Pareto-Optimal Clustering with the Primal Deterministic Information Bottleneck.

At the heart of both lossy compression and clustering is a trade-off between the fidelity and size of the learned representation. Our goal is to map out and study the Pareto frontier that quantifies this trade-off. We focus on the optimization of the Deterministic Information Bottleneck (DIB) objective over the space of hard clusterings. To this end, we introduce the primal DIB problem, which we show results in a much richer frontier than its previously studied Lagrangian relaxation when optimized over discrete search spaces. We present an algorithm for mapping out the Pareto frontier of the primal DIB trade-off that is also applicable to other two-objective clustering problems. We study general properties of the Pareto frontier, and we give both analytic and numerical evidence for logarithmic sparsity of the frontier in general. We provide evidence that our algorithm has polynomial scaling despite the super-exponential search space, and additionally, we propose a modification to the algorithm that can be used where sampling noise is expected to be significant. Finally, we use our algorithm to map the DIB frontier of three different tasks: compressing the English alphabet, extracting informative color classes from natural images, and compressing a group theory-inspired dataset, revealing interesting features of frontier, and demonstrating how the structure of the frontier can be used for model selection with a focus on points previously hidden by the cloak of the convex hull.

Cigarette Consumption Patterns of Non-Smokers, Occasional Smokers, and Daily Smokers in Selected ASEAN Countries.

BACKGROUND: This study investigates the socio-demographic factors associated with smoking status in five Southeast Asian countries: Malaysia, Thailand, Indonesia, Vietnam, and Philippines. METHODS: This cross-sectional study utilizes data of adults ≥15 years who completed the Global Adult Tobacco Surveys. Ordered probit analysis is used to account for the smoking statuses of non-smokers, occasional smokers, and daily smokers. RESULTS: Malaysian and Vietnamese households with more family members face lower smoking likelihoods than otherwise. Urbanites in Philippines and rural residents in Thailand and Indonesia are more likely to smoke on occasional and daily basis than others. Males are consistently more likely to smoke occasionally or daily and less likely to be non-smokers than females across all countries. Younger middle-age (retiree) individuals aged 30-35 (≥60) years in Malaysia and Thailand exhibit higher (lower) likelihoods to smoke occasionally or daily than their younger cohorts aged 15-29 years. Individuals aged 30 years and above in Indonesia, Vietnam, and Philippines display higher daily smoking propensities than others. Higher education levels dampens smoking likelihoods and increases non-smoking propensities in all countries. Non-government or self-employed workers in all countries are more likely to smoke occasionally or daily than unemployed persons. Being married is associated with higher non-smoking likelihoods in Thailand although this association is not evident in Malaysia. CONCLUSION: These findings suggest that a portfolio of targeted interventions is necessary to meet the needs of specific subpopulations within the various countries.
.

Body weight and physical activity of adolescents in Malaysia.

BACKGROUND: This study examined the factors associated with body mass index (BMI) categories of Malaysian adolescents by physical activity (PA) status. METHODS: Data were obtained from the 2012 Malaysia Global School-based Student Health Survey. Generalized ordered logit regression analysis was conducted on 24 339 adolescents by PA status. RESULTS: Early- (ages 11-13) and middle-stage (ages 14-16) adolescents were associated with higher overweight and obesity risks than their older peers (ages 17-18). Male adolescents faced higher underweight and obesity likelihoods than females. Hunger due to food shortage at home was associated with higher likelihoods of underweight and normal weight BMI categories. Smokers were more likely to be underweight or normal weight than non-smokers. Segmented-sample analysis by PA status indicated that, while the direction of associations was parallel across PA status, the magnitudes of association between age, hunger and smoking status with BMI status were greater among active than inactive adolescents. CONCLUSIONS: Male adolescents faced a dual burden of underweight and obesity. Other sociodemographic and dietary-lifestyle factors were associated with adolescent BMI categories. Segmented-sample analysis by PA status uncovered varying associations between factors that would otherwise be masked in pooled sample analysis. Public health authorities should take these factors into consideration when deliberating programs to ensure healthy adolescent body weight.

Awareness of diabetes, hypertension, and hypercholesterolemia in Malaysia.

BACKGROUND: Policy interventions for cardiovascular diseases require individual awareness of ailments. Such awareness is also key to individuals making changes to their lifestyle and dietary habits. The present study investigated the association of sociodemographic, health, and lifestyle factors with the awareness and prevalence of three ailments: diabetes, hypertension, and hypercholesterolemia. METHODS: Data were obtained from the Malaysia Non-Communicable Disease Surveillance-1. Logistic regressions were estimated and odds ratios of exposure variables calculated. RESULTS: Diabetes awareness was associated with work hours, age, family history of illnesses, and ethnicity. Individuals with diminished hypertension awareness included those who were younger, without family history of illnesses, not obese, working more hours, and not adhering to a healthy diet. Low awareness of hypercholesterolemia was associated with younger age, lower education level, living in rural areas, female gender, no family history of illnesses, non-obesity, and minority ethnic background. Prevalence generally had the same pattern of association with the exposure variables. CONCLUSIONS: Various sociodemographic and health and lifestyle characteristics were associated with diabetes, hypertension, and hypercholesterolemia awareness in Malaysia, albeit with varying outcomes. Therefore, programs focusing on lifestyle improvements should be targeted at high-risk subgroups, such as individuals working longer hours and young adults, who are less likely to be aware of their health risk factors.

Consumption of fruits and vegetables in Malaysia: profiling the daily and nondaily consumers.

This study examines the sociodemographic factors associated with daily fruit and vegetable (FV) consumption in Malaysia. Based on a cross-sectional sample of 2447 individuals from the Malaysia Non-Communicable Disease Surveillance-1, a multivariate sample selection system is developed and estimated, to accommodate high frequencies of daily FV consumption and the days of servings among nondaily consumers. Results indicate that the authors' attempt to account for endogenous sample selectivity and cross-equation correlations is justified. There exist positive correlations between FV consumption likelihoods and longer work hours, higher levels of education, high income, female gender, nonsmoking status, and being from East Malaysia. Among nondaily consumers, those with longer work hours, singles, and people with diabetes are less inclined to eat fruits on more days. Overall, higher-educated, affluent people, nonsmokers, and East Malaysians display more days of FV consumption. Based on these outcomes, several policy implications are recommended vis-à-vis FV consumption patterns in Malaysia.

Obesity inequality in Malaysia: decomposing differences by gender and ethnicity using quantile regression.

OBJECTIVE: Obesity prevalence is unequally distributed across gender and ethnic group in Malaysia. In this paper, we examine the role of socioeconomic inequality in explaining these disparities. DESIGN: The body mass index (BMI) distributions of Malays and Chinese, the two largest ethnic groups in Malaysia, are estimated through the use of quantile regression. The differences in the BMI distributions are then decomposed into two parts: attributable to differences in socioeconomic endowments and attributable to differences in responses to endowments. RESULTS: For both males and females, the BMI distribution of Malays is shifted toward the right of the distribution of Chinese, i.e., Malays exhibit higher obesity rates. In the lower 75% of the distribution, differences in socioeconomic endowments explain none of this difference. At the 90th percentile, differences in socioeconomic endowments account for no more than 30% of the difference in BMI between ethnic groups. CONCLUSIONS: Our results demonstrate that the higher levels of income and education that accrue with economic development will likely not eliminate obesity inequality. This leads us to conclude that reduction of obesity inequality, as well the overall level of obesity, requires increased efforts to alter the lifestyle behaviors of Malaysians.

Who are eating and not eating fruits and vegetables in Malaysia?

OBJECTIVE: To investigate the roles of sociodemographic factors in fruit and vegetable (FV) consumption in Malaysia. METHODS: Data are obtained from the Malaysia Non-Communicable Disease Surveillance-1. Logistic regressions are conducted using a multiracial (Malay, Chinese, Indian and other ethnic groups) sample of 2,447 observations to examine the factors affecting individual decisions to consume FV on a daily basis. RESULTS: Based on the binary outcomes of whether individuals consumed FV daily, results indicate that work hours, education, age ethnicity, income, gender, smoking status, and location of residence are significantly correlated with daily fruit consumption. Daily vegetable consumption is significantly correlated with income, gender, health condition, and location of residence. CONCLUSIONS: Our results imply the need for programs to educate and motivate consumers to make healthier dietary choices. Interventions to increase FV consumption by changing behaviors should be considered, as should those that increase public awareness of the dietary benefits of FV. These intervention programs should be targeted at and tailored toward individuals who are less educated, younger, less affluent, males, smokers, and metropolitan dwellers.

Determinants of body weight status in Malaysia: an ethnic comparison.

OBJECTIVE: To investigate the roles of sociodemographic and health lifestyle factors in affecting body mass index (BMI) across ethnic groups in Malaysia. METHODS: Data are obtained from 2,436 observations from the Malaysia Non-Communicable Disease Surveillance-1. The multi-ethnic sample is segmented into Malay, Chinese, and Indian/other ethnicities. Ordered probit analysis is conducted and marginal effects of sociodemographic and health lifestyle variables on BMI calculated. RESULTS: Malays between 41 and 58 years are more likely to be overweight or obese than their 31-40 years counterparts, while the opposite is true among Chinese. Retirees of Chinese and Indian/other ethnicities are less likely to be obese and more likely to have normal BMI than those between 31 and 40 years. Primary educated Chinese are more likely to be overweight or obese, while tertiary-educated Malays are less likely to suffer from similar weight issues as compared to those with only junior high school education. Affluent Malays and Chinese are more likely to be overweight than their low-middle income cohorts. Family illness history is likely to cause overweightness or obesity, irrespective of ethnicity. Malay cigarette smokers have lower overweight and obesity probabilities than non-cigarette smokers. CONCLUSIONS: There exists a need for flexible policies to address cross-ethnic differences in the sociodemographic and health-lifestyle covariates of BMI.

Ethnic disparities in metabolic syndrome in malaysia: an analysis by risk factors.

BACKGROUND: This study investigates ethnic disparities in metabolic syndrome in Malaysia. METHODS: Data were obtained from the Malaysia Non-Communicable Disease Surveillance-1 (2005/2006). Logistic regressions of metabolic syndrome health risks on sociodemographic and health-lifestyle factors were conducted using a multiracial (Malay, Chinese, and Indian and other ethnic groups) sample of 2,366 individuals. RESULTS: Among both males and females, the prevalence of metabolic syndrome amongst Indians was larger compared to both Malays and Chinese because Indians are more likely to exhibit central obesity, elevated fasting blood glucose, and low high-density lipoprotein cholesterol. We also found that Indians tend to engage in less physical activity and consume fewer fruits and vegetables than Malays and Chinese. Although education and family history of chronic disease are associated with metabolic syndrome status, differences in socioeconomic attributes do not explain ethnic disparities in metabolic syndrome incidence. The difference in metabolic syndrome prevalence between Chinese and Malays was not statistically significant. Whereas both groups exhibited similar obesity rates, ethnic Chinese were less likely to suffer from high fasting blood glucose. CONCLUSIONS: Metabolic syndrome disproportionately affects Indians in Malaysia. Additionally, fasting blood glucose rates differ dramatically amongst ethnic groups. Attempts to decrease health disparities among ethnic groups in Malaysia will require greater attention to improving the metabolic health of Malays, especially Indians, by encouraging healthful lifestyle changes.

Utilization of breast cancer screening methods in a developing nation: results from a nationally representative sample of Malaysian households.

As is the case in many developing nations, previous studies of breast cancer screening behavior in Malaysia have used relatively small samples that are not nationally representative, thereby limiting the generalizability of results. Therefore, this study uses nationally representative data from the Malaysia Non-Communicable Disease Surveillance-1 to investigate the role of socio-economic status on breast cancer screening behavior in Malaysia, particularly differences in screening behaviour between ethnic groups. The decisions of 816 women above age 40 in Malaysia to screen for breast cancer using mammography, clinical breast exams (CBE), and breast self-exams (BSE) are modeled using logistic regression. Results indicate that after adjusting for differences in age, education, household income, marital status, and residential location, Malay women are less likely than Chinese and Indian women to utilize mammography, but more likely to perform BSE. Education level and urban residence are positively associated with utilization of each method, but these relationships vary across ethnicity. Higher education levels are strongly related to using each screening method among Chinese women, but have no statistically significant relationship to screening among Malays.

Sociodemographic and health-lifestyle determinants of obesity risks in Malaysia.

The purpose of this study was to examine the sociodemographic and health-lifestyle factors that affect the likelihood of obesity among Malaysians. Data were obtained from the Malaysian Non-Communicable Disease Surveillance-1. The cross-sectional population-based survey consisted of 2447 observations, with an obesity prevalence rate of 17.2%. Based on logit regression analysis, the results suggest that obesity risks in Malaysia are affected by gender, education level, family history, health conditions, smoking status, and ethnic backgrounds. Specifically, Malaysians more likely to be obese are females (5.3%), lower educated (0.9%), those with history of family illnesses (4.8%), and nonsmokers (6.4%). However, Chinese (9.3%) and other (5.5%) ethnic groups are less likely to be obese when compared with Malays. Based on these results, several policy implications are discussed vis-à-vis obesity risks in Malaysia.

Cervical cancer screening in Malaysia: Are targeted interventions necessary?

This study examines the determinants of Papanicolaou Smear Test (PST) screening for cervical cancer among women in Malaysia. Attention is focused on the reasons different population subgroups give for non-screening. We find that Indian women are the least likely to have had a PST and also the least likely to know the reasons why one is screened. Malay women are less likely than Chinese women to have received a PST and are more likely to report embarrassment as the reason for not being tested. Urban women are less likely than rural women to have been tested and more likely to state lack of time as the reason. These results suggest targeted interventions may be necessary to increase screening rates in Malaysia.