Misclassification of smoking habits:An updated review of the literature

BACKGROUND Misclassification of smoking habits leads to underestimation of true relationships between diseases and active smoking, and overestimation of true relationships with passive smoking. Information on misclassification rates can be obtained from studies using cotinine as a marker.AIM To estimate overall misclassification rates based on a review and meta-analysis of the available evidence, and to investigate how misclassification rates depend on other factors.METHODS We searched for studies using cotinine as a marker which involved at least 200 participants and which provided information on high cotinine levels in selfreported non-, never, or ex-smokers or on low levels in self-reported smokers. We estimated overall misclassification rates weighted on sample size and investigated heterogeneity by various study characteristics. Misclassification rates were calculated for two cotinine cut points to distinguish smokers and nonsmokers, the higher cut point intended to distinguish regular smoking.RESULTS After avoiding double counting, 226 reports provided 294 results from 205 studies. A total of 115 results were from North America, 128 from Europe, 25 from Asia and 26 from other countries. A study on 6.2 million life insurance applicants was considered separately. Based on the lower cut point, true current smokers represented 4.96%(95% CI 4.32-5.60%) of reported non-smokers, 3.00%(2.45-3.54%) of reported never smokers, and 10.92%(9.23-12.61%) of reported exsmokers. As percentages of true current smokers, non-, never and ex-smokers formed, respectively, 14.50%(12.36-16.65%), 5.70%(3.20-8.20%), and 8.93%(6.57-11.29%). Reported current smokers represented 3.65%(2.84-4.45%) of true non-smokers. There was considerable heterogeneity between misclassification rates.Rates of claiming never smoking were very high in Asian women smokers, the individual studies reporting rates of 12.5%, 22.4%, 33.3%, 54.2% and 66.3%. False claims of quitting were relatively high in pregnant women, in diseased individuals who may recently have been advised to quit, and in studies considering cigarette smoking rather than any smoking. False claims of smoking were higher in younger populations. Misclassification rates were higher in more recently published studies. There was no clear evidence that rates varied by the body fluid used for the cotinine analysis, the assay method used, or whether the respondent was aware their statements would be validated by cotinine-though here many studies did not provide relevant information. There was only limited evidence that rates were lower in studies classified as being of good quality,based on the extent to which other sources of nicotine were accounted for.CONCLUSION It is important for epidemiologists to consider the possibility of bias due to misclassification of smoking habits, especially in circumstances where rates are likely to be high. The evidence of higher rates in more recent studies suggests that the extent of misclassification bias in studies relating passive smoking to smoking-related disease may have been underestimated.

Research Model of Churn Prediction Based on Customer Segmentation and Misclassification Cost in the Context of Big Data

Enterprises have vast amounts of customer behavior data in the era of big data. How to take advantage of these data to evaluate custom forfeit risks effectively is a common issue faced by enterprises. Most of traditional customer churn predicting models ignore customer segmentation and misclassification cost, which reduces the rationality of model. Dealing with these deficiencies, we established a research model of customer churn based on customer segmentation and misclassification cost. We utilized this model to analyze customer behavior data of a telecom company. The results show that this model is better than those models without customer segmentation and misclassification cost in terms of the performance, accuracy and coverage of model.

Improvement of Misclassification Rates of Classifying Objects under Box Cox Transformation and Bootstrap Approach

Discrimination and classification rules are based on different types of assumptions. Also, all most statistical methods are based on some necessary assumptions. Parametric methods are the best choice if it follows all the underlying assumptions. When assumptions are violated, parametric approaches do not provide a better solution and nonparametric techniques are preferred. After Box-Cox transformation, when assumptions are satisfied, parametric methods provide fewer misclassification rates. With this problem in mind, our concern is to compare the classification accuracy of parametric and non-parametric approaches with the aid of Box-Cox transformation and Bootstrapping. We carried Support Vector Machines (SVMs) and different discrimination and classification rules to classify objects. The attention is to critically compare the SVMs with Linear discrimination Analysis (LDA), and Quadratic discrimination Analysis (QDA) for measuring the performance of these techniques before and after Box-Cox transformation using misclassification rates. From the apparent error rates, we observe that before Box-Cox transformation, SVMs perform better than existing classification techniques, on the other hand, after Box-Cox transformation, parametric techniques provide fewer misclassification rates compared to nonparametric method. We also investigated the performances of classification techniques using the Bootstrap approach and observed that Bootstrap-based classification techniques significantly reduce the classification error rate than the usual techniques of small samples. Thus, this paper proposes to apply classification techniques under the Bootstrap approach for classifying objects in case of small sample. A real and simulated datasets application is carried out to see the performance.

Stochastic SIR Household Epidemic Model with Misclassification

In this work, we developed a theoretical framework leading to misclassification of the final size epidemic data for the stochastic SIR (Susceptible-Infective-Removed), household epidemic model, with false negative and false positive misclassification probabilities. Maximum likelihood based algorithm is then employed for its inference. We then analyzed and compared the estimates of the two dimensional model with those of the three and four dimensional models associated with misclassified final size data over arrange of theoretical parameters, local and global infection rates and corresponding proportion infected in the permissible region, away from its boundaries and misclassification probabilities. The adequacies of the three models to the final size data are examined. The four and three-dimensional models are found to outperform the two dimensional model on misclassified final size data.

Extent and predictors of grade upgrading and downgrading in an Australian cohort according to the new prostate cancer grade groupings

Object:To determine the extent and impact of upgrading and downgrading among men who underwent radical prostatectomy(RP)according to new grade groupings and to identify predictors of upgrading from biopsy grade Group Ⅰ and Ⅱ,and downgrading to grade Group I,in a community setting.Methods:Study participants included 2279 men with non-metastatic prostate cancer diagnosed 2006-2015 who underwent prostatectomy,from the multi-institutional South Australia Prostate Cancer Clinical Outcomes Collaborative registry.Extent of up-or down-grading was assessed by comparing biopsy and prostatectomy grade groupings.Risk of biochemical recurrence(BCR)with upgrading was assessed using multivariable competing risk regression.Binomial logistic regression was used to identify pre-treatment predictors of upgrading from grade Groups Ⅰ and Ⅱ,and risk group reclassification among men with low risk disease.Results:Upgrading occurred in 35%of cases,while downgrading occurred in 13%of cases.Sixty percent with grade Group I disease were upgraded following prostatectomy.Upgrading from grade Group I was associated with greater risk of BCR compared with concordant grading(Hazard ratio:3.1,95%confidence interval:1.7-6.0).Older age,higher prostate-specific antigen levels(PSA),fewer biopsy cores,higher number of positive cores and more recent diagnosis predicted upgrading from grade Group Ⅰ,while higher PSA and clinical stage predicted upgrading from grade Group Ⅱ.No clinical risk factors for reclassification were identified.Conclusion:Biopsy sampling errors may play an important role in upgrading from grade Group I.Improved clinical assessment of grade is needed to encourage greater uptake of active surveillance.

Ensemble-based active learning for class imbalance problem

In medical diagnosis, the problem of class imbalance is popular. Though there are abundant unlabeled data, it is very difficult and expensive to get labeled ones. In this paper, an ensemble-based active learning algorithm is proposed to address the class imbalance problem. The artificial data are created according to the distribution of the training dataset to make the ensemble diverse, and the random subspace re-sampling method is used to reduce the data dimension. In selecting member classifiers based on misclassification cost estimation, the minority class is assigned with higher weights for misclassification costs, while each testing sample has a variable penalty factor to induce the ensemble to correct current error. In our experiments with UCI disease datasets, instead of classification accuracy, F-value and G-means are used as the evaluation rule. Compared with other ensemble methods, our method shows best performance, and needs less labeled samples.

Improving the conduct of meta-analyses of observational studies

The author, who has published numerous meta-analyses of epidemiological studies, particularly on tobacco, comments on various aspects of their content. While such meta-analyses, even when well conducted, are more difficult to draw inferences from than are meta-analyses of clinical trials, they allow greater insight into an association than do simple qualitative reviews. This editorial starts with a discussion of some problems relating to hypothesis definition. These include the definition of the outcome, the exposure and the population to be considered, as well as the study inclusion and exclusion criteria. Under literature searching, the author argues against restriction to studies published in peer-reviewed journals, emphasising the fact that relevant data may be available from other sources. Problems of identifying studies and double counting are discussed, as are various issues in regard to data entry. The need to check published effect estimates is emphasised, and techniques to calculate estimates from material provided in the source publication are described. Once the data have been collected and an overall effect estimate obtained, tests for heterogeneity should be conducted in relation to different study characteristics. Though some meta-analysts recommend classifying studies by an overall index of study quality, the author prefers to separately investigate heterogeneity by those factors which contribute to the assessment of quality. Reasons why an association may not actually reflect a true causal relationship are also discussed, with the editorial describing techniques for investigating the relevance of confounding, and referring to problems resulting from misclassification of key variables. Misclassification of disease, exposure and confounding variables can all produce a spurious association, as can misclassification of the variable used to determine whether an individual can enter the study, and the author points to techniques to adjust for this. Issues relating to publication bias and the interpretation of "statistically significant" results are also discussed. The editorial should give the reader insight into the difficulties of producing a good meta-analysis.

Linear Dimension Reduction for Multiple Heteroscedastic Multivariate Normal Populations

For the case where all multivariate normal parameters are known, we derive a new linear dimension reduction (LDR) method to determine a low-dimensional subspace that preserves or nearly preserves the original feature-space separation of the individual populations and the Bayes probability of misclassification. We also give necessary and sufficient conditions which provide the smallest reduced dimension that essentially retains the Bayes probability of misclassification from the original full-dimensional space in the reduced space. Moreover, our new LDR procedure requires no computationally expensive optimization procedure. Finally, for the case where parameters are unknown, we devise a LDR method based on our new theorem and compare our LDR method with three competing LDR methods using Monte Carlo simulations and a parametric bootstrap based on real data.

Testing for a Zero Proportion

Tests for a proportion that may be zero are described. The setting is an environment in which there can be misclassifications or misdiagnoses, giving the possibility of nonzero counts from false positives even though no real examples may exist. Both frequentist and Bayesian tests and analyses are presented, and examples are given.

Bayesian Approach to Ranking and Selection for a Binary Measurement System

Binary measurement systems that classify parts as either pass or fail are widely used. Inspectors or inspection systems are often subject to error. The error rates are unlikely to be identical across inspectors. We propose a random effects Bayesian approach to model the error probabilities and overall conforming rate. We also introduce a feature-subset selection procedure to determine the best inspector in terms of overall classification accuracy. We provide simulation studies that demonstrate the viability of our proposed estimation ranking and subset-selection methods and apply the methods to a real data set.

An analysis of quantitative PCR reliability through replicates using the C_(t)method

There is considerable interest in quantitatively measuring nucleic acids from single cells to small populations. The most commonly employed laboratory method is the real-time polymerase chain reaction (PCR) analyzed with the crossing point or crossing threshold (Ct) method. Utilizing a multiwell plate reader we have performed hundreds of replicate reactions each at a set of initial conditions whose initial number of copies span a concentration range of ten orders of magnitude. The resultant Ct value distributions are analyzed with standard and novel statistical techniques to assess the variability/reliability of the PCR process. Our analysis supports the following conclusions. Given sufficient replicates, the mean and/or median Ct values are statistically distinguishable and can be rank ordered across ten orders of magnitude in initial template concentration. As expected, the variances in the Ct distributions grow as the number of initial copies declines to 1. We demonstrate that these variances are large enough to confound quantitative classi?cation of the initial condition at low template concentrations. The data indicate that a misclassi?cation transition is centered around 3000 initial copies of template DNA and that the transition region correlates with independent data on the thermal wear of the TAQ polymerase enzyme. We provide data that indicate that an alternative endpoint detection strategy based on the theory of well mixing and plate ?lling statistics is accurate below the mis- classi?cation transition where the real time method becomes unreliable.

Regression Analysis of Misclassified Current Status Data with Informative Observation Times

Misclassified current status data arises if each study subject can only be observed once and the observation status is determined by a diagnostic test with imperfect sensitivity and specificity.For the situation,another issue that may occur is that the observation time may be correlated with the interested failure time,which is often referred to as informative censoring or observation times.It is well-known that in the presence of informative censoring,the analysis that ignores it could yield biased or even misleading results.In this paper,the authors consider such data and propose a frailty-based inference procedure.In particular,an EM algorithm based on Poisson latent variables is developed and the asymptotic properties of the resulting estimators are established.The numerical results show that the proposed method works well in practice and an application to a set of real data is provided.

Q-learning based strategy analysis of cyber-physical systems considering unequal cost

This paper proposes a cyber security strategy for cyber-physical systems(CPS)based on Q-learning under unequal cost to obtain a more efficient and low-cost cyber security defense strategy with misclassification interference.The system loss caused by strategy selection errors in the cyber security of CPS is often considered equal.However,sometimes the cost associated with different errors in strategy selection may not always be the same due to the severity of the consequences of misclassification.Therefore,unequal costs referring to the fact that different strategy selection errors may result in different levels of system losses can significantly affect the overall performance of the strategy selection process.By introducing a weight parameter that adjusts the unequal cost associated with different types of misclassification errors,a modified Q-learning algorithm is proposed to develop a defense strategy that minimizes system loss in CPS with misclassification interference,and the objective of the algorithm is shifted towards minimizing the overall cost.Finally,simulations are conducted to compare the proposed approach with the standard Q-learning based cyber security strategy method,which assumes equal costs for all types of misclassification errors.The results demonstrate the effectiveness and feasibility of the proposed research.

On the rate of convergence for multi-category classification based on convex losses

The multi-category classification algorithms play an important role in both theory and practice of machine learning.In this paper,we consider an approach to the multi-category classification based on minimizing a convex surrogate of the nonstandard misclassification loss.We bound the excess misclassification error by the excess convex risk.We construct an adaptive procedure to search the classifier and furthermore obtain its convergence rate to the Bayes rule.

Power system transient stability assessment based on the multiple paralleled convolutional neural network and gated recurrent unit

In order to accurately evaluate power system stability in a timely manner after faults,and further improve the feature extraction ability of the model,this paper presents an improved transient stability assessment(TSA)method of CNN+GRU.This comprises a convolutional neural network(CNN)and gated recurrent unit(GRU).CNN has the feature extraction capability for a micro short-term time sequence,while GRU can extract characteristics contained in a macro long-term time sequence.The two are integrated to comprehensively extract the high-order features that are contained in a transient process.To overcome the difficulty of sample misclassification,a multiple parallel(MP)CNN+GRU,with multiple CNN+GRU connected in parallel,is created.Additionally,an improved focal loss(FL)func-tion which can implement self-adaptive adjustment according to the neural network training is introduced to guide model training.Finally,the proposed methods are verified on the IEEE 39 and 145-bus systems.The simulation results indicate that the proposed methods have better TSA performance than other existing methods.

Opinion on reconsideration of lung cancer risk from domestic radon exposure

Two extensive collaborative meta studies including 13 residential case-control studies from Europe and 7 from North America have demonstrated that domestic radon clearly poses a risk for lung cancer at exposure levels approaching those for underground miners.An excess risk for lung cancer of 0.08–0.16 per 100 Bq/m3 increase in radon concentration was obtained,with a positive trend of increased relative risk at above 200 Bq/m3.However,estimation of the cancer risk associated with domestic radon obtained by the pooling of multiple casecontrol studies has led to inaccuracies of the derived risk estimates in the low dose range,mainly due to smoking misclassification bias,uncertainties with respect to radon dosimetry,the histopathological characterization of lung cancers,as well as confounding by co-exposures with tobacco smoke and other indoor air pollutants.Together with a lack of biological plausibility,these deficiencies have a negative impact on the reliability of reported statistical relations between radon exposure and lung cancer at concentrations below 200 Bq/m3.It is vital that costly remedial actions to reduce domestic radon exposure are based on a scientifically robust risk assessment,and in view of aforementioned flaws,the hitherto proposed risk estimates need to be reconsidered.