Investigation of Hyperparameter Setting of a Long Short-Term Memory Model Applied for Imputation of Missing Discharge Data of the Daihachiga River

Missing observational data pose an unavoidable problem in the hydrological field. Deep learning technology has recently been developing rapidly, and has started to be applied in the hydrological field. Being one of the network architectures used in deep learning, Long Short-Term Memory (LSTM) has been applied largely in related research, such as flood forecasting and discharge prediction, and the performance of an LSTM model has been compared with other deep learning models. Although the tuning of hyperparameters, which influences the performance of an LSTM model, is necessary, no sufficient knowledge has been obtained. In this study, we tuned the hyperparameters of an LSTM model to investigate the influence on the model performance, and tried to obtain a more suitable hyperparameter combination for the imputation of missing discharge data of the Daihachiga River. A traditional method, linear regression with an accuracy of 0.903 in Nash–Sutcliffe Efficiency (NSE), was chosen as the comparison target of the accuracy. The results of most of the trainings that used the discharge data of both neighboring and estimation points had better accuracy than the regression. Imputation of 7 days of the missing period had a minimum value of 0.904 in NSE, and 1 day of the missing period had a lower quartile of 0.922 in NSE. Dropout value indicated a negative correlation with the accuracy. Setting dropout as 0 had the best accuracy, 0.917 in the lower quartile of NSE. When the missing period was 1 day and the number of hidden layers were more than 100, all the compared results had an accuracy of 0.907–0.959 in NSE. Consequently, the case, which used discharge data with backtracked time considering the missing period of 1 day and 7 days and discharge data of adjacent points as input data, indicated better accuracy than other input data combinations. Moreover, the following information is obtained for this LSTM model: 100 hidden layers are better, and dropout and recurrent dropout levels equaling 0 are also better. The obtained optimal combination of hyperparameters exceeded the accuracy of the traditional method of regression analysis.

Effect of lunar cycle on rainfall

ABSTRACT. Rainfall data for a period of 50 years from 1931 onwards have been analysed for three west, coast stations in Kerala for the southwest monsoon period, The period is divided into two halves, the first half, i.e.June-July, providing comparatively more rainfall and the second half, i.e. August-September, providing comparatively lesser rainfall. Rainy days, having rain amounts>6.25 cm/day, have only been utilized for this study. The lunar cycle, which is having 29.53 days, is divided into ten phases, each phase constituting of around three days. To consider the effect of solar activity, the period is divided into active and quiet sun by considering those years with sunspot number greater than the upper quartile and those with sunspot number less than the lower quartile respectively. The data were analysed using x2 test. It describes the magnitude of the discrepancy between theory and observation. Analysis has shown that there is some statistical significance between heavy rainfall and lunar cycle. The effect is more significant in active sun period which shows the effect of solar activity also.

Acircularity and circularity indexes of the foveal avascular zone in high myopia

This study explored the association between foveal avascular zone (FAZ) parameters and high myopia using optical coherence tomography angiography. We divided 106 eyes of 106 patients into quartiles based on the axial length. The upper quartile was then defined as the high myopia group (n = 27), while the lower quartile was the non-high myopia group (n = 26). The areas and minor axis lengths of superficial and deep FAZ, the perimeters and major axis lengths of deep FAZ were significantly larger in eyes with high myopia than in eyes with non-high myopia (P < 0.05). Inversely, the subfoveal choroidal thickness was significantly thinner in eyes with high myopia than in those with non-high myopia. Linear regression analyses showed that no significant correlation was observed between FAZ areas and acircularity and circularity indexes of FAZ in non-high myopia group. Conversely, FAZ areas strongly correlated with acircularity and circularity indexes of FAZ in high myopia group. We found that an increase in the FAZ area in highly myopic eyes was accompanied by a significant variation in FAZ acircularity and circularity indexes. Further research should address whether these findings are associated with future disease development in highly myopic eyes.

Physical Activity, Fitness and Cognitive Performance of Estonian First-Grade Schoolchildren According Their MVPA Level in Kindergarten: A Longitudinal Study

Little is known about the longitudinal trajectories and associations regarding physical activity (PA), physical fitness (PF), and cognitive skills in childhood. Accelerometer-based PA, sedentary behavior (SB), PF, and cognitive skills were measured in Estonian children (n = 147) in kindergarten (6.6 years) and again at school (7.6 years). Children were subgrouped into lower and upper quartiles by their moderate-to-vigorous PA (MVPA) at 6.6 years. Children in the upper quartile had lower SB, higher PA, and greater muscular strength. Attending school, MVPA in the lower quartile improved. In both subgroups, most strength values and cognitive skills improved, while balance deteriorated in first grade. In the upper quartile, a greater MPA at 6.6 years predicted lower perceptual skills at 7.6 years. A greater SB at 6.6 years predicted higher verbal skills, light and moderate PA and MVPA, and lower verbal skills at 7.6 years after taking into account confounding factors such as the child’s sex, age, awake wear time (AWT), maternal education, and/or child’s sports participation. A vigorous PA at 6.6 years predicted perceptual (in upper quartile) or verbal (in lower quartile) skills at 7.6 years after controlling for similar confounders. No correlation for PF at 6.6 years and cognitive skills at 7.6 years existed; after adjusting for the above-mentioned confounders relative to upper/lower-limb strength, the 4 × 10 m shuttle run results predicted higher perceptual or verbal skills; static balance and cardiorespiratory fitness predicted lower verbal skills. Cardiorespiratory fitness predicted higher perceptual skills after controlling for sex, age, and AWT. Overall, PA and strength were constantly better and SB lower in the upper quartile, yet the lower quartile demonstrated improved MVPA in first grade, and both subgroups increased most components of their strength and cognitive skills in first grade. Higher levels of VPA at kindergarten predicted either better perceptual or verbal skills in first grade after controlling for confounders; the opposite associations were found for other PA levels and cognitive skills in the higher quartile. PF components at kindergarten predicted either superior or inferior cognitive skills in first grade after adjusting for confounders.

Dynamics of size classes of cities in Poland in 1950–2018

The study aims to present variability in the population of individual size classes of Polish cities in 1950-2018, its determinants and consequences for the formation of the settlement system. The application of quartiles as the basic statistical measure allowed identifying which parts of particular size classes of cities had the greatest impact (progressively or regressively) on changes in the settlement system. In this respect, strong dynamics of the upper quartile of Polish cities, lasting until the end of the 1970s, becomes apparent. In the 1980s, the dynamics of the upper quartile of large cities showed a shift during development waves in comparison to the culmination of these waves for medium-sized cities. However, since the end of the 1980s, there has been an increasing role of the lower quartile in determining the direction of changes in the size structure of cities, especially in medium and small cities.

The Predictive Role of QEEG in the Post-Operative Delirium and Prognosis of Cardiac Surgery Patients

Objective To determine whether qEEG has the capable of predicting the onset of post-operative delirium (POD) and prognosis of cardiac surgery patients. Methods We prospectively studied a cohort of cardiac surgery patients undergoing EEG for evaluation of altered mental status. Patients were assessed for delirium with the CAM-ICU or CAM. EEG were interpreted clinically by clinician, and reports were reviewed to identify features such as aEEG，alpha，beta，theta or delta relative band energy, alpha variability and spectral entropy. Generalized linear models were used to quantify associations among EEG findings, delirium, and clinical outcomes, including length of stay and mortality. Results 60 patients were evaluated and 29 (48.33%) met delirium criteria. The EEG finding most strongly associated with delirium presence was aEEG, whether the peak value or valley value less than lower quartile or greater than upper quartile. Conclusions The peak or valley value of aEEG in F3-P3/F4-P4 derivation is a good predictor of post-operative delirium in cardiac surgery patients.Trial registration: Clinical Trials.gov Identifier, NCT03351985. Registered 1 December 201

Humoral Response to the Pfizer BNT162b2 Vaccine in Patients Undergoing Maintenance Hemodialysis

Background and objectivesCoronavirus disease 2019 (COVID-19) is associated with higher morbidity and mortality in patients on maintenance hemodialysis. Patients on dialysis tend to have a reduced immune response to infection or vaccination. We aimed to assess, for the first time to the best of our knowledge, the humoral response following vaccination with the BNT162b2 vaccine in patients on maintenance hemodialysis and the factors associated with it.Design, setting, participants, & measurementsThe study included 56 patients on maintenance hemodialysis (dialysis group) and a control group composed of 95 health care workers. All participants had received two doses of the BNT162b2 (Pfizer-BioNTech) vaccine. The serology testing was done using Quant II IgG anti-Spike severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) assay by Abbott a median of 30 days after receipt of the second dose of the vaccine.ResultsAll subjects in the control group developed an antibody response compared with 96% (54 of 56) positive responders in the dialysis group. The IgG levels in the dialysis group (median, 2900; interquartile range, 1128–5651) were significantly lower than in the control group (median, 7401; interquartile range, 3687–15,471). A Mann–Whitney U test indicated that this difference was statistically significant (U=1238; P<0.001). There was a significant inverse correlation of age and IgG levels in both groups. The odds of being in the lower quartile were significantly higher for older individuals (odds ratio, 1.11 per year of age; 95% confidence interval, 1.08 to 1.20; P=0.004) and for the dialysis group compared with the control group (odds ratio, 2.7; 95% confidence interval, 1.13 to 7.51; P=0.05). Within the dialysis group, older age and lower lymphocyte count were associated with antibody response in the lower quartile (odds ratio, 1.22 per 1-year older; 95% confidence interval, 1.13 to 1.68; P=0.03 and odds ratio, 0.83 per 10-e3/µl-higher lymphocyte count; 95% confidence interval, 0.58 to 0.97; P=0.05).ConclusionsAlthough most patients on maintenance hemodialysis developed a substantial humoral response following the BNT162b2 vaccine, it was significantly lower than controls. Age was an important factor in the humoral response, regardless of chronic medical conditions.

Effects of the COVID-19 Lockdown on Portuguese Children’s Motor Competence

During long periods without school, children are more susceptible to unhealthy behaviors, such as an increase in sedentary behaviors, which has a negative impact on children’s motor competence (MC). The COVID-19 lockdown offered us a unique opportunity to test, in a quasi-experimental setting, the impact of lockdown movement restrictions on children’s MC. We assessed the motor competence of 114 children aged 6–9 years using the motor competence assessment. All children were tested before and after the COVID-19 lockdown. Chi-square and 2 × 2 ANOVA (sex by moment) were used to further analyze the data. Regardless of sex, motor performances in all tests (except for jumping sideways in boys) were lower when compared with performances before lockdown. There was a marked decreasing trend in children’s levels of MC, shifting from an upper to a lower quartile in different tests. The results after the lockdown were always significantly inferior to the results before lockdown in all motor tests (except jumping sideways), in the three components of MC, and in global MC. Children’s global MC score decreased by an average of 13 points in boys and 16 points in girls. The imposed movement restrictions had a negative effect on children’s motor competence development.

UK Internet antenatal dietary advice: a content accuracy and readability analysis

The aim of the study was to assess the accuracy and readability of Internet prenatal nutrition advice. Between August and December 2018, 130 Internet pages returned from Google searches on foods to avoid, foods to eat and supplements use were compared with UK government advice for pregnant women. Readability was assessed using the Flesch Readability Ease (FRE) tool. Descriptive and non-parametric tests were used. Spearman’s correlation explored associations between accuracy and readability. Kruskal–Wallis tests with Bonferroni correction were used for multiple pairwise tests and Mann–Whitney U tests for two-sample differences in medians. A total of 130 Internet pages were examined: 48 % from publishers, 27 % from other commercial organisations, 22 % from charities and 3 % from governments. Eighty-three (64 %) pages contained inaccurate and accurate advice, twenty-three (18 %) were accurate and complete, twenty-one (16 %) were inaccurate, and three (2 %) lacked any relevant advice. The median percentage accuracy of all advice was 83 (lower quartile, upper quartile: 48, 100). Median FRE was 55 (46, 61) ‘fairly difficult’. Eighty-seven pages (67 %) scored below the recommended FRE for public Internet pages. There was a weak positive correlation between accuracy and readability of Internet pages (rho = 0·241, P = 0·006). Accuracy of Internet pages did not differ by dietary theme. Pages on supplements were the most difficult to read. Internet pages from publishers and other commercial organisations were significantly less accurate than those from not-for-profit organisations (median percentage difference –8 (–29, 0·00), P = 0·019). Much pregnancy-related dietary advice online is inaccurate and difficult to read. Advice should be developed in consultation with qualified nutritionists and dietitians.

P0762THE SERUM URIC ACID HAS A INDEPENDENT ASSOCIATION WITH 1-YEAR RENAL OUTCOMES IN CHRONIC KIDNEY DISEASE PATIENT WITH HYPERTENSION

Background and Aims Serum uric acid (UA) is associated with renal disease. Hyperuricemia can be a risk of hypertension, intrarenal vascular disease and renal injury. We investigate the serum UA has an association with renal disease progression in patients with chronic kidney disease (CKD) with hypertension. Method We recruited 270 CKD patients with hypertension from 4 centers in Korea through the APrODiTe study and followed for 1 year. Serum UA was evaluated as a continuous value and groups divided by quartiles. The renal outcomes were an increase in random urine protein/creatinine ratio (PCR) than baseline value or estimated glomerular filtration rate (eGFR) deterioration which means a decrease in eGFR ≥ 5 (ml/min/1.73m2). Results Baseline serum UA was 6.58 ± 1.73 mg/dl and 6.52 ± 3.59 mg/dl after 1 year. For proteinuria progression, a 1 mg/dl higher serum UA has independent correlation in multivariate regression (odds ratio (OR): 1.272; 95% confidence interval (CI): 1.031-1.568; P = 0.024). The higher quartile of serum UA showed a correlation with the odd ratio than lower quartile (OR: 2.243; 95% CI: 0.862-5.837; P = 0.098, OR:3.417; 95% CI: 1.275-9.152; P= 0.015, OR: 2.754; 95% CI: 1.013-7.488; P &lt; 0.047). In subgroup analysis, the patients with late CKD stage (3-5) showed serum UA has a positive correlation with proteinuria progression (OR: 1.311; 95% CI: 1.022-1.682; P= 0.033) and the top quartile group was correlated with the increased odds ratio compared to lower quartile (OR: 3.811; 95% CI: 1.153-12.59; P = 0.028). For eGFR deterioration, the higher quartile of UA was positively correlated with the odd ratio in only univariate analysis. Conclusion Serum UA level has an independent correlation with proteinuria progression in especially late CKD patient with hypertension. Whereas for eGFR deterioration, serum UA did not show a significant correlation.