Time filter

Source Type

Waltham, MA, United States

A central challenge in high dimensional data settings in biomedical investigations involves the estimation of an optimal prediction algorithm to distinguish between different disease phenotypes. A significant complicating aspect in these analyses can be attributed to the presence of features that exhibit statistical interactions. Indeed, in several clinical investigations such as genetic studies of complex diseases, it is of interest to specifically identify such features. In this paper, we compare the performance of four commonly used classifiers (K-Nearest Neighbors, Prediction Analysis for Microarrays, Random Forests and Support Vector Machines) in settings involving high dimensional datasets including statistically interacting feature subsets. We evaluate the performance of these classifiers under conditions of varying sample size, levels of signal-to-noise ratio and strength of statistical interactions among features. We summarize two datasets from studies in diabetes and cardiovascular disease involving gene expression, metabolomics and proteomics measurements and compare results obtained using the four classifiers. Simulation studies revealed that the classifier Prediction Analysis of Microarrays had the highest classification accuracy in the absence of noise, statistical interactions and when feature distributions were multivariate Gaussian within each class. In the presence of statistical interactions, modest effect sizes and the absence of noise, Support Vector Machines achieved the best performance followed closely by Random Forests. Random Forests was optimal in settings that included both significant levels of high dimensional noise features and statistical interactions between biomarker pairs. The data applications revealed similar trends in the relative performances of each classifier. Random Forests had the highest classification accuracy among the four classifiers and was successful in incorporating interaction effects between features in the presence of noise in high dimensional datasets. Source

A quantitative proteomics workflow was implemented that provides extended plasma protein coverage by extensive protein depletion in combination with the sensitivity and breadth of analysis of twodimensional LC-MS/MS shotgun analysis. Abundant proteins were depleted by a two-stage process using IgY and Supermix depletion columns in series. Samples are then extensively fractionated by two-dimensional chromatography with fractions directly deposited onto MALDI plates. Decoupling sample fractionation from mass spectrometry facilitates a targeted MS/MS precursor selection strategy that maximizes measurement of a consistent set of peptides across experiments. Multiplexed stable isotope labeling provides quantification relative to a common reference sample and ensures an identical set of peptides measured in the set of samples (set of eight) combined in a single experiment. The more extensive protein depletion provided by the addition of the Supermix column did not compromise overall reproducibility of the measurements or the ability to reliably detect changes in protein levels between samples. The implementation of this workflow is presented for a case study aimed at generating molecular signatures for prediction of first heart attack. © 2011 American Chemical Society. Source

Pogatzki-Zahn E.M.,University Hospital of Muenster | Schnabel A.,University Hospital of Muenster | Zahn P.K.,BG Medicine
Expert Review of Neurotherapeutics | Year: 2012

Postoperative pain treatment is an important healthcare issue. However, the management of pain in patients after surgery remains insufficient. In the present review, several key areas important for postoperative pain management are discussed. New findings about efficacy and side effects of nonopioid analgesics, such as paracetamol, NSAIDs and COX-2 inhibitors, are presented and discussed in light of acute, short-term application in the perioperative period. Second, new findings about postoperative pain management in patients with preoperative pain and chronic opioid consumption are reported. Third, feasibility of the transversus abdominal plane block as a new and promising regional anesthesia technique is discussed. Finally, potential predictors, mechanisms and preventive treatment strategies of persistent chronic pain after surgery are presented. © 2012 Expert Reviews Ltd. Source

BG Medicine | Date: 2010-11-15

Biomarkers and methods are disclosed for diagnosing the risk of a myocardial infarction in an individual by measuring the levels of a set of biomarkers in a sample from an individual. A risk score is calculated for the individual by weighting the measured levels of the biomarkers. The risk score then is used to identify whether the individual is likely to experience a myocardial infarction. In addition, kits are disclosed that include a set of reagents for specifically measuring biomarker levels in a sample from an individual.

Methods are provided for diagnosing the risk of a cardiovascular event in a patient. In some embodiments, the method includes measuring the level of proprotein convertase subtilisin kexin type 9 (PCSK9) in a sample obtained from a patient and comparing the measured level of PCSK9 to a control. Also provided are methods of selecting a therapy for a patient prior to administration of the therapy. In some embodiments, the method includes measuring a PCSK9 blood concentration in a sample from a patient to determine the presence or absence of a PCSK9 blood concentration indicative of responsiveness to an inhibitor of 3-hydroxy-3-methylglutaryl-coenzyme A reductase. Further provided are methods for assessing the efficacy of a therapy being administered to a patient. In certain embodiments, the method includes detecting a change in a PCSK9 blood concentration in a sample from a patient, wherein a change in detected levels is indicative of whether the therapy is efficacious.

Discover hidden collaborations