RESUMEN
Adversarial training is effective in improving the robustness of deep neural networks. However, existing studies still exhibit significant drawbacks in terms of the robustness, generalization, and fairness of models. In this study, we validate the importance of different perturbation directions (i.e., adversarial and anti-adversarial) and bounds from both theoretical and practical perspectives. The influence of adversarial training on deep learning models in terms of fairness, robustness, and generalization is theoretically investigated under a more general perturbation scope that different samples can have different perturbation directions and varied perturbation bounds. Our theoretical explorations suggest that combining adversaries and anti-adversaries with varied bounds in training can be more effective in achieving better fairness among classes and a better tradeoff among robustness, accuracy, and fairness in some typical learning scenarios compared with standard adversarial training. Inspired by our theoretical findings, a more general learning objective that combines adversaries and anti-adversaries with varied bounds on each training sample is presented. To solve this objective, two adversarial training frameworks based on meta-learning and reinforcement learning are proposed, in which the perturbation direction and bound for each sample are determined by its training characteristics. Furthermore, the role of the combination strategy with varied bounds is explained from a regularization perspective. Extensive experiments under different learning scenarios verify our theoretical findings and the effectiveness of the proposed methodology.
RESUMEN
An adversarial attack is typically implemented by solving a constrained optimization problem. In top-k adversarial attacks implementation for multi-label learning, the attack failure degree (AFD) and attack cost (AC) of a possible attack are major concerns. According to our experimental and theoretical analysis, existing methods are negatively impacted by the coarse measures for AFD/AC and the indiscriminate treatment for all constraints, particularly when there is no ideal solution. Hence, this study first develops a refined measure based on the Jaccard index appropriate for AFD and AC, distinguishing the failure degrees/costs of two possible attacks better than the existing indicator function-based scheme. Furthermore, we formulate novel optimization problems with the least constraint violation via new measures for AFD and AC, and theoretically demonstrate the effectiveness of weighting slack variables for constraints. Finally, a self-paced weighting strategy is proposed to assign different priorities to various constraints during optimization, resulting in larger attack gains compared to previous indiscriminate schemes. Meanwhile, our method avoids fluctuations during optimization, especially in the presence of highly conflicting constraints. Extensive experiments on four benchmark datasets validate the effectiveness of our method across different evaluation metrics.
RESUMEN
BACKGROUND: Obesity and hypertension are major risk factors for cardiovascular diseases that affect millions of people worldwide. Both conditions are associated with chronic low-grade inflammation, which is mediated by adipokines such as adiponectin. Adiponectin is the most abundant adipokine that has a beneficial impact on metabolic and vascular biology, while high serum concentrations are associated with some syndromes. This "adiponectin paradox" still needs to be clarified in obesity-associated hypertension. The aim of this study was to investigate how adiponectin affects blood pressure, inflammation, and metabolic function in obesity hypertension using a Chinese adult case-control study. METHODS: A case-control study that had finished recruiting 153 subjects divided as four characteristic groups. Adiponectin serum levels were tested by ELISA in these subjects among these four characteristic Chinese adult physical examination groups. Waist circumference (WC), body mass index (BMI), systolic blood pressure (SB), diastolic blood pressure (DB), and other clinical laboratory data were collected. Analyzation of correlations between the research index and differences between groups was done by SPSS. RESULTS: Serum adiponectin levels in the| normal healthy group (NH group) were significantly higher than those in the newly diagnosed untreated just-obesity group (JO group), and negatively correlated with the visceral adiposity index. With multiple linear egression analysis, it was found that, for serum adiponectin, gender, serum albumin (ALB), alanine aminotransferase (ALT) and high-density lipoprotein cholesterol (HDLC) were the significant independent correlates, and for SB, age and HDLC were the significant independent correlates, and for DB, alkaline phosphatase (ALP) was the significant independent correlate. The other variables did not reach significance in the model. CONCLUSIONS: Our study reveals that adiponectin's role in obesity-hypertension is multifaceted and is influenced by the systemic metabolic homeostasis signaling axis. In obesity-related hypertension, compensatory effects, adiponectin resistance, and reduced adiponectin clearance from impaired kidneys and liver all contribute to the "adiponectin paradox".
Asunto(s)
Adiponectina , Hipertensión , Adulto , Humanos , Estudios de Casos y Controles , Hipertensión/diagnóstico , Obesidad/complicaciones , Obesidad/diagnóstico , HDL-Colesterol , Inflamación , China/epidemiologíaRESUMEN
To examine whether circulating interleukin-6 (IL-6) levels (CirIL6) have a causal effect on blood pressure using Mendelian randomization (MR) methods. We used data from genome-wide association studies (GWAS) of European ancestry to obtain genetic instruments for circulating IL-6 levels and blood pressure measurements. We applied several robust MR methods to estimate the causal effects and to test for heterogeneity and pleiotropy. We found that circulating IL-6 had a significant positive causal effect on systolic blood pressure (SBP) and pulmonary arterial hypertension (PAH), but not on diastolic blood pressure (DBP) or hypertension. We found that as CirIL6 genetically increased, SBP increased using Inverse Variance Weighted (IVW) method (for ukb-b-20175, ß = 0.082 with SE = 0.032, P = 0.011; for ukb-a-360, ß = 0.075 with SE = 0.031, P = 0.014) and weighted median (WM) method (for ukb-b-20175, ß = 0.061 with SE = 0.022, P = 0.006; for ukb-a-360, ß = 0.065 with SE = 0.027, P = 0.014). Moreover, CirIL6 may be associated with an increased risk of PAH using WM method (odds ratio (OR) = 15.503, 95% CI, 1.025-234.525, P = 0.048), but not with IVW method. Our study provides novel evidence that circulating IL-6 has a causal role in the development of SBP and PAH, but not DBP or hypertension. These findings suggest that IL-6 may be a potential therapeutic target for preventing or treating cardiovascular diseases and metabolic disorders. However, more studies are needed to confirm the causal effects of IL-6 on blood pressure and to elucidate the underlying mechanisms and pathways.
Asunto(s)
Hipertensión , Interleucina-6 , Humanos , Presión Sanguínea/genética , Interleucina-6/genética , Estudio de Asociación del Genoma Completo , Análisis de la Aleatorización Mendeliana , Hipertensión/genéticaRESUMEN
Treating each training sample unequally is prevalent in many machine-learning tasks. Numerous weighting schemes have been proposed. Some schemes take the easy-first mode, whereas others take the hard-first one. Naturally, an interesting yet realistic question is raised. Given a new learning task, which samples should be learned first, easy or hard? To answer this question, both theoretical analysis and experimental verification are conducted. First, a general objective function is proposed and the optimal weight can be derived from it, which reveals the relationship between the difficulty distribution of the training set and the priority mode. Two novel findings are subsequently obtained: besides the easy-first and hard-first modes, there are two other typical modes, namely, medium-first and two-ends-first; the priority mode may be varied if the difficulty distribution of the training set changes greatly. Second, inspired by the findings, a flexible weighting scheme (FlexW) is proposed for selecting the optimal priority mode when there is no prior knowledge or theoretical clues. The four priority modes can be flexibly switched in the proposed solution, thus suitable for various scenarios. Third, a wide range of experiments is conducted to verify the effectiveness of our proposed FlexW and further compare the weighting schemes in different modes under various learning scenarios. On the basis of these works, reasonable and comprehensive answers are obtained for the easy-or-hard question.
RESUMEN
BACKGROUND: Obesity and hypertension are major risk factors for cardiovascular diseases that affect millions of people worldwide. Both conditions are associated with chronic low-grade inflammation, which is mediated by cytokines such as interleukin-6 (IL-6). IL-6 is a multifunctional cytokine that can have pro-inflammatory or anti-inflammatory effects depending on the context. The exact role of IL-6 in obesity-associated hypertension is unclear. OBJECTIVE: To investigate how IL-6 affects blood pressure, inflammation, and metabolic function in obesity-hypertension using a Chinese adult case-control study. METHODS: A total of 153 participants were sorted into four subgroups according to their body mass index (BMI) and blood pressure (BP): normal healthy group (NH), just obesity group (JO), just-hypertension group (JH), and obesity-hypertension group (OH). Serum IL-6 concentrations were measured by Enzyme-linked Immunosorbent Assay (ELISA) and their correlations with anthropometric and laboratory parameters and their differences across the subgroups were examined. Multiple linear regression analysis was performed to identify the predictors of serum IL-6 concentrations in each group. RESULTS: Serum IL-6 concentrations were higher in NH group than in JO group and correlated positively with diastolic blood pressure in NH and JO groups, but not in JH and OH groups. Serum IL-6 concentrations also correlated with albumin in NH group, alkaline phosphatase in JO group, serum creatinine and fasting blood glucose in JH group. The influencing factors of serum IL-6 concentrations varied among the four groups, with gender, diastolic blood pressure and albumin being significant predictors in NH group, alkaline phosphatase in JO group, age and serum creatinine in JH group, and none in OH group. CONCLUSIONS: These results suggest that IL-6 may play diverse effects in the pathogenesis of obesity- hypertension, depending on the presence or absence of obesity and hypertension. Further studies are needed to elucidate the underlying mechanisms of IL-6 signaling and function in these diseases.
Asunto(s)
Hipertensión , Interleucina-6 , Humanos , Adulto , Estudios de Casos y Controles , Fosfatasa Alcalina , Creatinina , Pueblos del Este de Asia , Obesidad , Citocinas , Índice de Masa Corporal , Inflamación , AlbúminasRESUMEN
Features, logits, and labels are the three primary data when a sample passes through a deep neural network (DNN). Feature perturbation and label perturbation receive increasing attention in recent years. They have been proven to be useful in various deep learning approaches. For example, (adversarial) feature perturbation can improve the robustness or even generalization capability of learned models. However, limited studies have explicitly explored for the perturbation of logit vectors. This work discusses several existing methods related to class-level logit perturbation. A unified viewpoint between regular/irregular data augmentation and loss variations incurred by logit perturbation is established. A theoretical analysis is provided to illuminate why class-level logit perturbation is useful. Accordingly, new methodologies are proposed to explicitly learn to perturb logits for both the single-label and multilabel classification tasks. Meta-learning is also leveraged to determine the regular or irregular augmentation for each class. Extensive experiments on benchmark image classification datasets and their long-tail versions indicated the competitive performance of our learning method. As it only perturbs on logit, it can be used as a plug-in to fuse with any existing classification algorithms. All the codes are available at https://github.com/limengyang1992/lpl.
RESUMEN
BACKGROUND: University students with suicidal ideation are at high risk of suicide, which constitutes a significant social and public health problem in China. However, little is known about the prevalence and associated risk factors of suicidal ideation among first-year university students in China, especially during the COVID-19 pandemic. OBJECTIVES: To investigate the prevalence of suicidal ideation and its factors in first-year Chinese university students from a vocational college in Zhejiang during the COVID-19 pandemic. METHODS: Using a cluster sampling technique, a university-wide survey was conducted of 686 first-year university students from Hangzhou in March 2020 using University Personality Inventory (UPI). UPI includes an assessment for suicidal ideation and possible risk factors. Suicidal ideation prevalence was calculated for males and females. Univariate analysis and multivariable logistic regression models were conducted, adjusting for age and sex. Analyses were carried out using the SPSS version 22.0 software. RESULTS: The prevalence of 12-month suicidal ideation among first-year university students during March 2020 was 5.2%, and there was no significant difference between males and females (4.8% vs. 6.0%, x2 = 0.28, p = 0.597). Multivariable logistic regression analysis identified social avoidance (B = 0.78, OR = 2.17, p < 0.001) and emotional vulnerability (B = 0.71, OR = 2.02, p < 0.001) as positively associated with suicidal ideation. CONCLUSIONS: Social avoidance and emotional vulnerabilities are unique factors associated with greater suicidal ideation among first-year university students during the COVID-19 pandemic. UPI serves as a validated tool to screen suicide risks among Chinese university students. Encouraging social engagement and improving emotional regulation skills are promising targets to reduce suicidal ideation among first-year university students.
Asunto(s)
COVID-19 , Inventores , COVID-19/epidemiología , China/epidemiología , Femenino , Humanos , Masculino , Pandemias , Personalidad , Prevalencia , Factores de Riesgo , Estudiantes/psicología , Ideación Suicida , UniversidadesRESUMEN
BACKGROUND: According to the validation literature on items of Young's Internet Addiction Test (IAT), this study rephrased disputable items to improve the psychometric properties of this Chinese version of IAT and identify the presence of differential item function (DIF) among demographic and Internet use factors; detect the effect of demographic and Internet use factors on IAT after adjusting for DIF. METHODS: An online questionnaire was distributed to college students in Zhe Jiang province in two stage. The 1st phase study collected 384 valid responses to examine the quality of IAT items by using Rasch Model analysis and exploring factor analysis (EFA). The online questionnaire was modified according to the 1st phase study and distributed online for the 2nd phase study which collected a total of 1131 valid responses. The 2nd phase study applied confirmatory factor analysis (CFA) and a multiple indicator multiple causes (MIMIC) model to verify the construct of IAT, potential effect of covariates on IAT latent factors, as well as the effect of differential item functioning (DIF). RESULTS: Rasch model analysis in the 1st phase study indicated a 5-point rating scale was performed better, no sever misfit was found on item. The overall property of Chinese version IAT with the 5-point scale was good to excellent person and item separation (2.66 and 6.86). A three-factor model was identified by EFA. In the 2nd phase study, IAT 13 were detected with DIF for gender in MIMIC model. After correcting DIF effect, the significant demographic and Internet use factors on IAT were time spent online per day, year 3, year 2, general users. CONCLUSION: Item improvement was efficient that the problematic items found in literature was performed good in this study. The overall psychometric property of this Chinese version IAT was good with limited DIF effect in one item. Item improvement on IAT13 was encouraged in the future study to avoid gender bias and benefit for epidemiology on PIU.
Asunto(s)
Conducta Adictiva , Trastorno de Adicción a Internet , Conducta Adictiva/diagnóstico , Conducta Adictiva/epidemiología , Demografía , Femenino , Humanos , Uso de Internet , Masculino , Psicometría , SexismoRESUMEN
OBJECTIVE: To explore the role of leptin in the onset and development of obesity-associated hypertension. SUBJECTS AND METHODS: A case-control study that had finished recruiting 153 subjects divided as four characteristic groups. Leptin serum levels were tested by ELISA in these subjects among these four characteristic Chinese adult physical examination groups. Waist circumference (WC), body mass index (BMI), systolic blood pressure (SB), diastolic blood pressure (DB), and other clinical laboratory data were collected. Analyzation of correlations between the research index and differences between groups was done by SPSS. RESULTS: Serum leptin levels statistically significantly positively correlated with BMI and WC, and negatively with the HDLC (high-density lipoprotein cholesterol), even after adjustment for age and gender. There was no significant difference in the serum leptin levels between the normal healthy group (NH group) and the newly diagnosed untreated just-hypertension group (JH group). And the same is between the newly diagnosed untreated obesity-hypertension group (OH group) and the newly diagnosed untreated just-obesity group (JO group). Multiple linear regression analysis indicated BMI and gender as significant independent correlates of serum leptin. CONCLUSIONS: These results show leptin may not be essential but play an additive effect in the development of obesity-associated hypertension. Leptin may only play an additive effect role in the intricate interwoven network of regulators contributing to the development of hypertension in obese patients.
Asunto(s)
Hipertensión , Leptina , Adulto , Índice de Masa Corporal , Estudios de Casos y Controles , China , Humanos , Hipertensión/complicaciones , ObesidadRESUMEN
OBJECTIVE: To explore the role of adiponectin-resistin (AR) index as a better indicator of obesity-related hypertension. METHOD(S): This study continued a case control study that had finished recruiting 153 subjects divided as four characteristic groups. Fasting serum resistin levels (FSR) and Fasting serum adiponectin levels (FSA) were tested by ELISA. And, other related anthropometric clinical and metabolic data were collected. Analyzation on correlations between research index and differences between groups were done by SPSS. AR index's performance was also validated by the receiver operating characteristic (ROC) curves, the net reclassification improvement (NRI), and the integrated discrimination improvement (IDI). RESULT(S): The AR index was defined as 1+ log10(R0)-log10(A0). AUC of the AR index was 0.660 and NRI and IDI indicated AR index outperformed FSA alone. AR index statistically significantly negatively correlated with SB and DB and positively with ALB and SCR. AR index was statistically significantly different between the NH group and OH group and more specific than FSR alone as a biomarker of obesity-related hypertension. CONCLUSION(S): The AR index was more strongly associated with increased risk of obesity-related hypertension than the solely index of FSR or FSA and was useful for early diagnosis of obesity-related hypertension.
Asunto(s)
Hipertensión , Resistencia a la Insulina , Adiponectina , Adulto , Índice de Masa Corporal , Estudios de Casos y Controles , China/epidemiología , Humanos , Hipertensión/diagnóstico , Hipertensión/etiología , Obesidad/complicaciones , ResistinaRESUMEN
Sentiment analysis is a key component in various text mining applications. Numerous sentiment classification techniques, including conventional and deep-learning-based methods, have been proposed in the literature. In most existing methods, a high-quality training set is assumed to be given. Nevertheless, constructing a high-quality training set that consists of highly accurate labels is challenging in real applications. This difficulty stems from the fact that text samples usually contain complex sentiment representations, and their annotation is subjective. We address this challenge in this study by leveraging a new labeling strategy and utilizing a two-level long short-term memory network to construct a sentiment classifier. Lexical cues are useful for sentiment analysis, and they have been utilized in conventional studies. For example, polar and negation words play important roles in sentiment analysis. A new encoding strategy, that is, ρ -hot encoding, is proposed to alleviate the drawbacks of one-hot encoding and, thus, effectively incorporate useful lexical cues. Moreover, the sentimental polarity of a word may change in different sentences due to label noise or context. A flipping model is proposed to model the polar flipping of words in a sentence. We compile three Chinese datasets on the basis of our label strategy and proposed methodology. Experiments demonstrate that the proposed method outperforms state-of-the-art algorithms on both benchmark English data and our compiled Chinese data.
Asunto(s)
Semántica , Análisis de Sentimientos , Algoritmos , Minería de DatosRESUMEN
Objective: To explore the role of resistin in the onset and development of obesity-related hypertension.Methods: Resistin serum levels were tested by ELISA in 153 adult subjects among four characteristic Chinese adult physical examination groups. Waist circumference (WC), body mass index (BMI), systolic blood pressure (SB), diastolic blood pressure (DB), and other clinical laboratory data were collected. Following, correlations between research index and differences between groups were analyzed using SPSS.Results: Serum resistin levels statistically significantly negatively correlated with SB, DB and BMI, but statistically significantly positively correlated with serum creatinine (SCR) and serum albumin (ALB), even after adjustment for age and/or gender. The serum level of resistin in the normal healthy subject group (NH) was higher than in other groups.Conclusions: Resistin's role in the onset of obesity-related hypertension may be more important than what has been previously assumed. More pathway substances in the early onset of obesity-related hypertension should be tested.Abbreviations: WC, waist circumference; GGT, Gamma-glutamyltransferase; ALB, Albumin; ALT, Alanine aminotransferase; LDL, Low density lipoprotein cholesterol; TG, Triglyceride; HDLC, High density lipoprotein cholesterol; FA Fructosamine; SCR, serum creatinine; IB, Indirect bilirubin; ALP, Alkaline phosphatase; CB, Conjugated bilirubin; UREA, Urea; Ua, Uric acid; FBG, fasting blood glucose; TC, Total cholesterol; TB, Total bilirubin; TP, Total protein; TC/HDLC, TC/HDLC ratio; SB, systolic blood pressure; DB, diastolic blood pressure.
Asunto(s)
Pueblo Asiatico , Hipertensión/sangre , Obesidad/sangre , Resistina/sangre , Adulto , Biomarcadores/sangre , Presión Sanguínea , Índice de Masa Corporal , China , Femenino , Humanos , Hipertensión/complicaciones , Hipertensión/fisiopatología , Masculino , Persona de Mediana Edad , Obesidad/complicaciones , Circunferencia de la CinturaRESUMEN
OBJECTIVE: The objective was to investigate the expression of macrophage migration inhibitory factor (MIF) in non-small cell lung cancer (NSCLC), as well as the effects of macrophage MIF on tumor cells. METHODS: The human NSCLC cell strains H358 and H524 were selected as research objects. The Real-Time Polymerase Chain Reaction (RT-PCR) and Western Blot were utilized to detect the expression levels of MIF in human NSCLC cell strains. The lentiviral plasmid was utilized for MIF-mRNA interference. The expression levels of MIF before and after transfection were compared. The cell strains were cultured and proliferated for cell count and comparison. RESULTS: H358 showed MIF high expression while H524 showed MIF low expression. Once the H358 cells were constructed as silent MIF expression, compared with the original H358 cells, the difference was statistically significant. Once the H524 cells were constructed as high MIF expression, compared with original H524 cells, the difference was statistically significant. Being cultured for respective 3, 5, and 7 days, the transfected H358 cells showed a significant decrease in proliferative activity compared with original H358 cells, while the transfected H524 cells showed a significant increase in proliferative activity compared with original H524 cells. CONCLUSION: MIF has high expression in H358 cells while low expression in H524 cells. The expression of MIF could enhance the proliferative activity of NSCLC tumor cells.
RESUMEN
BACKGROUND: The obesity-hypertension pathogenesis is complex. From the phenotype to molecular mechanism, there is a long way to clarify the mechanism. To explore the association between obesity and hypertension, we correlate the phenotypes such as the waist circumference (WC), body mass index (BMI), systolic blood pressure (SB), and diastolic blood pressure (DB) with the clinical laboratory data between four specific Chinese adult physical examination groups (newly diagnosed untreated just-obesity group, newly diagnosed untreated obesity-hypertension group, newly diagnosed untreated just-hypertension group, and normal healthy group), and the results may show something. OBJECTIVE: To explore the mechanisms from obesity to hypertension by analyzing the correlations and differences between WC, BMI, SB, DB, and other clinical laboratory data indices in four specific Chinese adult physical examination groups. METHODS: This cross-sectional study was conducted from September 2012 to July 2014, and 153 adult subjects, 34 women and 119 men, from 21 to 69 years, were taken from four characteristic Chinese adult physical examination groups (newly diagnosed untreated just-obesity group, newly diagnosed untreated obesity-hypertension group, newly diagnosed untreated just-hypertension group, and normal healthy group). The study was approved by the ethics committee of Hangzhou Center for Disease Control and Prevention. WC, BMI, SB, DB, and other clinical laboratory data were collected and analyzed by SPSS. RESULTS: Serum levels of albumin (ALB)ï¼alanine aminotransferase (ALT), low density lipoprotein cholesterol (LDLC), triglyceride (TG), high density lipoprotein cholesterol (HDLC), alkaline phosphatase (ALP), uric acid (Ua), and TC/HDLC (odds ratio) were statistically significantly different between the four groups. WC statistically significantly positively correlated with BMI, ALT, Ua, and serum levels of glucose (GLU), and TC/HDLC, and negatively with ALB, HDLC, and serum levels of conjugated bilirubin (CB). BMI was statistically significantly positively related to ALT, Ua, LDLC, WC, and TC/HDLC, and negatively to ALB, HDLC, and CB. DB statistically significantly positively correlated with ALP, BMI, and WC. SB was statistically significantly positively related to LDLC, GLU, serum levels of fructosamine (FA), serum levels of the total protein (TC), BMI, and WC. CONCLUSION: The negative body effects of obesity are comprehensive. Obesity may lead to hypertension through multiple ways by different percents. GGT, serum levels of gamma glutamyltransferase; ALB, serum levels of albumin; ALT, serum levels of alanine aminotransferase; LDLC, serum levels of low density lipoprotein cholesterol; TG, serum levels of triglyceride; HDLC, serum levels of high density lipoprotein cholesterol; FA, serum levels of fructosamine; S.C.R, serum levels of creatinine; IB, serum levels of indirect bilirubin; ALP, serum levels of alkaline phosphatase; CB, serum levels of conjugated bilirubin; UREA, Urea; Ua, serum levels of uric acid; GLU, serum levels of glucose; TC, serum levels of the total cholesterol; TB, serum levels of the total bilirubin; TP, serum levels of the total protein; TC/HDLC, TC/HDLC ratio.
Asunto(s)
Presión Sanguínea , Índice de Masa Corporal , Hipertensión/fisiopatología , Obesidad/fisiopatología , Circunferencia de la Cintura , Adulto , Anciano , Alanina Transaminasa/sangre , Fosfatasa Alcalina/sangre , Bilirrubina/sangre , Glucemia/metabolismo , China , HDL-Colesterol/sangre , LDL-Colesterol/sangre , Estudios Transversales , Diástole , Femenino , Fructosamina/sangre , Humanos , Hipertensión/sangre , Hipertensión/complicaciones , Masculino , Persona de Mediana Edad , Obesidad/sangre , Obesidad/complicaciones , Fenotipo , Factores de Riesgo , Albúmina Sérica/metabolismo , Sístole , Triglicéridos/sangre , Ácido Úrico/sangre , Adulto Joven , gamma-Glutamiltransferasa/sangreRESUMEN
Obtaining a sufficient number of accurate labels to form a training set for learning a classifier can be difficult due to the limited access to reliable label resources. Instead, in real-world applications, less-accurate labels, such as labels from nonexpert labelers, are often used. However, learning with less-accurate labels can lead to serious performance deterioration because of the high noise rate. Although several learning methods (e.g., noise-tolerant classifiers) have been advanced to increase classification performance in the presence of label noise, only a few of them take the noise rate into account and utilize both noisy but easily accessible labels and less-noisy labels, a small amount of which can be obtained with an acceptable added time cost and expense. In this brief, we propose a learning method, in which not only noisy labels but also auxiliary less-noisy labels, which are available in a small portion of the training data, are taken into account. Based on a flipping probability noise model and a logistic regression classifier, this method estimates the noise rate parameters, infers ground-truth labels, and learns the classifier simultaneously in a maximum likelihood manner. The proposed method yields three learning algorithms, which correspond to three prior knowledge states regarding the less-noisy labels. The experiments show that the proposed method is tolerant to label noise, and outperforms classifiers that do not explicitly consider the auxiliary less-noisy labels.
RESUMEN
BACKGROUND: The study was to investigate the prevalence of metabolic health in subjects with obesity in the Chinese population and to identify the determinants related to metabolic abnormality in obese individuals. METHODS: 5013 subjects were recruited from seven provincial capitals in China. The obesity and metabolic status were classified based on body mass index (BMI) and the number of abnormalities in common components of metabolic syndrome. RESULTS: 27.9% of individuals with obesity were metabolically healthy. The prevalence of the metabolically healthy obese (MHO) phenotype was significantly decreased with age in women (p trend < 0.001), but not significantly in men (p trend = 0.349). Central obesity (odds ratio [OR] = 4.07, 95% confidence interval [CI] = 1.93-8.59), longer sedentary time (OR = 1.97, 95%CI = 1.27-3.06), and with a family history of obesity related diseases (hypertension, diabetes, dyslipidemia) (OR = 1.85, 95%CI = 1.26-2.71) were significantly associated with having metabolic abnormality in obese individuals. Higher levels of physical activity and more fruit/vegetable intake had decreased ORs of 0.67 (95%CI = 0.45-0.98) and 0.44 (95%CI = 0.28-0.70), respectively. CONCLUSION: 27.9% of obese participants are in metabolic health. Central obesity, physical activity, sedentary time, fruits/vegetables intake and family history of diseases are the determinants associated with metabolic status in obesity.
Asunto(s)
Síndrome Metabólico/epidemiología , Obesidad/epidemiología , Adulto , Anciano , Pueblo Asiatico , Índice de Masa Corporal , China/epidemiología , Ciudades/epidemiología , Diabetes Mellitus , Dieta , Dislipidemias , Femenino , Humanos , Hipertensión/epidemiología , Estilo de Vida , Masculino , Persona de Mediana Edad , Obesidad/metabolismo , Obesidad Abdominal/epidemiología , Oportunidad Relativa , PrevalenciaRESUMEN
BACKGROUND: Both environmental and genetic factors are attributable to the incidence of noise-induced hearing loss (NIHL). The purpose of this study was to examine the associations between genetic variations in the EYA4, GRHL2 and DFNA5 genes and the risk to noise-induced hearing loss (NIHL) in a Chinese population. METHODS: A case-control study was conducted with 476 NIHL workers and 475 normal hearing workers matched with gender, years of noise exposure, and intensity of noise exposure. Twelve tag single-nucleotide polymorphisms (SNP) in the EYA4, GRHL2 and DFNA5 genes were genotyped using nanofluidic dynamic arrays on the Fluidigm platform. Multiple logistic regression was used to analyze the associations of genetic variations with NIHL adjusted by age, smoking/drinking status, and cumulative noise exposure and their interactions with noise exposure. RESULTS: The SNPs of rs3777781and rs212769 in the EYA4 gene were significantly associated with NIHL risk. In rs3777781, comparing with the subjects carrying with TT types, the carriers with AT and AA genotypes had the decreased risk of NIHL (OR = 0.721, 95% CI = 0.522 - 0.996). In rs212769, the AG and AA carriers had increased NIHL risk (OR = 1.430, 95% CI = 1.014 - 2.016) compared with the subjects with GG genotype. Rs666026 in the associated GRHL2 gene and rs2521758 in the DFNA5 gene were marginally t associated with NIHL (P = 0.065 and 0.052, respectively). Rs2521758 and rs212769 had significantly interacted with noise exposure (P < 0.05). CONCLUSIONS: Genetic variations in the EYA4, GRHL2 and DFNA5 genes and their interactions with occupational noise exposure may play an important role in the incidence of NIHL.
Asunto(s)
Proteínas de Unión al ADN/genética , Exposición a Riesgos Ambientales , Pérdida Auditiva Provocada por Ruido/epidemiología , Ruido/efectos adversos , Polimorfismo de Nucleótido Simple , Receptores de Estrógenos/genética , Transactivadores/genética , Factores de Transcripción/genética , Adulto , Estudios de Casos y Controles , China , Proteínas de Unión al ADN/metabolismo , Pérdida Auditiva Provocada por Ruido/genética , Humanos , Modelos Logísticos , Masculino , Persona de Mediana Edad , Receptores de Estrógenos/metabolismo , Riesgo , Transactivadores/metabolismo , Factores de Transcripción/metabolismo , Adulto JovenRESUMEN
Horror content sharing on the Web is a growing phenomenon that can interfere with our daily life and affect the mental health of those involved. As an important form of expression, horror images have their own characteristics that can evoke extreme emotions. In this paper, we present a novel context-aware multi-instance learning (CMIL) algorithm for horror image recognition. The CMIL algorithm identifies horror images and picks out the regions that cause the sensation of horror in these horror images. It obtains contextual cues among adjacent regions in an image using a random walk on a contextual graph. Borrowing the strength of the fuzzy support vector machine (FSVM), we define a heuristic optimization procedure based on the FSVM to search for the optimal classifier for the CMIL. To improve the initialization of the CMIL, we propose a novel visual saliency model based on the tensor analysis. The average saliency value of each segmented region is set as its initial fuzzy membership in the CMIL. The advantage of the tensor-based visual saliency model is that it not only adaptively selects features, but also dynamically determines fusion weights for saliency value combination from different feature subspaces. The effectiveness of the proposed CMIL model is demonstrated by its use in horror image recognition on two large-scale image sets collected from the Internet.
Asunto(s)
Procesamiento de Imagen Asistido por Computador/métodos , Aprendizaje Automático , Reconocimiento de Normas Patrones Automatizadas/métodos , Miedo , Lógica Difusa , Humanos , Películas Cinematográficas , Máquina de Vectores de SoporteRESUMEN
OBJECTIVE: This study is designed to investigate the treatment approach and prognosis of pulmonary ground-glass-like shadow, especially solitary nodular ground-glass opacity (SNGGO). METHODS: Forty-nine cases of SNGGO that persisted after anti-inflammatory treatment in our hospital were retrospectively studied. These patients received thoracoscopic surgery due to indefinitive diagnosis and a tendency of canceration (some cases were followed up for 1-24 months before surgery). Intraoperative rapid frozen section was performed for pathological diagnosis, and surgery method was chosen according to pathological results and the health status of the patients. RESULTS: Forty-three cases showed malignancy, among which 36 cases received thoracoscopic total resection of the lung cancer and seven received simple wedge resection or pulmonary segment resection due to poor lung function; two cases were atypical adenomatous hyperplasia (AAH) and received wedge resection; and four cases were benign and received lesion resection only. Intraoperative frozen section results were in line with postoperative pathological analysis. No lymph node metastasis was detected in any malignant cases as indicated by lymph node dissection or sampling. All malignant cases were staged Ia by postoperative pathological analysis. Neither recurrence nor metastasis occurred during the 1-30 months' follow-up. CONCLUSIONS: SNGGO that persists after anti-inflammatory treatment tend to be adenocarcinoma, which can hardly be diagnosed in the early stage through non-invasive examination. If there's no contraindication for surgery, video-assisted thoracoscopy (VATS)-guided resection of the lesion plus intraoperative rapid frozen section should be performed to synchronize diagnosis and treatment, which could achieve satisfactory prognosis.