The MIDAS Touch: Accurate and Scalable Missing-Data Imputation with Deep Learning

Principled methods for analyzing missing values, based chiefly on multiple imputation, have become increasingly popular yet can struggle to handle the kinds of large and complex data that are also becoming common. We propose an accurate, fast, and scalable approach to multiple imputation, which we call MIDAS (Multiple Imputation with Denoising Autoencoders). MIDAS employs a class of unsupervised neural networks known as denoising autoencoders, which are designed to reduce dimensionality by corrupting and attempting to reconstruct a subset of data. We repurpose denoising autoencoders for multiple imputation by treating missing values as an additional portion of corrupted data and drawing imputations from a model trained to minimize the reconstruction error on the originally observed portion. Systematic tests on simulated as well as real social science data, together with an applied example involving a large-scale electoral survey, illustrate MIDAS’s accuracy and efficiency across a range of settings. We provide open-source software for implementing MIDAS.

Consumption of Red Meat Is Negatively Associated with Cognitive Function: A Cross-Sectional Analysis of UK Biobank

Objectives In the largest study of its type, we tested for associations between red meat consumption and cognitive function using data from half a million participants enrolled into the UK Biobank cohort study. Methods Baseline data was obtained from the UK Biobank cohort, comprising half a million participants aged 37–73 years recruited between 2006 and 2010. The intake of red meat (frequency per week) was obtained using a self-reported food frequency questionnaire. Cognitive tests included the reaction-time (RT) test (reaction ability, N = 496, 695), fluid intelligence (FI) (reasoning ability, N = 165,467), the numeric memory test (short-term memory, N = 50,364), the pairs-matching (PM) test (visual-spatial memory, N = 482,650) and the prospective memory test (N = 171,509). Logistic and linear regression modelling was conducted with adjustment for potential confounders including age at recruitment, sex, ethnicity, Townsend deprivation index, smoking, alcohol, education, body mass index, physical activity level, sleeping hours, stroke history, and family history of dementia. Results Each additional portion per week of red-meat intake was associated with slower reaction time by 0.26 milliseconds (95% CI: 0.02, 0.50), lower FI score by 0.01 points (−0.02, −0.00), reduced numeric memory by 0.02 digits (−0.03, −0.01), and increased odds of incorrect prospective memory by 1% (0%, 2%). In men these associations were larger regarding the RT test (β = 0.54, [0.21, 0.87]), FI score (β = −0.02, [−0.03, −0.01]), and prospective memory (OR = 1.03, [1.01, 1.04]), while in women these were not significant. In terms of the PM test, a single additional portion of red-meat intake was associated with reduced incorrect matches by 0.004 pairs (−0.003, −0.006), both in men (β = −0.003, [−0.001, −0.005]) and women (β = −0.006, [−0.004, −0.008]). Conclusions In this cross-sectional analysis of the adult UK population, higher intake of red meat was associated with poorer cognitive function including reaction and reasoning ability, short-term and prospective memory especially among men; but not visual-spatial memory which showed a weak protective effect of red meat. Funding Sources The joint scholarship of University of Leeds and China Scholarship Council.

Do Older Adults with Multimorbidity Meet the Recommended Levels of Physical Activity? An Analysis of Scottish Health Survey

There is a positive association between physical activity (PA) and improved health in older adults. The objective of this study was to assess the prevalence and determinants of meeting recommended levels of PA among older adults with multimorbidity. Data has been derived from the nationally representative Scottish Health Surveys (2014–2017). A sub-sample of 2230 older adults (aged 65+) with multimorbidity were the study participants. Physical activity was evaluated using current recommended guidelines. Overall, 32.3% of the participants met the recommended levels of PA. Independent predictors of meeting the recommended levels of PA include male gender [odds ratio (OR) 2.00 (95% confidence interval (CI) 1.58–2.54)], living in the least deprived areas [OR 1.79 (95% CI 1.20–2.69)]; being a non-smoker [OR 2.22 (95% CI 1.48–3. 34)]. Also, meeting recommended PA decreased with age [OR 0.92 (95% CI 0.90–0.94)] and body mass index [OR 0.93 (95% CI 0.91–0.95]; but increased per additional portion of fruit and vegetables taken [OR 1.19 (95% CI 1.12–1.25)] and with increase in well-being scale score [OR 1.05 (95% CI 1.03 to 1.06)]. Adherence to PA guidelines seems to be more related to age, BMI, gender (i.e. higher PA adherence in men vs. women), social support (i.e. social deprivation), dietary habits (i.e. fruit and vegetable intake) and social isolation among the elderly. In the one-third of older population, adherence to PA was associated to better mental health. Therefore, adaptation of PA guideline to suit theses determinants factors would reduce the gap difference among older adults with multimorbidity and enhance their mental well-being.

Fruit and Vegetable Consumption and Their Polyphenol Content Are Inversely Associated with Sleep Duration: Prospective Associations from the UK Women’s Cohort Study

This study aims to investigate the prospective associations between fruit and vegetable (FV) intakes and their polyphenol content with subsequent sleep duration in UK women. In this study, 13,958 women with ~4 years of follow-up in the UK Women’s Cohort Study were included in the analyses. FV intakes were assessed at baseline using a food frequency questionnaire (FFQ), and average hours of sleep per day were self-reported in follow-up. Polyphenol intake was calculated by matching FV items from the FFQ with the Phenol-Explorer database. Linear regression models, adjusting for confounders, were used for the analyses. Consuming an additional portion of apples, kiwi, oranges, pineapple, and 100% pure juice were associated with shorter sleep. Similarly, an additional portion of cabbage, celery, aubergine, olives, and peppers were inversely associated with sleep duration. An additional gram of total polyphenols was associated with shorter sleep by 18 min (99% CI −31 to −4, p < 0.001). FV consumption and total polyphenol content were inversely associated with sleep duration; however, effect sizes were small, and polyphenol classes from FV intakes were not associated with sleep duration. Future intervention studies considering the time of FV consumption in relation to sleep are needed to clarify the underlying mechanisms.

Concentration of Credit Exposure as a Significant Source of Risk in Banking Activities: The Idea and Methods of Estimation

The simultaneous activation of many sources of risk can slow bank operations and even lead to bankruptcy. Credit risk is the greatest threat to the orderly functioning of a bank. To protect against its materialization banks spend nearly 90% of their total capital requirement. Concentration of credit exposure to single entities, as well as to single economic sectors, can be a source of additional risks. Estimation of the additional portion of the capital requirement in selected banks in Poland in 2008-2013 indicates that banks should assign additional 4% and 2% of the capital requirement to cover the risk of exposure concentrations in: respectively, individual entities and individual economic sectors. For banks with a retail profile more important was the risk of large exposures in individual economic sectors, and for banks with a corporate profile in individual entities. Estimates were carried out according to the procedure used by the Bank of Spain and the Bank of Slovenia, and the data derived from the annual financial reports of selected banks listed on the WSE.

Aquaporin 2: Identification and analysis of expression in calves’ urine during their first month of life

Michałek, K., Dratwa-Chałupnik, A., Ciechanowicz, A. K. and Malinowski, E. 2014. Aquaporin 2: Identification and analysis of expression in calves’ urine during their first month of life. Can. J. Anim. Sci. 94: 653–659. Aquaporin 2 (AQP2) is a vasopressin-regulated water channel of the renal collecting duct and is excreted into urine. Renal excretion of AQP2 has been studied in humans, dogs, rats and mice. The aim of this study was to identify and analyze expression of AQP2 in calves’ urine. Short and controlled diarrhea was induced in the calves in order to determine any change in renal excretion of AQP2 and to evaluate a possible role of this channel in water balance during the neonatal period. For this purpose, an additional portion of lactose was administered to the calves with a milk replacer. The present work shows that after transient diarrhea associated with excessive supply of lactose in the diet, a significant increase in renal AQP2 excretion was reported in the first 3 wk of the calves’ lives. Renal excretion of this protein in the urine increased significantly despite the absence of changes in the concentration of vasopressin or the osmotic pressure of the blood plasma. The increase observed in the expression of the protein might be associated with the reduction of the total Ca2+ concentration in the blood plasma of calves. The results clearly indicate a different mechanism regulating urinary excretion of AQP2 in newborn calves, hitherto unknown. Although elevated renal excretion of AQP2 was observed, there was no increase in the osmolality of the urine in the calves tested.

Effect of division of initial dose on combustion parameters of a dual fuel Cl engine

Worsening of economic and environmental parameters of a dual fuel engine at partial engine loads results from protracted combustion of leaned gas air mixture. This leads to reduction of overall efficiency and growth of CO and THC emissions. Such problems are difficult to solve in traction engines due to theirfrequent operation at partial loads and rapid changes of engine load and rotational speed. Negative effects of leaning of gaseous mixture can be partially diminished through division of initial dose and activation of combustion of gaseous mixture through delayed injection of additional portion of liquid fuel. In the paper are presented test results of experimental SB3.1 engine run on CNG with divided dose of the Diesel oil. There were analyzed overall efficiency and selected parameters of combustion process. Division of the dose influenced advantageously on parameters of the engine. The overall efficiency increased, while maximal pressure pmx, heat release rate (dQ/da)muxe and pressure growth rate (dp/da)max decreased. The engine with divided dose operated more smoothly with less noise.

The PVT gene frequently amplifies with MYC in tumor cells

The line of human colon carcinoma cells known as COLO320-DM contains an amplified and abnormal allele of the proto-oncogene MYC (DMMYC). Exon 1 and most of intron 1 of MYC have been displaced from DMMYC by a rearrangement of DNA. The RNA transcribed from DMMYC is a chimera that begins with an ectopic sequence of 176 nucleotides and then continues with exons 2 and 3 of MYC. The template for the ectopic sequence represents exon 1 of a gene known as PVT, which lies 50 kilobase pairs downstream of MYC. We encountered three abnormal configurations of MYC and PVT in the cell lines analyzed here: (i) amplification of the genes, accompanied by insertion of exon 1 and an undetermined additional portion of PVT within intron 1 of MYC to create DMMYC; (ii) selective deletion of exon 1 of PVT from amplified DNA that contains downstream portions of PVT and an intact allele of MYC; and (iii) coamplification of MYC and exon 1 of PVT, but not of downstream portions of PVT. We conclude that part or all of PVT is frequently amplified with MYC and that intron 1 of PVT represents a preferred boundary for amplification affecting MYC.

The PVT gene frequently amplifies with MYC in tumor cells.

The line of human colon carcinoma cells known as COLO320-DM contains an amplified and abnormal allele of the proto-oncogene MYC (DMMYC). Exon 1 and most of intron 1 of MYC have been displaced from DMMYC by a rearrangement of DNA. The RNA transcribed from DMMYC is a chimera that begins with an ectopic sequence of 176 nucleotides and then continues with exons 2 and 3 of MYC. The template for the ectopic sequence represents exon 1 of a gene known as PVT, which lies 50 kilobase pairs downstream of MYC. We encountered three abnormal configurations of MYC and PVT in the cell lines analyzed here: (i) amplification of the genes, accompanied by insertion of exon 1 and an undetermined additional portion of PVT within intron 1 of MYC to create DMMYC; (ii) selective deletion of exon 1 of PVT from amplified DNA that contains downstream portions of PVT and an intact allele of MYC; and (iii) coamplification of MYC and exon 1 of PVT, but not of downstream portions of PVT. We conclude that part or all of PVT is frequently amplified with MYC and that intron 1 of PVT represents a preferred boundary for amplification affecting MYC.