Classification Using Generalized Partial Least Squares

The advances in computational biology have made simultaneous monitoring of thousands of features possible. The high throughput technologies not only bring about a much richer information context in which to study various aspects of gene functions but they also present challenge of analyzing data with large number of covariates and few samples. As an integral part of machine learning, classification of samples into two or more categories is almost always of interest to scientists. In this paper, we address the question of classification in this setting by extending partial least squares (PLS), a popular dimension reduction tool in chemometrics, in the context of generalized linear regression based on a previous approach, Iteratively ReWeighted Partial Least Squares, i.e. IRWPLS (Marx, 1996). We compare our results with two-stage PLS (Nguyen and Rocke, 2002A; Nguyen and Rocke, 2002B) and other classifiers. We show that by phrasing the problem in a generalized linear model setting and by applying bias correction to the likelihood to avoid (quasi)separation, we often get lower classification error rates.

Cadaver dogs--a study on detection of contaminated carpet squares

INTRODUCTION: Cadaver dogs are known as valuable forensic tools in crime scene investigations. Scientific research attempting to verify their value is largely lacking, specifically for scents associated with the early postmortem interval. The aim of our investigation was the comparative evaluation of the reliability, accuracy, and specificity of three cadaver dogs belonging to the Hamburg State Police in the detection of scents during the early postmortem interval. MATERIAL AND METHODS: Carpet squares were used as an odor transporting media after they had been contaminated with the scent of two recently deceased bodies (PMI<3h). The contamination occurred for 2 min as well as 10 min without any direct contact between the carpet and the corpse. Comparative searches by the dogs were performed over a time period of 65 days (10 min contamination) and 35 days (2 min contamination). RESULTS: The results of this study indicate that the well-trained cadaver dog is an outstanding tool for crime scene investigation displaying excellent sensitivity (75-100), specificity (91-100), and having a positive predictive value (90-100), negative predictive value (90-100) as well as accuracy (92-100).

Magic angle spinning magnetic resonance: a novel method opening up translational research into NAFLD?

NAFLD (non-alcoholic fatty liver disease) and NASH (non-alcoholic steatohepatitis) are of increasing importance, both in connection with insulin resistance and with the development of liver cirrhosis. Histological samples are still the 'gold standard' for diagnosis; however, because of the risks of a liver biopsy, non-invasive methods are needed. MAS (magic angle spinning) is a special type of NMR which allows characterization of intact excised tissue without need for additional extraction steps. Because clinical MRI (magnetic resonance imaging) and MRS (magnetic resonance spectroscopy) are based on the same physical principle as NMR, translational research is feasible from excised tissue to non-invasive examinations in humans. In the present issue of Clinical Science, Cobbold and co-workers report a study in three animal strains suffering from different degrees of NAFLD showing that MAS results are able to distinguish controls, fatty infiltration and steatohepatitis in cohorts. In vivo MRS methods in humans are not obtainable at the same spectral resolution; however, know-how from MAS studies may help to identify characteristic changes in crowded regions of the magnetic resonance spectrum.

From "magic bullets" to specific cancer immunotherapy

The immune system is able to specifically target antigen-expressing cancer cells. The promise of immunotherapy was to eliminate cancer cells without harming normal tissue and, therefore, with no or very few side effects. Immunotherapy approaches have, for several decades, been tested against several tumours, most often against malignant melanoma. However, although detectable immune responses have regularly been induced, the clinical outcome has often been disappointing. The development of molecular methods and an improved understanding of tumour immunosurveillance led to novel immunotherapy approaches in the last few years. First randomised phase III trials proved that immunotherapy can prolong survival of patients with metastatic melanoma or prostate cancer. The development in the field is very rapid and various molecules (mainly monoclonal antibodies) that activate the immune system are currently being tested in clinical trials and will possibly change our treatment of cancer. The ultimate goal of any cancer therapy and also immunotherapy is to cure cancer. However, this depends on the elimination of the disease originating cancer stem cells. Unfortunately, cancer stem cells seem resistant to most available treatment options. Recent developments in immunotherapy may allow targeting these cancer stem cells specifically in the future. In this review, we summarise the current state of immunotherapy in clinical routine and the expected developments in the near future.

Comparison of partial least squares regression and principal component regression for pelvic shape prediction.

This paper studied two different regression techniques for pelvic shape prediction, i.e., the partial least square regression (PLSR) and the principal component regression (PCR). Three different predictors such as surface landmarks, morphological parameters, or surface models of neighboring structures were used in a cross-validation study to predict the pelvic shape. Results obtained from applying these two different regression techniques were compared to the population mean model. In almost all the prediction experiments, both regression techniques unanimously generated better results than the population mean model, while the difference on prediction accuracy between these two regression methods is not statistically significant (α=0.01).