Screening for breast cancer.

Support vector machine recursive feature elimination (SVM-RFE) is a powerful feature selection algorithm. However, when the candidate feature set contains highly correlated features, the ranking criterion of SVM-RFE will be biased, which would hinder the application of SVM-RFE on gas sensor data. In this paper, the linear and nonlinear SVM-RFE algorithms are studied. After investigating the correlation bias, an improved algorithm SVM-RFE + CBR is proposed by incorporating the correlation bias reduction (CBR) strategy into the feature elimination procedure. Experiments are conducted on a synthetic dataset and two breath analysis datasets, one of which contains temperature modulated sensors. Large and comprehensive sets of transient features are extracted from the sensor responses. The classification accuracy with feature selection proves the efficacy of the proposed SVM-RFE + CBR. It outperforms the original SVM-RFE and other typical algorithms. An ensemble method is further studied to improve the stability of the proposed method. By statistically analyzing the features’ rankings, some knowledge is obtained, which can guide future design of e-noses and feature extraction algorithms.

Feature selection and analysis on correlated gas sensor data with recursive feature elimination

In the past few years, all manner of storage applications, ranging from disk array systems to distributed and wide-area systems, have started to grapple with the reality of tolerating multiple simultaneous failures of storage nodes. Unlike the single failure case, which is optimally handled with RAID Level-5 parity, the multiple failure case is more difficult because optimal general purpose strategies are not yet known. Erasure Coding is the field of research that deals with these strategies, and this field has blossomed in recent years. Despite this research, the decades-old Reed- Solomon erasure code remains the only space-optimal (MDS) code for all but the smallest storage systems. The best performing implementations of Reed-Solomon coding employ a variant called Cauchy Reed-Solomon coding, developed in the mid 1990s [4]. In this paper, we present an improvement to Cauchy Reed-Solomon coding that is based on optimizing the Cauchy distribution matrix. We detail an algorithm for generating good matrices and then evaluate the performance of encoding using all implementations Reed- Solomon codes, plus the best MDS codes from the literature. The improvements over the original Cauchy Reed-Solomon codes are as much as 83% in realistic scenarios, and average roughly 10% over all cases that we tested.

/pdf/optimizing-cauchy-reed-solomon-codes-for-fault-tolerant-59wsbbdioh.pdf

Optimizing Cauchy Reed-Solomon Codes for Fault-Tolerant Network Storage Applications

A deep feature based framework for breast masses classification

Background
Many systematic reviews exist on interventions to improve safe and effective medicines use by consumers, but research is distributed across diseases, populations and settings. The scope and focus of such reviews also vary widely, creating challenges for decision-makers seeking to inform decisions by using the evidence on consumers’ medicines use.

This is an update of a 2011 overview of systematic reviews, which synthesises the evidence, irrespective of disease, medicine type, population or setting, on the effectiveness of interventions to improve consumers' medicines use.

Objectives
To assess the effects of interventions which target healthcare consumers to promote safe and effective medicines use, by synthesising review-level evidence.

Methods
Search methods: We included systematic reviews published on the Cochrane Database of Systematic Reviews and the Database of Abstracts of Reviews of Effects. We identified relevant reviews by handsearching databases from their start dates to March 2012.

Selection criteria: We screened and ranked reviews based on relevance to consumers’ medicines use, using criteria developed for this overview.

Data collection and analysis: We used standardised forms to extract data, and assessed reviews for methodological quality using the AMSTAR tool. We used standardised language to summarise results within and across reviews; and gave bottom-line statements about intervention effectiveness. Two review authors screened and selected reviews, and extracted and analysed data. We used a taxonomy of interventions to categorise reviews and guide syntheses.

Main results
We included 75 systematic reviews of varied methodological quality. Reviews assessed interventions with diverse aims including support for behaviour change, risk minimisation and skills acquisition. No reviews aimed to promote systems-level consumer participation in medicines-related activities. Medicines adherence was the most frequently-reported outcome, but others such as knowledge, clinical and service-use outcomes were also reported. Adverse events were less commonly identified, while those associated with the interventions themselves, or costs, were rarely reported.

Looking across reviews, for most outcomes, medicines self-monitoring and self-management programmes appear generally effective to improve medicines use, adherence, adverse events and clinical outcomes; and to reduce mortality in people self-managing antithrombotic therapy. However, some participants were unable to complete these interventions, suggesting they may not be suitable for everyone.

Other promising interventions to improve adherence and other key medicines-use outcomes, which require further investigation to be more certain of their effects, include:

· simplified dosing regimens: with positive effects on adherence;

· interventions involving pharmacists in medicines management, such as medicines reviews (with positive effects on adherence and use, medicines problems and clinical outcomes) and pharmaceutical care services (consultation between pharmacist and patient to resolve medicines problems, develop a care plan and provide follow-up; with positive effects on adherence and knowledge).

Several other strategies showed some positive effects, particularly relating to adherence, and other outcomes, but their effects were less consistent overall and so need further study. These included:

· delayed antibiotic prescriptions: effective to decrease antibiotic use but with mixed effects on clinical outcomes, adverse effects and satisfaction;

· practical strategies like reminders, cues and/or organisers, reminder packaging and material incentives: with positive, although somewhat mixed effects on adherence;

· education delivered with self-management skills training, counselling, support, training or enhanced follow-up; information and counselling delivered together; or education/information as part of pharmacist-delivered packages of care: with positive effects on adherence, medicines use, clinical outcomes and knowledge, but with mixed effects in some studies;

· financial incentives: with positive, but mixed, effects on adherence.

Several strategies also showed promise in promoting immunisation uptake, but require further study to be more certain of their effects. These included organisational interventions; reminders and recall; financial incentives; home visits; free vaccination; lay health worker interventions; and facilitators working with physicians to promote immunisation uptake. Education and/or information strategies also showed some positive but even less consistent effects on immunisation uptake, and need further assessment of effectiveness and investigation of heterogeneity.

There are many different potential pathways through which consumers' use of medicines could be targeted to improve outcomes, and simple interventions may be as effective as complex strategies. However, no single intervention assessed was effective to improve all medicines-use outcomes across all diseases, medicines, populations or settings.

Even where interventions showed promise, the assembled evidence often only provided part of the picture: for example, simplified dosing regimens seem effective for improving adherence, but there is not yet sufficient information to identify an optimal regimen.

In some instances interventions appear ineffective: for example, the evidence suggests that directly observed therapy may be generally ineffective for improving treatment completion, adherence or clinical outcomes.

In other cases, interventions may have variable effects across outcomes. As an example, strategies providing information or education as single interventions appear ineffective to improve medicines adherence or clinical outcomes, but may be effective to improve knowledge; an important outcome for promoting consumers' informed medicines choices.

Despite a doubling in the number of reviews included in this updated overview, uncertainty still exists about the effectiveness of many interventions, and the evidence on what works remains sparse for several populations, including children and young people, carers, and people with multimorbidity.

Authors' conclusions
This overview presents evidence from 75 reviews that have synthesised trials and other studies evaluating the effects of interventions to improve consumers' medicines use.

Systematically assembling the evidence across reviews allows identification of effective or promising interventions to improve consumers’ medicines use, as well as those for which the evidence indicates ineffectiveness or uncertainty.

Decision makers faced with implementing interventions to improve consumers' medicines use can use this overview to inform decisions about which interventions may be most promising to improve particular outcomes. The intervention taxonomy may also assist people to consider the strategies available in relation to specific purposes, for example, gaining skills or being involved in decision making. Researchers and funders can use this overview to identify where more research is needed and assess its priority. The limitations of the available literature due to the lack of evidence for important outcomes and important populations, such as people with multimorbidity, should also be considered in practice and policy decisions.

/pdf/interventions-to-improve-safe-and-effective-medicines-use-by-4v83thdzsz.pdf

Interventions to improve safe and effective medicines use by consumers: an overview of systematic reviews.

List of Figures. List of Tables. Preface. 1: Digital Communication. 1. Basics. 2. Algorithms and Complexity. 3. Encoding and Decoding. 4. Bounds. 5. Overview of the Text. 2: Abstract Algebra. 1. Sets and Groups. 2. Rings, Domains, and Fields. 3. Vector Spaces and GF(pm). 4. Polynomials over Galois Fields. 5. Frequency Domain Analysis of Polynomials over GF(q) [x]/(xn-1). Linear Block Codes. 1. Basic Structure of Linear Codes. 2. Repetition and Parity Check Codes. 3. Hamming Codes. 4. Reed-Muller Codes. 5. Cyclic Codes. 6. Quadratic Residue Codes. 7. Golay Codes. 8. BCH and Reed-Solomon Codes. 4: Convolutional and Concatenated Codes. 1. Convolutional Encoders. 2. Analysis of Component Codes. 3. Concatenated Codes. 4. Analysis of Parallel Concatenated Codes. 5: Elements of Graph Theory. 1. Introduction. 2. Martingales. 3. Expansion. 6: Algorithms on Graphs. 1. Probability Models and Bayesian Networks. 2. Belief Propagation Algorithm. 3. Junction Tree Propagation Algorithm. 4. Message Passing and Error Control Decoding. 5. Message Passing in Loops. 7: Turbo Decoding. 1. Turbo Decoding. 2. Parallel Decoding. 3. Notes. 8: Low-Density Parity-Check Codes. 1. Basic Properties. 2. Simple Decoding Algorithms. 3. Explicit Construction. 4. Gallager's Decoding Algorithms. 5. Belief Propagation Decoding. 6. Notes. 9: Low-Density Generator Codes. 1. Introduction. 2. Decoding Analyses. 3. Good Degree Sequences. 4. Irregular Repeat-Accumulate Codes. 5. Cascaded Codes. 6. Notes. References. Index.

Fundamentals of Codes, Graphs, and Iterative Decoding

In this letter, we investigate an efficient Gaussian elimination decoding scheme of Raptor codes used over the binary erasure channel. It will be shown that the proposed incremental Gaussian elimination decoding significantly improves on the decoding time over the usual Gaussian elimination decoding while maintaining the same decoding performance.

Incremental Gaussian elimination decoding of raptor codes over BEC

In this letter, we propose an efficient algorithm for maximum-likelihood decoding of Raptor codes used over the binary erasure channel. The algorithm is inspired by the decoding scheme suggested in the 3GPP multimedia broadcast/multicast services standard and is an improved version of it.

An efficient algorithm for ML decoding of raptor codes over the binary erasure channel

Mutual information-based SVM-RFE for diagnostic classification of digitized mammograms

Digital mammography is one of the most promising options to diagnose breast cancer which is the most common cancer in women. However, its effectiveness is enfeebled due to the difficulty in distinguishing actual cancer lesions from benign abnormalities, which results in unnecessary biopsy referrals. To overcome this issue, computer aided diagnosis (CADx) using machine learning techniques have been studied worldwide. Since this is a classification problem and the number of features obtainable from a mammogram image is infinite, a feature selection method that is tailored for use in the CADx systems is needed. We propose a feature selection method based on multiple support vector machine recursive feature elimination (MSVM-RFE). We compared our method with four previously proposed feature selection methods which use support vector machine as the base classifier. Experiments were performed on lesions extracted from the Digital Database of Screening Mammography, the largest public digital mammography database available. We measured average accuracy over 5-fold cross validation on the 8 datasets we extracted. Selecting from 8 features, conventional algorithms like SVM-RFE and multiple SVM-RFE showed slightly better performance than others. However, when selecting from 22 features, our proposed modified multiple SVM-RFE using boosting outperformed or was at least competitive to all others. Our modified method may be a possible alternative to SVM-RFE or the original MSVM-RFE in many cases of interest. In the future, we need a specific method to effectively combine models trained during the feature selection process and a way to combine feature subsets generated from individual SVM-RFE instances.

Saejoon Kim

Papers

Fundamentals of Codes, Graphs, and Iterative Decoding

Incremental Gaussian elimination decoding of raptor codes over BEC

An efficient algorithm for ML decoding of raptor codes over the binary erasure channel

Mutual information-based SVM-RFE for diagnostic classification of digitized mammograms

AdaBoost-based multiple SVM-RFE for classification of mammograms in DDSM