Misclassification Error Research Articles

Validating the indicator "maternal death review coverage" to improve maternal mortality data: A retrospective analysis of district, facility, and individual medical record data.

Understanding causes and contributors to maternal mortality is critical from a quality improvement perspective to inform decision making and monitor progress toward ending preventable maternal mortality. The indicator "maternal death review coverage" is defined as the percentage of maternal deaths occurring in a facility that are audited. Both the numerator and denominator of this indicator are subject to misclassification errors, underreporting, and bias. This study assessed the validity of the indicator by examining both its numerator-the number and quality of death reviews-and denominator-the number of facility-based maternal deaths and comparing estimates of the indicator obtained from facility- versus district-level data. We collected data on the number of maternal deaths and content of death reviews from all health facilities serving as birthing sites in 12 districts in three countries: Argentina, Ghana, and India. Additional data were extracted from health management information systems on the number and dates of maternal deaths and maternal death reviews reported from health facilities to the district-level. We tabulated the percentage of facility deaths with evidence of a review, the percentage of reviews that met the World Health Organization defined standard for maternal and perinatal death surveillance and response. Results were stratified by sociodemographic characteristics of women and facility location and type. We compared these estimates to that obtained using district-level data. and looked at evidence of the review at the district/provincial level. Study teams reviewed facility records at 34 facilities in Argentina, 51 facilities in Ghana, and 282 facilities in India. In total, we found 17 deaths in Argentina, 14 deaths in Ghana, and 58 deaths in India evidenced at facilities. Overall, >80% of deaths had evidence of a review at facilities. In India, a much lower percentage of deaths occurring at secondary-level facilities (61.1%) had evidence of a review compared to deaths in tertiary-level facilities (92.1%). In all three countries, only about half of deaths in each country had complete reviews: 58.8% (n = 10) in Argentina, 57.2% (n = 8) in Ghana, and 41.1% (n = 24) in India. Dramatic reductions in indicator value were seen in several subnational geographic areas, including Gonda and Meerut in India and Sunyani in Ghana. For example, in Gonda only three of the 18 reviews conducted at facilities met the definitional standard (16.7%), which caused the value of the indicator to decrease from 81.8% to 13.6%. Stratification by women's sociodemographic factors suggested systematic differences in completeness of reviews by women's age, place of residence, and timing of death. Our study assessed the validity of an important indicator for ending preventable deaths: the coverage of reviews of maternal deaths occurring in facilities in three study settings. We found discrepancies in deaths recorded at facilities and those reported to districts from facilities. Further, few maternal death reviews met global quality standards for completeness. The value of the calculated indicator masked inaccuracies in counts of both deaths and reviews and gave no indication of completeness, thus undermining the ultimate utility of the measure in achieving an accurate measure of coverage.

PermDroid a framework developed using proposed feature selection approach and machine learning techniques for Android malware detection

The challenge of developing an Android malware detection framework that can identify malware in real-world apps is difficult for academicians and researchers. The vulnerability lies in the permission model of Android. Therefore, it has attracted the attention of various researchers to develop an Android malware detection model using permission or a set of permissions. Academicians and researchers have used all extracted features in previous studies, resulting in overburdening while creating malware detection models. But, the effectiveness of the machine learning model depends on the relevant features, which help in reducing the value of misclassification errors and have excellent discriminative power. A feature selection framework is proposed in this research paper that helps in selecting the relevant features. In the first stage of the proposed framework, t-test, and univariate logistic regression are implemented on our collected feature data set to classify their capacity for detecting malware. Multivariate linear regression stepwise forward selection and correlation analysis are implemented in the second stage to evaluate the correctness of the features selected in the first stage. Furthermore, the resulting features are used as input in the development of malware detection models using three ensemble methods and a neural network with six different machine-learning algorithms. The developed models’ performance is compared using two performance parameters: F-measure and Accuracy. The experiment is performed by using half a million different Android apps. The empirical findings reveal that malware detection model developed using features selected by implementing proposed feature selection framework achieved higher detection rate as compared to the model developed using all extracted features data set. Further, when compared to previously developed frameworks or methodologies, the experimental results indicates that model developed in this study achieved an accuracy of 98.8%.

Misclassification of causes of death among a small all-autopsied group of former nuclear workers: Death certificates vs. autopsy reports.

The U.S. Transuranium and Uranium Registries performs autopsies on each of its deceased Registrants as a part of its mission to follow up occupationally-exposed individuals. This provides a unique opportunity to explore death certificate misclassification errors, and the factors that influence them, among this small population of former nuclear workers. Underlying causes of death from death certificates and autopsy reports were coded using the 10th revision of the International Classification of Diseases (ICD-10). These codes were then used to quantify misclassification rates among 268 individuals for whom both full autopsy reports and death certificates with legible underlying causes of death were available. When underlying causes of death were compared between death certificates and autopsy reports, death certificates correctly identified the underlying cause of death's ICD-10 disease chapter in 74.6% of cases. The remaining 25.4% of misclassified cases resulted in over-classification rates that ranged from 1.2% for external causes of mortality to 12.2% for circulatory disease, and under-classification rates that ranged from 7.7% for external causes of mortality to 47.4% for respiratory disease. Neoplasms had generally lower misclassification rates with 4.3% over-classification and 13.3% under-classification. A logistic regression revealed that the odds of a match were 2.8 times higher when clinical history was mentioned on the autopsy report than when it was not. Similarly, the odds of a match were 3.4 times higher when death certificates were completed using autopsy findings than when autopsy findings were not used. This analysis excluded cases where it could not be determined if autopsy findings were used to complete death certificates. The findings of this study are useful to investigate the impact of death certificate misclassification errors on radiation risk estimates and, therefore, improve the reliability of epidemiological studies.

Effect of fluoridated water on invasive NHS dental treatments for adults: the LOTUS retrospective cohort study and economic evaluation.

Most water fluoridation studies were conducted on children before the widespread introduction of fluoride toothpastes. There is a lack of evidence that can be applied to contemporary populations, particularly adolescents and adults. To pragmatically assess the clinical and cost effectiveness of water fluoridation for preventing dental treatment and improving oral health in a contemporary population of adults, using a natural experiment design. Retrospective cohort study using routinely collected National Health Service dental claims (FP17) data. National Health Service primary dental care: general dental practices, prisons, community dental services, domiciliary settings, urgent/out-of-hours and specialised referral-only services. Dental patients aged 12 years and over living in England (n = 6,370,280). Individuals exposed to drinking water with a fluoride concentration ≥ 0.7 mg F/l between 2010 and 2020 were matched to non-exposed individuals on key characteristics using propensity scores. Primary: number of National Health Service invasive dental treatments (restorations/'fillings' and extractions) received per person between 2010 and 2020. Secondary: decayed, missing and filled teeth, missing teeth, inequalities, cost effectiveness and return on investment. National Health Service Business Services Authority dental claims data. Water quality monitoring data. Predicted mean number of invasive dental treatments was 3% lower in the optimally fluoridated group than in the sub/non-optimally fluoridated group (incidence rate ratio 0.969, 95% CI 0.967 to 0.971), a difference of -0.173 invasive dental treatments (95% CI -0.185 to -0.161). This magnitude of effect is smaller than what most stakeholders we engaged with (n = 50/54) considered meaningful. Mean decayed, missing and filled teeth were 2% lower in the optimally fluoridated group, with a difference of -0.212 decayed, missing and filled teeth (95% CI -0.229 to -0.194). There was no statistically significant difference in the mean number of missing teeth per person (0.006, 95% CI -0.008 to 0.021). There was no compelling evidence that water fluoridation reduced social inequalities in treatments received or missing teeth; however, decayed, missing and filled teeth data did not demonstrate a typical inequalities gradient. Optimal water fluoridation in England in 2010-20 was estimated to cost £10.30 per person (excluding original setup costs). Mean National Health Service treatment costs for fluoridated patients 2010-20 were 5.5% lower per person, by £22.26 (95% CI -£23.09 to -£21.43), and patients paid £7.64 less in National Health Service dental charges per person (2020 prices). Pragmatic, observational study with potential for non-differential errors of misclassification in fluoridation assignment and outcome measurement and residual and/or unmeasured confounding. Decayed, missing and filled teeth data have not been validated. Water fluoridation cost estimates are based on existing programmes between 2010 and 2020, and therefore do not include the potentially significant capital investment required for new programmes. Receipt of optimal water fluoridation between 2010 and 2020 resulted in very small health effects, which may not be meaningful for individuals, and we could find no evidence of a reduction in social inequalities. Existing water fluoridation programmes in England produced a positive return on investment between 2010 and 2020 due to slightly lower National Health Service treatment costs. These relatively small savings should be evaluated against the projected costs and lifespan of any proposed capital investment in water fluoridation, including new programmes. National Health Service dental data are a valuable resource for research. Further validation and measures to improve quality and completeness are warranted. This trial is registered as ISRCTN96479279, CAG: 20/CAG/0072, IRAS: 20/NE/0144. This award was funded by the National Institute for Health and Care Research (NIHR) Public Health Research programme (NIHR award ref: NIHR128533) and is published in full in Public Health Research; Vol. 12, No. 5. See the NIHR Funding and Awards website for further award information.

Revisiting Confidence Estimation: Towards Reliable Failure Prediction.

Reliable confidence estimation is a challenging yet fundamental requirement in many risk-sensitive applications. However, modern deep neural networks are often overconfident for their incorrect predictions, i.e., misclassified samples from known classes, and out-of-distribution (OOD) samples from unknown classes. In recent years, many confidence calibration and OOD detection methods have been developed. In this paper, we find a general, widely existing but actually-neglected phenomenon that most confidence estimation methods are harmful for detecting misclassification errors. We investigate this problem and reveal that popular calibration and OOD detection methods often lead to worse confidence separation between correctly classified and misclassified examples, making it difficult to decide whether to trust a prediction or not. Finally, we propose to enlarge the confidence gap by finding flat minima, which yields state-of-the-art failure prediction performance under various settings including balanced, long-tailed, and covariate-shift classification scenarios. Our study not only provides a strong baseline for reliable confidence estimation but also acts as a bridge between understanding calibration, OOD detection, and failure prediction.

LR-GLASSO Method for Solving Multiple Explanatory Variables of the Village Development Index

Sustainable Development Goals (SDGs) are developments that maintain sustainable improvement in society’s economic, social, and environmental welfare. Kemendes PDTT RI has issued the Village Development Index (VDI) to provide information and the status of village progress to support village development to improve the National SDGS. Modelling with multiple explanatory variables causes a high correlation between explanatory variables, multicollinearity, and coefficient estimation results, which have major variance and overfitting in prediction results. The modeling solution uses LASSO and GLASSO. Binary categorical response data uses binary logistic regression (LR), so LR-LASSO and LR-GLASSO are used. North Maluku Province has a VDI ranking that tends to fall in 2018-2022. Based on the mean and variance of the coefficient estimation results and misclassification errors, LR-GLASSO is better than LR-LASSO and LR. LR-GLASSO is recommended for analyzing VDI data because it has many explanatory variables, and the correlation between them is relatively high. The recommendation given by the Indonesian government, if it is to increase the status of VDI in Indonesia, especially in North Maluku province, is to increase the number of electricity users, food and beverage stalls, and other cooperatives. The Indonesian government also needs to pay attention to villages relatively far from the regent's office, between food and beverage stalls, and supporting health centres, because they still need to be developed compared to other villages, and more than 50% of villages are underdeveloped. If the Village SDGs are formulated through increasing VDI status, it will support the achievement of SDGs goals nationally.

A model-free variable screening method for optimal treatment regimes with high-dimensional survival data

Summary We propose a model-free variable screening method for the optimal treatment regime with high-dimensional survival data. The proposed screening method provides a unified framework to select the active variables in a prespecified target population, including the treated group as a special case. Based on this framework, the optimal treatment regime is exactly the optimal classifier that minimizes a weighted misclassification error rate, with weights associated with survival outcome variables, the censoring distribution and a prespecified target population. Our main contribution involves reformulating the weighted classification problem into a classification problem within a hypothetical population, where the observed data can be viewed as a sample obtained from outcome-dependent sampling, with the selection probability inversely proportional to the weights. Consequently, we introduce the weighted Kolmogorov–Smirnov approach for selecting active variables in the optimal treatment regime, extending the conventional Kolmogorov–Smirnov method for binary classification. Additionally, the proposed screening method exhibits two levels of robustness. The first level of robustness is achieved because the proposed method does not require any model assumptions for the survival outcome on treatment and covariates, whereas the other is attained as the form of treatment regimes is allowed to be unspecified even without requiring convex surrogate loss, such as logit loss or hinge loss. As a result, the proposed screening method is robust to model misspecifications, and nonparametric learning methods such as random forests and boosting can be applied to those selected variables for further analysis. The theoretical properties of the proposed method are established. The performance of the proposed method is examined through simulation studies and illustrated by a lung cancer dataset.

Emulating two qubits with a four-level transmon qudit for variational quantum algorithms

Using quantum systems with more than two levels, or qudits, can scale the computational space of quantum processors more efficiently than using qubits, which may offer an easier physical implementation for larger Hilbert spaces. However, individual qudits may exhibit larger noise, and algorithms designed for qubits require to be recompiled to qudit algorithms for execution. In this work, we implemented a two-qubit emulator using a 4-level superconducting transmon qudit for variational quantum algorithm applications and analyzed its noise model. The major source of error for the variational algorithm was readout misclassification error and amplitude damping. To improve the accuracy of the results, we applied error-mitigation techniques to reduce the effects of the misclassification and qudit decay event. The final predicted energy value is within the range of chemical accuracy.

Distribution-free Bayesian regularized learning framework for semi-supervised learning

In machine learning it is often necessary to assume or know the distribution of the data, however it is difficult to do so in practical applications. Aiming to this problem, this work, we propose a novel distribution-free Bayesian regularized learning framework for semi-supervised learning, which is called Hessian regularized twin minimax probability extreme learning machine (HRTMPELM). In this framework, we attempt to construct two non-parallel hyperplanes by introducing the high separation probability assumption, such that each hyperplane separates samples from one class with maximum probability while moving away from samples from the other class. Subsidiently, the framework can be utilized to construct reasonable semi-supervised classifiers by using the information of the inherent geometric distribution of the samples through the Hessian regularization term. Additionally, the proposed framework controls the misclassification error of samples by minimizing the upper limit of the worst-case misclassification probability, and improves the generalization performance of the model by introducing the idea of regularization to avoid the occurrence of ill-posedness and overfitting problems. More importantly, the framework has no hyperparameters, making the learning process very simplified and efficient. Finally, a simple and reliable algorithm with globally optimal solutions via multivariate Chebyshev inequalities is designed for solving the proposed learning framework. Experiments on multiple datasets demonstrate the reliability and effectiveness of the proposed learning framework compared to other methods. Especially, we applied the framework to Ningxia wolfberry quality detection, which greatly enriches and facilitates the application of machine learning algorithms in the agricultural field.

The impact of early large-scale screening on the evolution of pandemics

We study the problem of large-scale screening in the early stages of a pandemic. In this setting, resources such as testing kits, budget, and hospital beds are scarce, and testing in early stages has the potential to control the dynamics of disease spread. We thus devise a screening framework in which subjects are randomly sampled from the population for screening. We investigate two screening strategies - individual and group testing - and identify settings in which each is more effective. To model disease spread dynamics, we utilize a compartmental susceptible-infected-quarantined-removed-deceased model in which we embed sampling and screening decisions that account for testing misclassification errors. We perform a sensitivity analysis over the parameters that govern the disease dynamics and show the efficacy of our proposed screening solutions. We then calibrate our model using data on the COVID-19 pandemic in the United States and demonstrate the effectiveness of our model in reducing total number of infections while meeting healthcare system capacity.

Locality is the strongest predictor of expert performance in image-based differentiation of bacterial and fungal corneal ulcers from India.

This study sought to identify the sources of differential performance and misclassification error among local (Indian) and external (non-Indian) corneal specialists in identifying bacterial and fungal keratitis based on corneal photography. This study is a secondary analysis of survey data assessing the ability of corneal specialists to identify acute bacterial versus fungal keratitis by using corneal photography. One-hundred images of 100 eyes from 100 patients with acute bacterial or fungal keratitis in South India were previously presented to an international cohort of cornea specialists for interpretation over the span of April to July 2021. Each expert provided a predicted probability that the ulcer was either bacterial or fungal. Using these data, we performed multivariable linear regression to identify factors predictive of expert performance, accounting for primary practice location and surrogate measures to infer local fungal ulcer prevalence, including locality, latitude, and dew point. In addition, Brier score decomposition was used to determine experts' reliability ("calibration") and resolution ("boldness") and were compared between local (Indian) and external (non-Indian) experts. Sixty-six experts from 16 countries participated. Indian practice location was the only independently significant predictor of performance in multivariable linear regression. Resolution among Indian experts was significantly better (0.08) than among non-Indian experts (0.01; P < 0.001), indicating greater confidence in their predictions. There was no significant difference in reliability between the two groups ( P = 0.40). Local cornea experts outperformed their international counterparts independent of regional variability in tropical risk factors for fungal keratitis. This may be explained by regional characteristics of infectious ulcers with which local corneal specialists are familiar.

Efficacy of Regularized Multitask Learning Based on SVM Models.

This article investigates the efficacy of a regularized multitask learning (MTL) framework based on SVM (M-SVM) to answer whether MTL always provides reliable results and how MTL outperforms independent learning. We first find that the M-SVM is Bayes risk consistent in the limit of a large sample size. This implies that despite the task dissimilarities, the M-SVM always produces a reliable decision rule for each task in terms of the misclassification error when the data size is large enough. Furthermore, we find that the task-interaction vanishes as the data size goes to infinity, and the convergence rates of the M-SVM and its single-task counterpart have the same upper bound. The former suggests that the M-SVM cannot improve the limit classifier's performance; based on the latter, we conjecture that the optimal convergence rate is not improved when the task number is fixed. As a novel insight into MTL, our theoretical and experimental results achieved an excellent agreement that the benefit of the MTL methods lies in the improvement of the preconvergence-rate (PCR) factor (to be denoted in Section III) rather than the convergence rate. Moreover, this improvement of PCR factors is more significant when the data size is small. In addition, our experimental results of five other MTL methods demonstrate the generality of this new insight.

Rethinking Cost-Sensitive Classification in Deep Learning via Adversarial Data Augmentation

Cost-sensitive classification is critical in applications where misclassification errors widely vary in cost. However, overparameterization poses fundamental challenges to the cost-sensitive modeling of deep neural networks (DNNs). The ability of a DNN to fully interpolate a training data set can render a DNN, evaluated purely on the training set, ineffective in distinguishing a cost-sensitive solution from its overall accuracy maximization counterpart. This necessitates rethinking cost-sensitive classification in DNNs. To address this challenge, this paper proposes a cost-sensitive adversarial data augmentation (CSADA) framework to make overparameterized models cost sensitive. The overarching idea is to generate targeted adversarial examples that push the decision boundary in cost-aware directions. These targeted adversarial samples are generated by maximizing the probability of critical misclassifications and used to train a model with more conservative decisions on costly pairs. Experiments on well-known data sets and a pharmacy medication image (PMI) data set, made publicly available, show that our method can effectively minimize the overall cost and reduce critical errors while achieving comparable overall accuracy. Funding: Research reported in this publication was supported by the National Library of medicine of the National Institutes of Health in the United States under award number R01LM013624. Data Ethics &amp; Reproducibility Note: This paper abides by data ethics requirements. Data used are publicly available online at https://deepblue.lib.umich.edu/data/concern/data_sets/6d56zw997 . Codes to replicate the results of this paper are available at https://doi.org/10.24433/CO.2139841.v1 .

Spatial pattern of urban forest diversity and its potential drivers in a snow climate city, Northeast China

Urban forest diversity is the cornerstone of providing ecosystem services and improving urban environment. However, the spatial pattern of urban forest (UF) diversity and its driving mechanisms are still not well understood. In our study, using 330 UF sample plots and the corresponding plot-based standard deviation values of remote sensing parameters with different spatial resolutions, we developed prediction models for the spatial pattern of UF diversity, and the driving factors of UF diversity were further explored in our study. Our findings indicated that the 3-m spatial resolution of remote sensing was optimal for predicting UF diversity. Both excessively high and excessively low resolutions can increase misclassification errors. In addition, among the three prediction models, the boosted regression trees (BRT) model exhibited greater robustness than the support vector machine (SVM) and random forest (RF) models for predicting UF diversity. UF species were abundant, with 112 tree species in Changchun City. Spatially, the UF diversity showed a significant gradient decreasing trend from suburban areas (outer rings) to downtown areas. We also observed higher diversity in the older communities with high housing prices, supporting the “legacy effect and luxury effect”. Additionally, our research revealed that socioeconomic factors, landscape cover, and patterns collectively explained 41.25% of the UF diversity variation, wherein socioeconomic factors contributed the most variation (17.33%). We further found that UF diversity does not increase more with increasing house prices when the house prices are higher than ¥ 12,000/m2. The higher patch density reduced UF diversity, while a higher UF landscape shape index (LSI) promoted diversity. The LSI threshold, which significantly impacted UF diversity, was about 12. Our research showed that high-resolution remote sensing offers a rapid, cost-effective approach to acquire UF diversity patterns, which could enhance urban forest diversity and contribute to sustainable urban development.

Exact-matching algorithms using administrative health claims database equivalence factors for real-world data analysis based on the target trial emulation framework

AbstractReal-world data have become increasingly important in medical science and healthcare. A new, effective, and practically feasible statistical design is needed to unlock the potential of real-world data that decision-makers and practitioners can use to meet people’s healthcare needs. In the first half of the study, we validated our proposed new method by simulation, and in the second half, we conducted a clinical study on actual real-world data. We proposed the “Exact Matching Algorithm Using Administrative Health Claims Database Equivalence Factors (AHCDEFs)” using a target trial emulation framework. The simulation trials were conducted 500 times independently, considering the misclassification and chance errors of all variables and competing events of outcome. Two conventional methods, multivariate and propensity score analyses, were compared. Next, we estimated the effect of specific health guidance provided in Japan on the prevention of diabetes onset and medical expenditures. Our proposed novel method for real-world data returns improved estimates and fewer type I errors (the probability of erroneously determining that there is a difference when, in fact, there is no difference) than conventional methods. We quantitatively demonstrated the effectiveness of specific health guidance in Japan in preventing the onset of diabetes and reducing medical expenditures during five years. We proposed a new method for analyzing real-world data and an exact-matching algorithm using AHCDEFs. The larger the number of patients available for analysis, the more the AHCDEFs that can be matched, thereby removing the influence of confounding factors. This method will generate significant evidence when applied to real-world data.

Support matrix machine with truncated pinball loss for classification

With the expansion of vector-based classifiers to matrix-based classifiers, noise insensitivity and sparsity have always been the focal points. Existing SMM and Pin-SMM enjoy the former and the latter separately. To remedy the shortcoming, we propose a support matrix machine based on truncated pinball loss (TPin-SMM) in this paper, which integrates noise insensitivity and sparsity simultaneously. Thanks to the adding of two quantiles, it possesses precious properties including Bayes rule and bounding misclassification error as well. Concerning the non-convexity of TPin-SMM, a targeted CCCP-ADMM algorithm is established, which decomposes the problem into three sub-problems of each sub-iteration. To verify the validity of TPin-SMM, we have conducted numerical experiments on image datasets, EEG signal sets and Daimler Pedestrian Classification Benchmark dataset with different noises, all of them pass the statistical tests.

Harnessing Machine Learning for Accurate Cardiovascular Disease Prediction

Cardiovascular disease (CVD) is a life-threatening disease rising considerably in the world. Early detection and prediction of CVD as well as other heart diseases might protect many lives. This requires tact clinical data analysis. The potential of predictive machine learning algorithms to develop the doctor’s perception is essential to all stakeholders in the health sector since it can augment the efforts of doctors to have a healthier climate for patient diagnosis and treatment. We used the machine learning (ML) algorithm to carry out a significant explanation for accurate prediction and decision-making for CVD patients. Simple random sampling was used to select heart disease patients from the Khyber Teaching Hospital and Lady Reading Hospital, Pakistan. ML methods such as decision tree (DT), random forest (RF), logistic regression (LR), Naïve Bayes (NB), and support vector machine (SVM) were implemented for classification and prediction purposes for CVD patients in Pakistan. We performed exploratory analysis and experimental output analysis for all algorithms. We also estimated the confusion matrix and recursive operating characteristic curve for all algorithms. performance of the proposed ML algorithm was estimated using numerous conditions to recognize the most suitable machine learning algorithm in the class of models. RF algorithm had the highest accuracy of prediction, sensitivity, and recursive operative characteristic curve of 85.01%, 92.11%, and 87.73%, respectively, for CVD. It also had the least specificity and misclassification errors of 43.48% and 8.70%, respectively, for CVD. These results indicated that the RF algorithm is the most appropriate algorithm for CVD classification and prediction. Our proposed model can be implemented in all settings worldwide in the health sector for disease classification and prediction.

A Bayesian Approach to Estimate Maternal Mortality Globally Using National Civil Registration Vital Statistics Data Accounting for Reporting Errors

Estimation and monitoring of mortality, across multiple population-periods, is vital to improving national and global health outcomes. Reducing maternal mortality is a key part of the Sustainable Development Goals (SGDs), which commit countries and international agencies to monitor progress toward improvement of global maternal health outcomes. Country-specific estimates of maternal mortality are used for monitoring country progress. Civil registration vital statistics (CRVS) data provide valuable information on maternal mortality but are often subject to substantial reporting errors due to misclassification of maternal deaths or incompleteness of CRVS-reporting on deaths to women of reproductive age. Motivated by the challenge of estimating maternal mortality using error-prone CRVS data, this article introduces a Bayesian two-stage approach to obtain country-specific trends in maternal mortality ratios (MMRs), accounting for CRVS-related misclassification errors. In the first stage, we produce country-specific trends of CRVS-related data quality metrics, in terms sensitivity and specificity, through a Bayesian hierarchical misclassification model. In the second stage, we produce data-driven country trends in MMRs and proportion maternal deaths out of all-cause deaths, which are constructed using a Bayesian maternal mortality estimation model, accounting for CRVS-related data quality assessed in Stage 1. We present the approach to estimating maternal mortality and illustrate its use for several country case studies. This approach is used by the United National Maternal Mortality Interagency Group (UN-MMEIG) to produce maternal mortality estimates for global monitoring.

Assessment of adulteration of sage (Salvia sp.) with olive leaves using high‐performance thin‐layer chromatography, image analysis, and multivariate linear modeling

AbstractAccording to the study carried out at the University of Bristol, 60% of oregano spices present on the European Union (EU) market are adulterated with olive, myrtle, cistus, and hazelnut leaves. According to the same authors, the sage products are adulterated by similar bulking agents. The aim of this study was to assess possibilities for detection of sage adulteration by olive leaves using high‐performance thin‐layer chromatography (HPTLC) coupled with digital image analysis and multivariate linear regression/classification (partial least squares and partial least squares discriminant analysis). Twenty‐four samples (4 pure sage leaves, 4 pure olive leaves, and 16 mixtures of olive and sage leaves with content of added olive leaves varying in 5%, 10%, 20%, and 50%) have been prepared, extracted, and analyzed under normal‐phase conditions. Several derivatization methods were tested, and derivatized HPTLC plates were inspected under visible or ultraviolet light. Digital images of chromatograms were recorded. In order to minimize effects of intraplate and interplate peak shifts, background changes, and baseline drifts, correlation‐optimized warping, standard normal variate, and mean centering were applied to acquired signals. Partial least squares and partial least squares discriminant analysis models with moderate complexity (two to four latent variables) based on chromatographic signals obtained after derivatization by FeCl3, anisaldehyde–sulfuric acid, and 2,2‐diphenyl‐1‐picrylhydrazyl demonstrated good statistical performances with R2 ranging 0.894–0.998 and relative prediction error of 4–12%. Misclassification error &lt;4% was obtained in the case of 2,2‐diphenyl‐1‐picrylhydrazyl and anisaldehyde–sulfuric acid derivatization. Therefore, HPTLC combined with multivariate image analysis, signal processing, and linear modeling proved to be promising, cost‐effective chromatographic tool for assessment of sage adulteration by olive leaves.

Fast Convergence on Perfect Classification for Functional Data

We investigate the availability of approaching perfect classification on functional data with finite samples. The seminal work (Delaigle and Hall (2012)) showed that perfect classification for functional data is easier to achieve than for finite-dimensional data. This result is based on their finding that a sufficient condition for the existence of a perfect classifier, named a Delaigle--Hall condition, is only available for functional data. However, there is a danger that a large sample size is required to achieve the perfect classification even though the Delaigle--Hall condition holds, because a minimax convergence rate of errors with functional data has a logarithm order in sample size. This study solves this complication by proving that the Delaigle--Hall condition also achieves fast convergence of the misclassification error in sample size, under the bounded entropy condition on functional data. We study a reproducing kernel Hilbert space-based classifier under the Delaigle--Hall condition, and show that a convergence rate of its misclassification error has an exponential order in sample size. Technically, our proof is based on (i) connecting the Delaigle--Hall condition and a margin of classifiers, and (ii) handling metric entropy of functional data. Our experiments support our result, and also illustrate that some other classifiers for functional data have a similar property.