Labeling Errors Research Articles

Quality and Labelling of Meat Preparations on the Serbian Market According to the New Regulations

In the paper are shown the results of chemical quality parameters of meat preparations, sampled on the domestic market, which were analyzed from the aspect of regulations that define the quality of meat products as well as product labeling. A total of 30 samples of meat preparations including Minced meat with ingredients, Cevapcici, Pljeskavica, Barbecue meat, Fresh sausages, and fresh sausages under another name were investigated. The highest percentage of non- compliant samples concerning meat protein content as well as the content of collagen in meat proteins were determined in Fresh sausages (33 %). The highest percentage of the label non-compliances (60%) was concerning the missing nutritive declaration but also a huge problem seems to be inadequate specifying of proteins where 54 % of label errors referred to proteins such as missing information about meat protein content, use of non-meat protein sources where it is not allowed and inapropriate protein content designation. Other errors were represented in a smaller extent and refer to improper designation of the product group and additives, as well as missing information about the producer.

Electronic patient identification for sample labeling reduces wrong blood in tube errors.

Wrong blood in tube (WBIT) errors are a preventable cause of ABO-mismatched RBC transfusions. Electronic patient identification systems (e.g., scanning a patient's wristband barcode before pretransfusion sample collection) are thought to reduce WBIT errors, but the effectiveness of these systems is unclear. Part 1: Using retrospective data, we compared pretransfusion sample WBIT rates at hospitals using manual patient identification (n = 16 sites; >1.6 million samples) with WBIT rates at hospitals using electronic patient identification for some or all sample collections (n = 4 sites; >0.5 million samples). Also, we compared WBIT rates after implementation of electronic patient identification with preimplementation WBIT rates. Causes and frequencies of WBIT errors were evaluated at each site. Part 2: Transfusion service laboratories (n = 18) prospectively typed mislabeled (rejected) samples (n = 2844) to determine WBIT rates among samples with minor labeling errors. Part 1: The overall unadjusted WBIT rate at sites using manual patient identification was 1:10,110 versus 1:35,806 for sites using electronic identification (p < 0.0001). Correcting for repeat samples and silent WBIT errors yielded overall adjusted WBIT rates of 1:3046 for sites using manual identification and 1:14,606 for sites using electronic identification (p < 0.0001), with wide variation among individual sites. Part 2: The unadjusted WBIT rate among mislabeled (rejected) samples was 1:71 (adjusted WBIT rate, 1:28). In this study, using electronic patient identification at the time of pretransfusion sample collection was associated with approximately fivefold fewer WBIT errors compared with using manual patient identification. WBIT rates were high among mislabeled (rejected) samples, confirming that rejecting samples with even minor labeling errors helps mitigate the risk of ABO-incompatible transfusions.

Structural and functional analysis of the GABARAP interaction motif (GIM).

Corrigendum5 December 2018free access Structural and functional analysis of the GABARAP interaction motif (GIM) Vladimir V Rogov Search for more papers by this author Alexandra Stolz Search for more papers by this author Arvind C Ravichandran Search for more papers by this author Diana O Rios-Szwed Search for more papers by this author Hironori Suzuki Search for more papers by this author Andreas Kniss Search for more papers by this author Frank Löhr Search for more papers by this author Soichi Wakatsuki Search for more papers by this author Volker Dötsch Search for more papers by this author Ivan Dikic Search for more papers by this author Renwick CJ Dobson Search for more papers by this author David G McEwan orcid.org/0000-0003-0353-6532 Search for more papers by this author Vladimir V Rogov Search for more papers by this author Alexandra Stolz Search for more papers by this author Arvind C Ravichandran Search for more papers by this author Diana O Rios-Szwed Search for more papers by this author Hironori Suzuki Search for more papers by this author Andreas Kniss Search for more papers by this author Frank Löhr Search for more papers by this author Soichi Wakatsuki Search for more papers by this author Volker Dötsch Search for more papers by this author Ivan Dikic Search for more papers by this author Renwick CJ Dobson Search for more papers by this author David G McEwan orcid.org/0000-0003-0353-6532 Search for more papers by this author Author Information Vladimir V Rogov, Alexandra Stolz, Arvind C Ravichandran, Diana O Rios-Szwed, Hironori Suzuki, Andreas Kniss, Frank Löhr, Soichi Wakatsuki, Volker Dötsch, Ivan Dikic, Renwick CJ Dobson and David G McEwan EMBO Rep (2018)19:e47268https://doi.org/10.15252/embr.201847268 This article corrects the following: Structural and functional analysis of the GABARAP interaction motif (GIM)27 June 2017 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The authors recently noticed two labeling errors: In Table 1, the LIR motif of FUNDC1 was incorrectly labeled as “amino acids 152-160”. The correct annotation of the LIR motif is “aa 15-23”. In Tables 1 and 2, the protein names KTBD6 and KTBD7 were spelled incorrectly. The correct protein names are KBTBD6 and KBTBD7. These errors have been corrected inline on the journal website (https://doi.org/10.15252/embr.201643587), but these corrections have not been updated on PubMedCentral (PMC5538626). The authors apologize for these errors and any inconvenience caused by this oversight. Next ArticlePrevious Article Read MoreAbout the coverClose modalView large imageVolume 19,Issue 12,December 2018Cover: ADAR1 is responsible for editing dsRNAs, and loss of ADAR1 activates the innate immune response. ADAR1‐mediated editing is essential for thymic T cell development, and for counteracting autoimmunity via MDA5‐sensing of unedited dsRNA. From Taisuke Nakahama, Yukio Kawahara and colleagues: ADAR1‐mediated RNA editing is required for thymic self‐tolerance and inhibition of autoimmunity. For detail, see Article on page e46303. Cover concept by the authors. Cover illustration by Beata Edyta Mierzwa ( www.BeataScienceArt.com). Volume 19Issue 121 December 2018In this issue RelatedDetailsLoading ...

Contributing factors to outpatient pharmacy near miss errors: a Malaysian prospective multi-center study.

Background Detecting errors before medication dispensed or 'near misses' is a crucial step to combat the incidence of dispensing error. Despite this, no published evidence available in Malaysia relating to these issues. Objective To determine the incidence of medication labeling and filling errors, frequency of each type of the errors and frequency of the contributing factors at the final stage before dispensing. Setting Six Penang public funded hospitals outpatient pharmacies. Methods A prospective multicentre study, over 8week's period. Pharmacists identified and recorded the details of either medication labeling and/or filling error at the final stage of counter-checking before dispensing. Besides, the contributing factors for each error were determined and recorded in data collection form. Descriptive analysis was used to explain the study data. Main outcome measure The incidence of near misses. Results A total of 187 errors (near misses) detected, with 59.4% (n = 111) were medication filling errors and 40.6% (n = 76) were labeling errors. Wrong drug (n = 44, 39.6%) was identified as the highest type of filling errors while incorrect dose (n = 34, 44.7%) was identified as the highest type of labeling errors. Distracted and interrupted work environment was reported to lead the highest labeling and filling errors, followed by lack of knowledge and skills for filling errors and high workload for labeling errors. Conclusion The occurrence of near misses related to medication filling and labelling errors is substantial at outpatient pharmacy in Penang public funded hospitals. Further research is warranted to evaluate the intervention strategies needed to reduce the near misses.

QuickNAT: A fully convolutional network for quick and accurate segmentation of neuroanatomy

Whole brain segmentation from structural magnetic resonance imaging (MRI) is a prerequisite for most morphological analyses, but is computationally intense and can therefore delay the availability of image markers after scan acquisition. We introduce QuickNAT, a fully convolutional, densely connected neural network that segments a MRI brain scan in 20 s. To enable training of the complex network with millions of learnable parameters using limited annotated data, we propose to first pre-train on auxiliary labels created from existing segmentation software. Subsequently, the pre-trained model is fine-tuned on manual labels to rectify errors in auxiliary labels. With this learning strategy, we are able to use large neuroimaging repositories without manual annotations for training. In an extensive set of evaluations on eight datasets that cover a wide age range, pathology, and different scanners, we demonstrate that QuickNAT achieves superior segmentation accuracy and reliability in comparison to state-of-the-art methods, while being orders of magnitude faster. The speed up facilitates processing of large data repositories and supports translation of imaging biomarkers by making them available within seconds for fast clinical decision making.

Systematic Labeling Bias in Galaxy Morphologies

Abstract We present a metric to quantify systematic labeling bias in galaxy morphology data sets stemming from the quality of the labeled data. This labeling bias is independent from labeling errors and requires knowledge about the intrinsic properties of the data with respect to the observed properties. We conduct a relative comparison of label bias for different low-redshift galaxy morphology data sets. We show our metric is able to recover previous de-biasing procedures based on redshift as biasing parameter. By using the image resolution instead, we find biases that have not been addressed. We find that the morphologies based on supervised machine learning trained over features such as colors, shape, and concentration show significantly less bias than morphologies based on expert or citizen-science classifiers. This result holds even when there is underlying bias present in the training sets used in the supervised machine learning process. We use catalog simulations to validate our bias metric and show how to bin the multi-dimensional intrinsic and observed galaxy properties used in the bias quantification. Our approach is designed to work on any other labeled multi-dimensional data set, and the code is publicly available (https://github.com/guille-c/labeling_bias).

Genetic Characterization of the Apple Germplasm Collection in Central Italy: The Value of Local Varieties.

In the last 50 years, intensive farming systems have been boosted by modern agricultural techniques and newly bred cultivars. The massive use of few and related cultivars has dramatically reduced the apple genetic diversity of local varieties, confined to marginal areas. In Central Italy a limited spread of intensive fruit orchards has made it possible to preserve much of the local genetic diversity, but at the same time the coexistence of both modern and ancient varieties has generated some confusion. The characterization and clarification of possible synonyms, homonyms, and/or labeling errors in old local genetic resources is an issue in the conservation and management of living collections. 175 accessions provided by 10 apple collections, mainly local varieties, some of unknown origin, and well-known modern and ancient varieties, were studied by using 19 SSRs, analyzed by STRUCTURE, Ward’s clustering and parentage analysis. We were able to identify 25 duplicates, 9 synonyms, and 9 homonyms. As many as 37 unknown accession were assigned to well known local or commercial varieties. Polyploids made up 20%. Some markers were found to be significantly correlated with morphological traits and the loci associated with the fruit over color were related to QTLs for resistance to biotic stresses, aroma compounds, stiffness, and acidity. In conclusion the gene pool of Central Italy seems to be rather consistent and highly differentiated compared with other European studies (FST = 0.147). The importance of safeguarding this diversity and the impact on the management of the germplasm living collection is discussed.

High-throughput ovarian follicle counting by an innovative deep learning approach

The evaluation of the number of mouse ovarian primordial follicles (PMF) can provide important information about ovarian function, regulation of folliculogenesis or the impact of chemotherapy on fertility. This counting, usually performed by specialized operators, is a tedious, time-consuming but indispensable procedure.The development and increasing use of deep machine learning algorithms promise to speed up and improve this process. Here, we present a new methodology of automatically detecting and counting PMF, using convolutional neural networks driven by labelled datasets and a sliding window algorithm to select test data. Trained from a database of 9 millions of images extracted from mouse ovaries, and tested over two ovaries (3 millions of images to classify and 2 000 follicles to detect), the algorithm processes the digitized histological slides of a completed ovary in less than one minute, dividing the usual processing time by a factor of about 30. It also outperforms the measurements made by a pathologist through optical detection. Its ability to correct label errors enables conducting an active learning process with the operator, improving the overall counting iteratively. These results could be suitable to adapt the methodology to the human ovarian follicles by transfer learning.

Analysis of Computer-Generated Drug Label Errors in a Tertiary Care Hospital

Background: Electronic Systems and recommendations have been developed to facilitate the medication dispensing process, but on the other hand, they may generate dispensing errors themselves. This study is meant to assess the computer-generated drug label errors (CGDLEs). Objective: The main objective of this study is to identify the pattern and assess the significance of CGDLEs in a tertiary care hospital. Methodology: A total of 292 computer-generated drug label errors involving 169 patients and 61 medications were researched over 3 months (from January 1, 2017, to Mar 30, 2017) in a 500-bed tertiary care hospital. Collected data included patient demographics, type of error, error-related medications, time taken to resolve the error, station where error was identified and the clinical significance of the error. Analyses were conducted using STATA® v14.2 with descriptive and inferential statistics. Results: Thirty eight percent of the detected CGDLEs were considered major. Patients age ranged between 0.2 to 97 years (M+SD=36.4+24.3). Errors within the age group of 0.2 to 2 years represented 7.89% of the total errors. Errors within the age group of 61 years and above represented 21.91% of the total errors. Duration of therapy represents 34.59% of the total errors, followed by instructions for use (29.45%) and drug dose (19.86%). The major CGDLEs commonly include medications like tacrolimus (17.69%), methotrexate (16.81%) and ciclosporin (15.40%). Conclusion: A considerable proportion of all CGDLEs was observed. Many of these errors were serious and could directly affect the wellbeing of patients. Fortunately, these errors were captured before reaching patients. Urgent assessment of the system that generates such labels is required. Key words: Computer-generated drug label, Dispensing errors, Medication errors.

Glioma-derived cancer stem cells are hypersensitive to proteasomal inhibition.

Corrigendum5 September 2018free access Glioma-derived cancer stem cells are hypersensitive to proteasomal inhibition Young Dong Yoo Search for more papers by this author Dae-Hee Lee Search for more papers by this author Hyunjoo Cha-Molstad Search for more papers by this author Hyungsin Kim Search for more papers by this author Su Ran Mun Search for more papers by this author Changhoon Ji Search for more papers by this author Seong Hye Park Search for more papers by this author Ki Sa Sung Search for more papers by this author Seung A Choi Search for more papers by this author Joonsung Hwang Search for more papers by this author Deric M Park Search for more papers by this author Seung Ki Kim Search for more papers by this author Kyung-Jae Park Search for more papers by this author Shin-Hyuk Kang Search for more papers by this author Sang Cheul Oh Search for more papers by this author Aaron Ciechanover Search for more papers by this author Yong J Lee orcid.org/0000-0001-5914-8489 Search for more papers by this author Bo-Yeon Kim orcid.org/0000-0003-1105-5480 Search for more papers by this author Yong Tae Kwon orcid.org/0000-0002-8115-3150 Search for more papers by this author Young Dong Yoo Search for more papers by this author Dae-Hee Lee Search for more papers by this author Hyunjoo Cha-Molstad Search for more papers by this author Hyungsin Kim Search for more papers by this author Su Ran Mun Search for more papers by this author Changhoon Ji Search for more papers by this author Seong Hye Park Search for more papers by this author Ki Sa Sung Search for more papers by this author Seung A Choi Search for more papers by this author Joonsung Hwang Search for more papers by this author Deric M Park Search for more papers by this author Seung Ki Kim Search for more papers by this author Kyung-Jae Park Search for more papers by this author Shin-Hyuk Kang Search for more papers by this author Sang Cheul Oh Search for more papers by this author Aaron Ciechanover Search for more papers by this author Yong J Lee orcid.org/0000-0001-5914-8489 Search for more papers by this author Bo-Yeon Kim orcid.org/0000-0003-1105-5480 Search for more papers by this author Yong Tae Kwon orcid.org/0000-0002-8115-3150 Search for more papers by this author Author Information Young Dong Yoo, Dae-Hee Lee, Hyunjoo Cha-Molstad, Hyungsin Kim, Su Ran Mun, Changhoon Ji, Seong Hye Park, Ki Sa Sung, Seung A Choi, Joonsung Hwang, Deric M Park, Seung Ki Kim, Kyung-Jae Park, Shin-Hyuk Kang, Sang Cheul Oh, Aaron Ciechanover, Yong J Lee, Bo-Yeon Kim and Yong Tae Kwon EMBO Rep (2018)19:e46380https://doi.org/10.15252/embr.201846380 This article corrects the following: Glioma-derived cancer stem cells are hypersensitive to proteasomal inhibition19 December 2016 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The authors recently noticed a labeling error in Figure 7F, which has been replaced by one with the correct labels. The labeling does not affect the conclusions of the experiment. The authors apologize for any inconvenience caused by this oversight. Figure F. Download figure Download PowerPoint Next ArticlePrevious Article Volume 19Issue 91 September 2018In this issue FiguresRelatedDetailsLoading ...

Etikettierung von Laborröhrchen: ein komplexer Prozess mit hoher Fehleranfälligkeit

Errors associated with the labelling of laboratory specimens repeatedly occur in health facilities and are often described in Critical Incident Reporting Systems (CIRS). The analysis of these error reports and the complexity of the specimen collection process pose major challenges for responsible health care professionals. To examine which errors occur in the labelling of laboratory specimens, and what leads to their detection, the error reports of 42 Swiss hospitals within the Critical Incident Reporting & Reaching NETwork (CIRRNET) were systematically analysed. MethodWithin the scope of a reporting month in the year 2016, 42 hospitals were asked to pay particular attention to the issue of mislabelled laboratory specimens, to report these errors in the local CIRS system, and to forward the information to the CIRRNET database. In addition, using a systematic keyword search, a search for old error reports on the mislabelling of laboratory specimens was conducted in the CIRRNET database and the results extracted. 227 error reports were finally included in the analysis. All these error reports were analysed systematically and by content, and the problem areas described in these reports allocated to newly defined categories. The systematic analysis included: the time of error detection, the problem area described, the length of time between error occurrence and error detection, as well as possible safety barriers. ResultsThe majority (52 %) of labelling errors of laboratory specimens are detected in the laboratory, 21.1 % on the hospital ward before the specimens are sent to the laboratory to be processed, and a further 24.7 % at a very late stage when test results are associated with the patient on the ward. The analysis of the problem areas described showed that patient identification (7.9 %) was a key issue. The most frequent errors occurred in connection with the labelling of laboratory test tubes (45.4 %) and analysis forms or requisition slips (33 %). Numerous errors went undetected throughout the entire process, i.e. right up until the moment when the test result was finally assigned to the patient. The analysis also provided information about possible safety barriers for future prevention of laboratory specimen mislabelling. ConclusionsA content-analysis approach is essential to the evaluation of error reports as it is the only way to identify all the problems of laboratory specimen mislabelling described. Furthermore, a deductive approach also facilitates the identification of possible safety barriers. The findings provide valuable information for mislabelling prevention approaches that is not otherwise available. Further analysis of error reports from Critical Incident Reporting Systems is necessary in order to gain more experience with the methodical approach and to draw conclusions as to whether the complex process of analysing each individual error report yields information that is more useful than the analysis of the error reports of an identified problem area.

Joint skeleton estimation of multiple directed acyclic graphs for heterogeneous population.

The directed acyclic graph (DAG) is a powerful tool to model the interactions of high-dimensional variables. While estimating edge directions in a DAG often requires interventional data, one can estimate the skeleton of a DAG (i.e., an undirected graph formed by removing the direction of each edge in a DAG) using observational data. In real data analyses, the samples of the high-dimensional variables may be collected from a mixture of multiple populations. Each population has its own DAG while the DAGs across populations may have significant overlap. In this article, we propose a two-step approach to jointly estimate the DAG skeletons of multiple populations while the population origin of each sample may or may not be labeled. In particular, our method allows a probabilistic soft label for each sample, which can be easily computed and often leads to more accurate skeleton estimation than hard labels. Compared with separate estimation of skeletons for each population, our method is more accurate and robust to labeling errors. We study the estimation consistency for our method, and demonstrate its performance using simulation studies in different settings. Finally, we apply our method to analyze gene expression data from breast cancer patients of multiple cancer subtypes.

Eliminating Specimen Labeling Errors in Post-Anesthesia Care Unit (PACU)- Phase II

Eliminating Specimen Labeling Errors in Post-Anesthesia Care Unit (PACU)- Phase II

Optimal approaches to the quality control checking of product labels

Quality control checkers at fresh produce packaging facilities occasionally fail to detect incorrect information presented on labels. Despite being infrequent, such errors have significant financial and environmental repercussions. To understand why label-checking errors occur, observations and interviews were undertaken at a large packaging facility and followed up with a laboratory-based label-checking task. The observations highlighted the dynamic, complex environment in which label-checking took place, whilst the interviews revealed that operatives had not received formal training in label-checking. On the laboratory-based task, overall error detection accuracy was high but considerable individual differences were found between professional label-checkers. Response times were shorter when participants failed to detect label errors, suggesting incomplete checking or ineffective checking strategies. Furthermore, eye movement recordings indicated that checkers who adopted a systematic approach to checking were more successful in detecting errors. The extent to which a label checker adopted a systematic approach was not found to correlate with the number of years of experience that they had accrued in label-checking. To minimize the chances of label errors going undetected, explicit instruction and training, personnel selection and/or the use of software to guide performance towards a more systematic approach is recommended.

Multi-label semi-supervised classification through optimum-path forest

Multi-label classification consists of assigning one or multiple classes to each sample in a given dataset. However, the project of a multi-label classifier is usually limited to a small number of supervised samples as compared to the number of all possible label combinations. This scenario favors semi-supervised learning methods, which can cope with the absence of supervised samples by adding unsupervised ones to the training set. Recently, we proposed a semi-supervised learning method based on optimum connectivity for single-label classification. In this work, we extend it for multi-label classification with considerable effectiveness gain. After a single-label data transformation, the method propagates labels from supervised to unsupervised samples, as in the original approach, by assuming that samples from the same class are more closely connected through sequences of nearby samples than samples from distinct classes. Given that the procedure is more reliable in high-density regions of the feature space, an additional step repropagates labels from the maxima of a probability density function to correct possible labeling errors from the previous step. Finally, the data transformation is reversed to obtain multiple labels per sample. The new approach is experimentally validated on several datasets in comparison with state-of-the-art methods.

Deep learning for plant identification: how the web can compete with human experts

Automated identification of plants and animals has improved considerably in the last few years, in particular thanks to the recent advances in deep learning. In order to evaluate the performance of automated plant identification technologies in a sustainable and repeatable way, a dedicated system-oriented benchmark was setup in 2011 in the context of ImageCLEF (Goëau et al. 2011). Each year, since that time, several research groups participated in this large collaborative evaluation by benchmarking their image-based plant identification systems. In 2014, the LifeCLEF research platform (Joly et al. 2014) was created in the continuity of this effort so as to enlarge the evaluated challenges by considering birds and fishes in addition to plants, and audio and video contents in addition to images. The 2017-th edition of the LifeCLEF plant identification challenge (Joly et al. 2017) is an important milestone towards automated plant identification systems working at the scale of continental floras with 10.000 plant species living mainly in Europe and North America illustrated by a total of 1.1M images. Nowadays, such ambitious systems are enabled thanks to the conjunction of the dazzling recent progress in image classification with deep learning and several outstanding international initiatives, aggregating the visual knowledge on plant species coming from the main national botanical institutes. The PlantCLEF plant challenge that we propose to present at this workshop aimed at evaluating to what extent a large noisy training dataset collected through the web (then containing a lot of labelling errors) can compete with a smaller but trusted training dataset checked by experts. To fairly compare both training strategies, the test dataset was created from a third data source, the Pl@ntNet (Joly et al. 2015) mobile application that collects millions of plant image queries all over the world. Due to the good results obtained at the 2017-th edition of the LifeCLEF plant identification challenge, the next big question is how far such automated systems are from the human expertise. Indeed, even the best experts are sometimes confused and/or disagree with each other when validating images of living organism. A multimedia data actually contains only partial information that is usually not sufficient to determine the right species with certainty. Quantifying this uncertainty and comparing it to the performance of automated systems is of high interest for both computer scientists and expert naturalists. This work reports an experimental study following this idea in the plant domain. In total, 9 deep-learning systems implemented by 3 different research teams were evaluated with regard to 9 expert botanists of the French flora. The main outcome of this work is that the performance of state-of-the-art deep learning models is now close to the most advanced human expertise. This shows that automated plant identification systems are now mature enough for several routine tasks, and can offer very promising tools for autonomous ecological surveillance systems.

Semi-Supervised Image Classification With Self-Paced Cross-Task Networks

In a semi-supervised setting, direct training of a deep discriminative model on partially labeled images often suffers from overfitting and poor performance, because only a small number of labeled images are available, and errors in label propagation are, in many cases, inevitable. In this paper, we introduce an auxiliary clustering task to explore the structure of the image data, and judiciously weigh unlabeled data to alleviate the influence of ambiguous data on model training. For this purpose, we propose a cross-task network composed of two streams to jointly learn two tasks: classification and clustering. Based on the model predictions, a large number of pairwise constraints can be generated from unlabeled images, and are fed to the clustering stream. Since pairwise constraints encode weak supervision information, the clustering is tolerant of errors in labeling. Unlabeled images are weighted according to the distances to the clusters discovered, and a better discriminative model is trained on the classification stream associated with a weighted softmax loss. Furthermore, a self-paced learning paradigm is adopted to gradually train our deep model from easy examples to difficult ones. Experimental results on widely used image classification datasets confirm the effectiveness and superiority of the proposed approach.

Genetic identity and diversity of Nigerian cacao genebank collections verified by single nucleotide polymorphisms (SNPs): a guide to field genebank management and utilization

Nigeria is the sixth largest cacao producer in the world. Field performance and quality of cacao hybrid families is largely dependent on the genetic integrity of parental clones obtained in field genebank collections. However, information on the impact of mislabeling on seed garden output in Nigeria is lacking. Using 63 single nucleotide polymorphism (SNP) markers, we analyzed 1457 cacao trees sampled from seven major field genebank plots in Nigeria to assess the genetic integrity in Nigerian cacao germplasm. The procedure of multilocus matching with known reference clones revealed up to 78% mislabeling in recently introduced international germplasm. A high rate of mislabeling was also revealed in the West African local selections and breeding lines, using Bayesian assignment test. The problem of mislabeling has been attributed to errors from the sources of introduction, pre-planting labeling errors, and rootstocks overtaking budded scions due to poor field management. The analysis of genetic diversity revealed a good representation of the available cacao germplasm groups in Nigerian field genebanks, indicating that the genetic base of Nigeria cacao germplasm has been significantly widened through germplasm introductions. However, only a small proportion of the available germplasm in the genebank have been utilized for variety development. This study proved the utility of SNP markers for cleaning up the genebanks and reducing offtypes; thereby providing a strong basis for improving the accuracy and efficiency in cacao genebank management and breeding, as well as for mobilizing improved varieties to cacao farmers in Nigeria.

Drug syrups: Errors in drug labels with possible consequences in patients with hereditary fructose intolerance

Drug syrups: Errors in drug labels with possible consequences in patients with hereditary fructose intolerance

Erratum: Diet Quality Predicts Visceral Adiposity and Liver Fatness: The Multiethnic Cohort Study.

Obesity (2017) 25: 1442-1450. doi:10.1002/oby.21868 After publication of this article, the authors detected a labeling error for one of the four dietary index variables in the data set. The Healthy Eating Index (HEI) 2010 variable for cohort entry was actually the HEI 2005 for the dietary data collected at cohort entry. However, it was correctly labeled for the diet data collected at the clinic visit. (The HEI 2005 was developed based on the dietary recommendations made to the US population in 2005, whereas the 2010 HEI was developed based on the 2010 recommendations. These recommendations have not changed dramatically between these two dates.) Therefore, all the results in the tables, figures, and text of the manuscript referring to diet quality at cohort entry show associations for the HEI 2005 and not for the HEI 2010 as stated in the manuscript. The results for the other three diet quality indices were not affected. The authors subsequently concluded that the HEI 2005 represents an appropriate alternate description of diet quality, especially for the dietary data collected at cohort baseline in 1995 since it is the earliest density-based index available for these recommendations. Therefore, the error does not affect the validity of the analysis for HEI. In addition, after repeating all analyses with the correct variable, the authors found that the findings did not change substantially and that the overall conclusions of the manuscript remain the same.