Skewness Research Articles

P-721 Exploring the impact of geographical variation: A retrospective study examining semen phenotypes within a cohort of 19,612 men attending a fertility center in the UAE

Abstract Study question What are the epidemiological trends of semen phenotypes and their association with geographic regions among men seeking fertility treatment in the UAE? Summary answer Following covariate adjustment, UAE Nationals attending a fertility center displayed lower incidence of oligozoospermia and higher prevalence of asthenozoospermia and teratozoospermia compared to other nationalities. What is known already Conventional semen analysis remains the gold standard for diagnosing and managing male infertility. It is well documented that semen is heterogenous and can be influenced by diverse factors, including but not limited to race, ethnicity, sociocultural background, seasonality, and geographical region. Although higher infertility prevalence has been reported in regions like North Africa and the Middle East, existing estimates rely largely on data from female partners of infertile couples and studies dating back to 1988. This underscores a significant data paucity in the Middle East. Study design, size, duration In this retrospective study, the anthropometric data (age, weight, height, body mass index, nationality) and semen parameters (semen volume, semen pH, sperm concentration, total sperm count (TSC), total motility (TM), progressive motility (PM), normal morphology, viability) of 19,612 men (32,664 samples) who attended a fertility center in the UAE between 2011 and 2022 were assessed. Ethical approvals for this study were obtained from appropriate bodies (REC/2022/P27 and MBRU IRB-2021-12). Participants/materials, setting, methods Data from patients were categorized into different geographical regions (Sub-Saharan Africa, North Africa, North America, Central America, South America, Asia, Europe, Middle East, and Oceania) based on nationality. Patients were further grouped as MENA (Middle East and North Africa) and non-MENA, while data from UAE Nationals (UAE) were also compared to Other Nationalities (Global). Unadjusted analysis was performed using SPSS. While covariate adjustment and propensity score analyses were performed using R package and STATA. Main results and the role of chance All semen parameters, remained within normal WHO reference values when stratifying according to geographical regions, MENA and Non-MENA as well as UAE and Global. Interestingly, MENA displayed a significant reduction in semen volume (p < 0.0001) and sperm concentration (p = 0.017), and an increase in PM (p < 0.00001). Similarly, PM was higher in UAE versus Global. To address skewed data distribution between UAE and Global, we employed propensity score analysis and logistic regression models, with semen parameters categorized according to WHO lower reference limits. Following covariate adjustment, all models demonstrated no significant difference in PM and TM between the two subgroups. However, logistic regression indicated a non-significant increase (4%) in the prevalence of asthenozoospermia in UAE compared to Global. Categorical analysis indicated that the Global samples had significantly lower TSC compared to the UAE group (24.6% vs. 26.3%, p<.001). Modelling adjusted for propensity score and UAE Nationality further suggested that UAE were 13% less likely to have lower TSC compared to Global (p<.001). About 58% of UAE samples exhibited teratozoospermia, while it was 56% for Global. After adjusting for confounders, analysis revealed a significantly higher prevalence (12%) of teratozoospermia among UAE compared to Global, with a potential range of 9%-16% in the long term. Limitations, reasons for caution A limitation of the current study is data distribution skewness. Of the 32,664 samples from 19,612 patients (113 countries), 21,649 samples are from UAE Nationals. However, logistic regression, propensity score adjustment accounting for Nationality, propensity score matching (1:2 ratio), and inverse probability weighting were applied to mitigate this bias. Wider implications of the findings Paucity of data on male fertility in the UAE and MENA region does not allow for accurate estimation of male infertility. Owed to its sample size and statistical robustness, this study provides evidence addressing this gap, offering insights into interpretation of global data on reproductive health. Trial registration number not applicable

P-429 Measuring serum progesterone levels on the day of a fresh blastocyst transfer is debatable as there is no evidence for correlation with ongoing pregnancy results

Abstract Study question Does serum progesterone (P4) level on the day of a fresh blastocyst transfer relate to ongoing pregnancy rate (OPR)? Summary answer OPR after a fresh single blastocyst transfer seems not to be associated with serum P4 level on the day of transfer. What is known already Several studies have suggested an optimal serum P4 threshold for embryo transfers in frozen embryo transfer cycles. However, for fresh embryo transfers, data are scarce and conflicting. Study design, size, duration A retrospective single-center cohort study was performed on 598 complete data cycles, retained out of 631 consecutive fresh single blastocyst transfer cycles. Data were collected from patients undergoing a fresh IVF/ICSI cycle with autologous oocytes at a tertiary fertility center between June 2022 and August 2023. Participants/materials, setting, methods A total of 482 unique patients underwent one or more single fresh blastocyst transfers using micronized vaginal progesterone (MVP) 800mg BID for luteal phase support starting day after oocyte retrieval. Serum P4 levels were determined on day of transfer. The primary endpoint was OPR, positive hCG-test and early pregnancy loss (EPL) were secondary endpoints. Unadjusted and adjusted generalized estimating equations (GEE) assessed P4 level association with the endpoints, using an exchangeable working correlation matrix. Main results and the role of chance No association was identified between P4 levels on day of transfer and OPR after accounting for maternal age, blastocyst quality, P4 and estradiol level at last ultrasound, BMI and endometrial thickness. Data were grouped according to 10/50/90 and 25/50/75 percentiles, resulting in groups < = or > 27ng/ml and < = or > 40ng/ml, with omission of 90th and 75th percentile groups due to right-skewed distribution. With serum P4 levels higher than 40ng/ml, the probability of a positive hCG-test significantly increased (OR 1.79, CI 1.11-2.91; p = 0.018). No significant difference was observed in patients with serum P4 < = or > 27ng/ml (OR 1.23, CI 0.66-2.30; p = 0.506). Regarding OPR, no significant differences were observed when comparing patients with serum P4 < = or > 27/ng/ml or < = or > 40ng/ml (OR 0.88, CI 0.42-18.85; p = 0.736 and OR 1.41, CI 0.83-2.37; p = 0.203, respectively). Moreover, EPL did not significantly differ between serum P4 level groups (OR 1.93, CI 0.61-6.13; p = 0.265 and OR 1.2, CI 0.51-2.84; p = 0.674). Age (p < 0.001) and embryo quality (p < 0.001) are significantly related with OPR. Additionally, late follicular progesterone levels and OPR are significantly associated (OR 2.39, CI 1.16-4.95 in patients with P4 < = or > 27/ng/ml and OR 2.21, CI 1.07-4.55 in patients with < =40ng/ml and >40ng/ml). Limitations, reasons for caution Limitations include the retrospective design and the lack of live birth outcome data. Progesterone levels exceeding 60 ng/ml were not further analyzed to their precise value but grouped as 60 plus which did not allow to identify potential detrimental effects of very high P4 levels on treatment outcome. Wider implications of the findings Our data suggest that measuring P4 on the day of a single blastocyst transfer with MVP luteal phase support may not be indicative of OPR or the risk of EPL. This raises questions regarding the necessity for routine P4 measurements as well as the development of rescue strategies. Trial registration number ONZ-2023-0628

Adaptive-conditional loss and correction module enhanced informer network for long-tailed fault diagnosis of motor

Abstract This study focuses on the motor fault diagnosis facing the long-tailed distribution data, characterized by a multitude of fault types with limited data per category and the healthy state with massive data. This skewed distribution makes the traditional diagnostic models fail to identify less frequent faults. To this end, we introduce a novel fault diagnosis model, named Transformer- and gated-recurrent unit (GRU)-based network (TransGRU), to improve the diagnosis accuracy with the long-tailed distribution data. The TransGRU has two main modules, i.e., the feature extraction module and the correction module. The former is based on the Informer encoder with ProbSparse self-attention to extract features from the long-range multi-sensor data. The latter employs the GRU network addressing the long-tail effect by adjusting the diagnosis results via the gate mechanism. Besides, we informatively design an adaptive-conditional loss (ACL) function for the long-tailed fault diagnosis by integrating the properties of focal loss, class-tailored weights, and confusion weights. ACL concentrates on challenging classifications while balancing the representation and significance of various fault modes. Validation on experimental motor data confirms the capability of our TransGRU in identifying a wide range of fault types with limited fault data compared with the Transformer and state-of-the-art methods.

Predictive Assessment of the Interior Quality of Apartments Based on Multimodal Data with Variable Visual Input

Multimodal analyses in the context of automated real estate valuation (AVM) offer the possibility of enriching the models with additional information, which benefits the accuracy of the models. However, this variety of data can overwhelm common machine learning models, which generally only process certain data modalities and only a fixed data quantity. This leads to a bottleneck in information processing, as in many cases much more information is available per observation, of which only a single selection or sample is used to train the algorithm and the remaining information is disregarded. We propose a multimodal network architecture that incorporates both textual and visual inputs and fuses their information. Furthermore, we introduce a training strategy that can take advantage of a variable number of input images for each real estate object. In our experiments, we test and compare several unimodal (baseline) models with our multimodal architecture. Our approach shows several advantages in terms of model performance over unimodal approaches. The results show the best performance for the multimodal model with a variable number of visual inputs, as well as improved prediction for the underrepresented classes of indoor quality, mitigating the effects of unbalanced data. With the presented approach, which efficiently combines and merges multiple data modalities, we have shown how such a method can be easily adapted to an AVM for the extraction of supplementary information.

A Comparison of Bayesian Approximation Methods for Analyzing Large Spatial Skewed Data

A Comparison of Bayesian Approximation Methods for Analyzing Large Spatial Skewed Data

Confidence intervals for median absolute deviations

The median absolute deviation (MAD) is a robust measure of scale that is simple to implement, easy to interpret and often reported as a measure of dispersion when data is skewed. However, confidence intervals either for a single MAD or for the comparison of two MADs are lacking, meaning that interpretations may be subjective. We introduce interval estimators of the MAD to make reliable inferences for dispersion based on MADs. Our simulation results show that the coverages of the intervals are very close to the nominal for a variety of distributions, including those heavily skewed. We also present examples that clearly highlight the advantages of using these intervals instead of the heavily criticized intervals derived from sample variances for skewed data.

Cross-domain aspect-based sentiment classification with hybrid prompt

With the popularity of online shopping, an increasing number of buyers and sellers are basing their decisions on mass online reviews. As an efficient approach to obtain fine-grained sentiment information, cross-domain aspect-based sentiment classification has received extensive attention from academia and industry. However, this task encounters challenges such as limited labeled data or extreme differences in sentiment distributions between source and target domains. To address these issues, we propose a novel approach named CSC-PLDAT (Cross-domain Aspect-based Sentiment Classification with Prompt Learning and Domain Adversarial Training). Specifically, CSC-PLDAT incorporates a hybrid prompt comprising transferable and task-specific components. Based on this prompt, our model can effectively transfer domain-specific sentiment information. Meanwhile, it can adapt to the situation with limited labeled or skewed sentiment distributions. Additionally, considering various sentiment words can express identical sentiment polarities, we employ domain adversarial training to generate domain-independent sentiment representations for sentiment classification. The experiment is conducted on four popular benchmark datasets. The results demonstrate that our model outperforms the state-of-the-art methods, with an average micro-F1 score of 71.45% in cross-domain aspect-based sentiment classification. Furthermore, our model exhibits promising outcomes, particularly in scenarios characterized by limited labeled data and skewed sentiment distributions.

A deep learning drug screening framework for integrating local-global characteristics: A novel attempt for limited data

At the beginning of the "Disease X" outbreak, drug discovery and development are often challenged by insufficient and unbalanced data. To address this problem and maximize the information value of limited data, we propose a drug screening model, LGCNN, based on convolutional neural network (CNN), which enables rapid drug screening by integrating features of drug molecular structures and drug-target interactions at both local and global (LG) levels. Experimental results show that LGCNN exhibits better performance compared to other state-of-the-art classification methods under limited data. In addition, LGCNN was applied to anti-SARS-CoV-2 drug screening to realize therapeutic drug mining against COVID-19. LGCNN transcends the limitations of traditional models for predicting interactions between single drug targets and shows new advantages in predicting multi-target drug-target interactions. Notably, the cross-coronavirus generalizability of the model is also implied by the analysis of targets, drugs, and mechanisms in the prediction results. In conclusion, LGCNN provides new ideas and methods for rapid drug screening in emergency situations where data are scarce.

Privately Answering Queries on Skewed Data via Per-Record Differential Privacy

We consider the problem of the private release of statistics (like payroll) where it is critical to preserve the contribution made by a small number of outlying large entities. We propose a privacy formalism, per-record zero concentrated differential privacy (PzCDP), where the privacy loss associated with each record is a public function of that record's value. Unlike other formalisms which provide different privacy losses to different records, PzCDP's privacy loss depends explicitly on the confidential data. We define our formalism, derive its properties, and propose mechanisms which satisfy PzCDP that are uniquely suited to publishing skewed or heavy-tailed statistics, where a small number of records contribute substantially to query answers. This targeted relaxation helps overcome the difficulties of applying standard DP to these data products.

Methodology of Data Popularity Forecasting in High-Energy Physics Experiments on Unbalanced and Irregular Time-series Data

Methodology of Data Popularity Forecasting in High-Energy Physics Experiments on Unbalanced and Irregular Time-series Data

Comparative analysis of accounting information quality amongst French, Brazilian and Latin American companies

Purpose: This work aimed to compare the Accounting Information Quality (AIQ) of publicly traded companies in markets with different levels of economic development, represented by France, Brazil and other Latin American countries (these together). Methodology: The methodological approach was quantitative, using descriptive and exploratory statistics, and regression with unbalanced panel data. The data used encompassed the following periods: between 1987 and 2017, for the French market; between 1995 and 2017, for the Brazilian market; and between 1989 and 2015, for the Latin American market. Results: The results pointed to structural differences in AIQ in financial statements between the markets analyzed, indicating that the more developed market has better AIQ. Contributions of the Study: As a practical contribution, this research can support investor decisions, while the academic knowledge generated reveals that markets in different economic contexts have different levels of AIQ.

Nghiên cứu tác động của sức mạnh thị trường tới sự ổn định của các ngân hàng thương mại Việt Nam

This paper investigates the impact of market power on the stability of Vietnamese commercial banks using a quarterly unbalanced panel data set of 22 commercial banks in the period from Q2:2015 to Q3:2023, combined with panel data analysis models. Research results show that market power has a negative impact on the stability of Vietnamese commercial banks, implying that a more competitive banking market will have higher stability. In addition, a higher non-performing loan ratio, loan-to-deposit ratio and inflation also make banks less stable, as these factors increase bankruptcy risk, liquidity risk, and credit risk. On that basis, this paper proposes some policy recommendations to improve the stability of Vietnamese commercial banks, including creating a favorable banking business environment, encouraging diversification and innovation in the banking market (such as supporting small banks to increase capital or establish FinTech in a controlled manner). In addition, commercial banks need to proactively manage their internal factors such as credit risk and loan to deposit ratio to ensure stability.

Trustworthy Fault Diagnosis Method Based on Belief Rule Base With Multisource Uncertain Information for Vehicle

This paper develops a new trustworthy fault diagnosis (TFD) method based on belief rule base expert system with combining multi-source uncertainty information for vehicle, such as flying machine. To improve the performance of fault diagnosis for vehicle, a new belief rule base expert system with multi-source uncertainty information (BRB-MU) is developed, where the influence of the random environment disturbance is addressed by introducing uncertain factors of input characteristics. In BRB-MU, the uncertain factors of input characteristics are calculated by the signal to noise ratio based method. The new developed BRB-MU model aims to handle three problems: unbalanced observation data of system states, uncertain expert knowledge and random environment disturbance. The first two problems are solved by the combination of expert knowledge and observation data. In parallel, the third problem is handled by the new introduced uncertain factors based on singular value decomposition (SVD) in BRB-MU. To quantitatively analyze the influence of the uncertain information to the TFD output, the traceability analysis of BRB-MU is conduced that can provide support for decision making of vehicle optimization design. It is analyzed based on the modeling traceability and parameters interpretability. To demonstrate the effectiveness of the TFD method, an experiment illustration is conduced.

FedLAP-DP: Federated Learning by Sharing Differentially Private Loss Approximations

Conventional gradient-sharing approaches for federated learning (FL), such as FedAvg, rely on aggregation of local models and often face performance degradation under differential privacy (DP) mechanisms or data heterogeneity, which can be attributed to the inconsistency between the local and global objectives. To address this issue, we propose FedLAP-DP, a novel privacy-preserving approach for FL. Our formulation involves clients synthesizing a small set of samples that approximate local loss landscapes by simulating the gradients of real images within a local region. Acting as loss surrogates, these synthetic samples are aggregated on the server side to uncover the global loss landscape and enable global optimization. Building upon these insights, we offer a new perspective to enforce record-level differential privacy in FL. A formal privacy analysis demonstrates that FedLAP-DP incurs the same privacy costs as typical gradient-sharing schemes while achieving an improved trade-off between privacy and utility. Extensive experiments validate the superiority of our approach across various datasets with highly skewed distributions in both DP and non-DP settings. Beyond the promising performance, our approach presents a faster convergence speed compared to typical gradient-sharing methods and opens up the possibility of trading communication costs for better performance by sending a larger set of synthetic images. The source is available at https://github.com/a514514772/FedLAP-DP.

A Study of the Prevalence of Diabetic Retinopathy in Patients With Ischemic Heart Disease and Diabetes Mellitus.

Background Diabetes mellitus is one of the most important and common chronic diseases worldwide and is expected to increase in prevalence. Diabetic retinopathy (DR) is one of the most prevalent microvascular sequelae of diabetes mellitus (DM), and ischemic heart disease is a macrovascular sequela. This study was conducted to find out the relation between the degree of DR and ischemic heart disease severity in Indian patients. Materials and methods This cross-sectional, descriptive, hospital-based study was conducted in the ophthalmology department at Dr. D. Y. Patil Medical Hospital, Pune, Maharashtra, India, from September 2022 to June 2024. A total of 200 eyes from 100 patients who were diagnosed with cases of ischemic heart disease and diabetes mellitus were included in the study. Patients with corneal pathology like endothelial dystrophies, corneal degenerations, corneal scars, or trauma preventing good visualization of the posterior segment were excluded from the study. Patients with active uveitis, patients with a history of undergoing any previous vitreoretinal surgery or laser procedures, non-compliant patients, patients not willing to undergo the procedure, or those not consenting to the study were also excluded. Written informed consent was obtained from each patient. Data was entered in Microsoft Exceland statistical analysis was done using IBM Corp. Released 2019. IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp. As the continuous variables showed a skewed distribution, we used the Mann-Whitney test and the Kruskal-Wallis test to test the significance of the difference between continuous and categorical variables. A chi-square test was employed to check the association between categorical variables. Significance was assumed at an alpha error of 5%. Results The prevalence of diabetic retinopathy was found to be 95%. The mean age of patients with DR and patients with no diabetic retinopathy was 58.38 and 59.40 years, respectively, with the majority of the patients being in the age group of 60-69 years (46%). The majority of the patients were males (65%), while 35% were females. There was a significant association between the severity of diabetic retinopathy and the higher HbA1c levels, the use of insulin as a treatment modality, and the higher blood sugar levels in our study population. It was observed that the patients in our study with an ejection fraction of <40% had significantly higher severity of diabetic retinopathy in the form of PDR and high-risk PDR. The severity of the DR was directly correlated with the severity of IHD in our study, with most of the IHD patients with a 40-60% ejection fraction having moderate NPDR and patients with a >60% ejection fraction having mild or moderate NPDR. Conclusion The prevalence of diabetic retinopathy among the IHD patients with diabetes was 95% in our study, with moderate NPDR being the most common stage of DR seen among the patients. It was observed that more severe stages of diabetic retinopathy were seen in patients who were on treatment with insulin than in patients who were on treatment with OHA. Severe stages of diabetic retinopathy were associated with higher blood sugar levels (BSL) and higher glycated hemoglobin levels. In the present study, it was observed that a lower ejection fraction (<40%), which is a marker of reduced cardiac function, was associated with more severe stages of diabetic retinopathy.

Probability density and information entropy of machine learning derived intracranial pressure predictions.

Even with the powerful statistical parameters derived from the Extreme Gradient Boost (XGB) algorithm, it would be advantageous to define the predicted accuracy to the level of a specific case, particularly when the model output is used to guide clinical decision-making. The probability density function (PDF) of the derived intracranial pressure predictions enables the computation of a definite integral around a point estimate, representing the event's probability within a range of values. Seven hold-out test cases used for the external validation of an XGB model underwent retinal vascular pulse and intracranial pressure measurement using modified photoplethysmography and lumbar puncture, respectively. The definite integral ±1 cm water from the median (DIICP) demonstrated a negative and highly significant correlation (-0.5213±0.17, p< 0.004) with the absolute difference between the measured and predicted median intracranial pressure (DiffICPmd). The concordance between the arterial and venous probability density functions was estimated using the two-sample Kolmogorov-Smirnov statistic, extending the distribution agreement across all data points. This parameter showed a statistically significant and positive correlation (0.4942±0.18, p< 0.001) with DiffICPmd. Two cautionary subset cases (Case 8 and Case 9), where disagreement was observed between measured and predicted intracranial pressure, were compared to the seven hold-out test cases. Arterial predictions from both cautionary subset cases converged on a uniform distribution in contrast to all other cases where distributions converged on either log-normal or closely related skewed distributions (gamma, logistic, beta). The mean±standard error of the arterial DIICP from cases 8 and 9 (3.83±0.56%) was lower compared to that of the hold-out test cases (14.14±1.07%) the between group difference was statistically significant (p<0.03). Although the sample size in this analysis was limited, these results support a dual and complementary analysis approach from independently derived retinal arterial and venous non-invasive intracranial pressure predictions. Results suggest that plotting the PDF and calculating the lower order moments, arterial DIICP, and the two sample Kolmogorov-Smirnov statistic may provide individualized predictive accuracy parameters.

Validation and Implementation of a Somatic-Only Tumor Exome for Routine Clinical Application

Next-generation sequencing-based genomic testing is standard of care for tumor workflows. However, its application across different institutions continues to be challenging given the diversity of needs and resource availability among different institutions globally. Moreover, the use of a variety of different panels, including those from a few individual genes to those involving hundreds of genes, results in a relatively skewed distribution of care for patients. It is imperative to obtain a higher level of standardization without having to be restricted to specific kits or requiring repeated validations, which are generally expensive. We show the validation and clinical implementation of the DH-CancerSeq assay, a tumor-only whole-exome-based sequencing assay with integrated informatics, while providing similar input requirements, sensitivity, and specificity to a previously validated targeted gene panel and maintaining similar turnaround times for patient care.

Investigating the effectiveness of commercially available mouthwash on SARS-CoV-2 in vivo using viable virus titre as the primary outcome. A randomised controlled trial.

This multi-arm, parallel group, single-blinded randomised controlled trial aimed to assess three commercially available mouthwashes effectiveness against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This manuscript has been written in accordance with the CONSORT statement. Methods. Eligible participants were SARS-CoV-2 positive with a positive test in the last 72 h. All participants had mild to moderate symptoms and could provide five saliva samples over a 60 min period. Participants delivered a baseline saliva sample and then used a mouthwash as per manufacturer's instructions. They provided further saliva samples at minute 1, 10, 30 and 60. Participants were randomised to one of four groups; OraWize+, Total Care Listerine, Cool Mint Listerine and water (control). The lab-based research team were blind to the intervention. The research question was: can SARS-CoV-2 be rendered inactive in saliva by using a mouthwash and how long does this effect last? The primary outcome was the amount of viable infectious SARS-CoV-2 virus in the sample, compared to the baseline sample. The secondary outcome measure was the amount of genetic material from the SARS-CoV-2 virus in the sample, measured via PCR testing. Results. In total 100 participants were recruited (25 per group). Eight participants did not receive the allocated intervention and did not have saliva samples collected. There were no adverse events. In total 42 of the 92 participants had viable virus which could be cultured at baseline. Statistical analysis of the primary outcome was not advised due to the reduced level of viable virus at baseline and the positive skewness present in the distribution of log10(titre) data. Observational data of the primary outcome measure is presented. Analysis of the secondary outcome PCR measure showed that there was strong evidence for a decrease in SARS-CoV-2 RNA levels compared to water for all mouthwashes after 1 min, OraWize+ -0.49 (-0.92, -0.05), p-value 0.029, Cool Mint Listerine -0.81 (-1.25, -0.38), p-value<0.001, Total Care Listerine -1.05 (-1.48, -0.62), p-value<0.001. For the remaining timepoints there was generally no evidence of virus level reduction compared to water although there is weak evidence for a decrease at ten minutes using Total Care Listerine -0.44 (-0.88, 0.01), p-value 0.053. Conclusion. The three mouthwashes included in this trial observationally demonstrated a reduction in virus titre level 1 min after use, with virus levels normalising up to 60 min compared to the control. Although an interesting observation, this result could not be statistically analysed. Using the secondary outcome PCR measure all three included mouthwashes reduced virus levels compared to water at 1 min and these results were statistically significant. Clinically this result does not support the use of the included mouthwashes to reduce SARS-CoV-2 levels in saliva.

A Study on the Measurement of Relative Poverty in Developing Countries with Large Populations

Under the goal of global poverty eradication, it has become a forward-looking research aim to establish relative poverty criteria and identify people experiencing relative poverty in countries with different demographic characteristics. This paper introduces a new method to measure the relative poverty standard, which is to use the fuzzy decision tree algorithm to objectively estimate the relative poverty standard. The advantage of this algorithm lies in (1) it not only maintaining the regression idea of measuring absolute poverty, but also emphasizing the nonlinearity when the demand is increasing, which can reflect the change in human needs. (2) It overcomes the division of the traditional method which clearly distinguishes between those experiencing poverty and those who are not by means of a subjective threshold, and it also avoids the subjectivity of the selection of multidimensional indicators. (3) It overcomes the problems of data skewness and extreme value issues that traditional methods have, and can exhibit multi-dimensional characteristics. (4) Most importantly, this method can overcome the gap problem caused by the complex population structure in developing countries with huge populations, and is more adaptable under big data conditions than traditional methods. Taking China as an example, using data from the China Household Finance Survey for validation, the validation results show that the relative poverty standard in China in 2019 can be approximately delineated as 5288.5 RMB; this result is higher than the absolute poverty standard line delineated in China in that year, lower than the relative poverty standard line measured using the proportion method, and it can satisfy the average per capita food, tobacco, and alcohol consumption expenditure of Chinese residents in that year. Thus, compared with other methods, the fuzzy decision tree algorithm can better match the identification of relative poverty in developing countries with large populations.

Boosting semi‐supervised learning under imbalanced regression via pseudo‐labeling

SummaryImbalanced samples are widespread, which impairs the generalization and fairness of models. Semi‐supervised learning can overcome the deficiency of rare labeled samples, but it is challenging to select high‐quality pseudo‐label data. Unlike discrete labels that can be matched one‐to‐one with points on a numerical axis, labels in regression tasks are consecutive and cannot be directly chosen. Besides, the distribution of unlabeled data is imbalanced, which easily leads to an imbalanced distribution of pseudo‐label data, exacerbating the imbalance in the semi‐supervised dataset. To solve this problem, this article proposes a semi‐supervised imbalanced regression network (SIRN), which consists of two components: A, designed to learn the relationship between features and labels (targets), and B, dedicated to learning the relationship between features and target deviations. To measure target deviations under imbalanced distribution, the target deviation function is introduced. To select continuous pseudo‐labels, the deviation matching strategy is designed. Furthermore, an adaptive selection function is developed to mitigate the risk of skewed distributions due to imbalanced pseudo‐label data. Finally, the effectiveness of the proposed method is validated through evaluations of two regression tasks. The results show a great reduction in predicted value error, particularly in few‐shot regions. This empirical evidence confirms the efficacy of our method in addressing the issue of imbalanced samples in regression tasks.