% of the data were used for training, and the remainder were used to assess the generalization performance. In general, the SVM methodology was able to produce models with the best predictive performance, followed by BRANN and then PLSDA. However, a small number of data sets showed either equivalent or better predictability using PLSDA, which may indicate relatively linear relationships in these data sets. All SVM models showed predictive ability (>60% of test set predicted correctly) and five out of the 12 test sets showed excellent prediction (>80% prediction accuracy). These models represent the first use of pattern recognition methods to discriminate between substrates and nonsubstrates of human drug metabolizing enzymes and the first thorough assessment of three classification algorithms using multiple metabolic data sets.

The profile of women who consult alternative health practitioners in Australia

OBJECTIVES

To compare the characteristics of complementary and alternative medicine (CAM) users and non-users among Australian women.

DESIGN

Cross-sectional postal questionnaire conducted during 1996, forming the baseline survey of the Australian Longitudinal Study on Women's Health.

PARTICIPANTS

Women aged 18-23 years (n = 14 779), 45-50 years (n = 14 099) and 70-75 years (n = 12 939), randomly selected from the Health Insurance Commission database, with over-sampling of women from rural and remote areas of Australia.

MAIN OUTCOME MEASURES

Consultation with an alternative health practitioner in the 12 months before the survey.

RESULTS

Women in the mid-age cohort were more likely to have consulted an alternative health practitioner in the previous year (28%) than women in the younger cohort (19%) or older cohort (15%). In all age groups, CAM users were more likely than CAM non-users to reside in non-urban areas, to report poorer health, have more symptoms and illness, and be higher users of conventional health services.

CONCLUSIONS

Women in non-urban Australia are more likely to use CAM but do so in in parallel with conventional health services.

Bayesian Trigonometric Support Vector Classifier

This letter describes Bayesian techniques for support vector classification. In particular, we propose a novel differentiable loss function, called the trigonometric loss function, which has the desirable characteristic of natural normalization in the likelihood function, and then follow standard gaussian processes techniques to set up a Bayesian framework. In this framework, Bayesian inference is used to implement model adaptation, while keeping the merits of support vector classifier, such as sparseness and convex programming. This differs from standard gaussian processes for classification. Moreover, we put forward class probability in making predictions. Experimental results on benchmark data sets indicate the usefulness of this approach.

Dynamical components analysis of fMRI data through kernel PCA

In parallel with standard model-based methods for the analysis of fMRI data, exploratory methods--such as PCA, ICA, and clustering--have been developed to give an account of the dataset with minimal priors: no assumption is made on the data content itself, but the data structure is assumed to show some properties (decorrelation, independence) that allow for the detection of structures of interest. In this paper, we present an alternative that tries to take into account some relevant knowledge for the analysis of the dataset, e.g., the experimental paradigm, while keeping the flexibility of exploratory methods: we use a prior temporal modeling of the data that characterizes each voxel time course. Two implementations are proposed: one based on the General Linear Model, the other one on more flexible short-term predictors, whose complexity is controlled by a Minimum Description Length approach. However, our main concern here is the construction of a multivariate model; the latter is performed with the help of a kernel PCA method that builds a redundant representation of the data through the nonlinearity of the kernel. This allows for a refinement in the description of the (temporal) patterns of interest. In particular, this helps in the characterization of subtle variations in the response to different experimental conditions. We illustrate the usefulness of nonlinearity through the analysis of a synthetic dataset and show on a real dataset how it helps to interpret the experimental results.

Detection of prostate cancer by integration of line-scan diffusion, T2-mapping and T2-weighted magnetic resonance imaging; a multichannel statistical classifier

A multichannel statistical classifier for detecting prostate cancer was developed and validated by combining information from three different magnetic resonance (MR) methodologies: T2-weighted, T2-mapping, and line scan diffusion imaging (LSDI). From these MR sequences, four different sets of image intensities were obtained: T2-weighted (T2W) from T2-weighted imaging, Apparent Diffusion Coefficient (ADC) from LSDI, and proton density (PD) and T2 (T2 Map) from T2-mapping imaging. Manually segmented tumor labels from a radiologist, which were validated by biopsy results, served as tumor "ground truth." Textural features were extracted from the images using co-occurrence matrix (CM) and discrete cosine transform (DCT). Anatomical location of voxels was described by a cylindrical coordinate system. A statistical jack-knife approach was used to evaluate our classifiers. Single-channel maximum likelihood (ML) classifiers were based on 1 of the 4 basic image intensities. Our multichannel classifiers: support vector machine (SVM) and Fisher linear discriminant (FLD), utilized five different sets of derived features. Each classifier generated a summary statistical map that indicated tumor likelihood in the peripheral zone (PZ) of the prostate gland. To assess classifier accuracy, the average areas under the receiver operator characteristic (ROC) curves over all subjects were compared. Our best FLD classifier achieved an average ROC area of 0.839(+/-0.064), and our best SVM classifier achieved an average ROC area of 0.761(+/-0.043). The T2W ML classifier, our best single-channel classifier, only achieved an average ROC area of 0.599(+/-0.146). Compared to the best single-channel ML classifier, our best multichannel FLD and SVM classifiers have statistically superior ROC performance (P=0.0003 and 0.0017, respectively) from pairwise two-sided t-test. By integrating the information from multiple images and capturing the textural and anatomical features in tumor areas, summary statistical maps can potentially aid in image-guided prostate biopsy and assist in guiding and controlling delivery of localized therapy under image guidance.

Bayesian networks for knowledge discovery in large datasets: basics for nurse researchers

The growth of nursing databases necessitates new approaches to data analyses. These databases, which are known to be massive and multidimensional, easily exceed the capabilities of both human cognition and traditional analytical approaches. One innovative approach, knowledge discovery in large databases (KDD), allows investigators to analyze very large data sets more comprehensively in an automatic or a semi-automatic manner. Among KDD techniques, Bayesian networks, a state-of-the art representation of probabilistic knowledge by a graphical diagram, has emerged in recent years as essential for pattern recognition and classification in the healthcare field. Unlike some data mining techniques, Bayesian networks allow investigators to combine domain knowledge with statistical data, enabling nurse researchers to incorporate clinical and theoretical knowledge into the process of knowledge discovery in large datasets. This tailored discussion presents the basic concepts of Bayesian networks and their use as knowledge discovery tools for nurse researchers.

Minimum description length block finder, a method to identify haplotype blocks and to compare the strength of block boundaries

We describe a new probabilistic method for finding haplotype blocks that is based on the use of the minimum description length (MDL) principle. We give a rigorous definition of the quality of a segmentation of a genomic region into blocks and describe a dynamic programming algorithm for finding the optimal segmentation with respect to this measure. We also describe a method for finding the probability of a block boundary for each pair of adjacent markers: this gives a tool for evaluating the significance of each block boundary. We have applied the method to the published data of Daly and colleagues. The results expose some problems that exist in the current methods for the evaluation of the significance of predicted block boundaries. Our method, MDL block finder, can be used to compare block borders in different sample sets, and we demonstrate this by applying the MDL-based method to define the block structure in chromosomes from population isolates.

Kernel-Based Nonlinear Blind Source Separation

We propose kTDSEP, a kernel-based algorithm for nonlinear blind source separation (BSS). It combines complementary research fields: kernel feature spaces and BSS using temporal information. This yields an efficient algorithm for nonlinear BSS with invertible nonlinearity. Key assumptions are that the kernel feature space is chosen rich enough to approximate the nonlinearity and that signals of interest contain temporal information. Both assumptions are fulfilled for a wide set of real-world applications. The algorithm works as follows: First, the data are (implicitly) mapped to a high (possibly infinite)—dimensional kernel feature space. In practice, however, the data form a smaller submanifold in feature space—even smaller than the number of training data points—a fact that has already been used by, for example, reduced set techniques for support vector machines. We propose to adapt to this effective dimension as a preprocessing step and to construct an orthonormal basis of this submanifold. The latter dimension-reduction step is essential for making the subsequent application of BSS methods computationally and numerically tractable. In the reduced space, we use a BSS algorithm that is based on second-order temporal decorrelation. Finally, we propose a selection procedure to obtain the original sources from the extracted nonlinear components automatically.

Experiments demonstrate the excellent performance and efficiency of our kTDSEP algorithm for several problems of nonlinear BSS and for more than two sources.

Classification of rheumatoid joint inflammation based on laser imaging

We describe a classification system for a novel imaging method for arthritic finger joints. The basis of this system is a laser imaging technique which is sensitive to the optical characteristics of finger joint tissue. From the laser images acquired at baseline and follow-up, finger joints can automatically be classified according to whether the inflammatory status has improved or worsened. To perform the classification task, various linear and kernel-based systems were implemented and their performances were compared. Based on the results presented in this paper, we conclude that the laser-based imaging permits a reliable classification of pathological finger joints, making it a sensitive method for detecting arthritic changes.