Forward Selection Procedure Research Articles

Fundamental factors and the level of underpricing

Underpricing has been the subject of substantial research at a global scale and has been thoroughly described in the literature, but it is still unclear what factors influence its level in the Polish capital market, particularly in the context of fundamental factors. Since investors have very limited information of the companies going public, the companies must convey or signal information that can be used for valuing their shares. The aim of the article is to identify the impact of a company’s fundamental factors on the level of underpricing of an initial public offering. The research is based on the sample of 227 companies debuting in the regulated market of the Warsaw Stock Exchange between 2005 and 2022. The empirical analysis and regression results estimation were based on the data coming from prospectuses of the companies going public. To prepare the econometric linear regression model, the stepwise forward variable selection procedure was carried out. The evidence indicates that underpricing is significantly affected by a few financial variables, such as assets, ROA, BVS, and EPS.

Integrating very high resolution environmental proxies in genotype-environment association studies.

Landscape genomic analyses associating genetic variation with environmental variables are powerful tools for studying molecular signatures of species' local adaptation and for detecting candidate genes under selection. The development of landscape genomics over the past decade has been spurred by improvements in resolutions of genomic and environmental datasets, allegedly increasing the power to identify putative genes underlying local adaptation in non-model organisms. Although these associations have been successfully applied to numerous species across a diverse array of taxa, the spatial scale of environmental predictor variables has been largely overlooked, potentially limiting conclusions to be reached with these methods. To address this knowledge gap, we systematically evaluated performances of genotype-environment association (GEA) models using predictor variables at multiple spatial resolutions. Specifically, we used multivariate redundancy analyses to associate whole-genome sequence data from the plant Arabis alpina L. collected across four neighboring valleys in the western Swiss Alps, with very high-resolution topographic variables derived from digital elevation models of grain sizes between 0.5 m and 16 m. These comparisons highlight the sensitivity of landscape genomic models to spatial resolution, where the optimal grain sizes were specific to variable type, terrain characteristics, and study extent. To assist in selecting variables at appropriate spatial resolutions, we demonstrate a practical approach to produce, select, and integrate multiscale variables into GEA models. After generalizing fine-grained variables to multiple spatial resolutions, a forward selection procedure is applied to retain only the most relevant variables for a particular context. Depending on the spatial resolution, the relevance for topographic variables in GEA studies calls for integrating multiple spatial scales into landscape genomic models. By carefully considering spatial resolutions, candidate genes under selection by a more realistic range of pressures can be detected for downstream analyses, with important applied implications for experimental research and conservation management of natural populations.

Analyzing market basket data through sparse multivariate logit models

AbstractUsing multivariate logit models, we analyze purchases of product categories made by individual households. We introduce a sparse multivariate logit model that considers only a subset of all two-way interactions. A combined forward and backward selection procedure based on a cross-validated performance measure excludes about 74 % of the possible two-way interactions. We also specify random coefficient versions of both the non-sparse and the sparse model. The fact that the random coefficient models lead to better values of the Bayesian information criterion demonstrates the importance of latent heterogeneity. The random coefficients sparse model attains the best statistical performance if we consider model complexity and offers a better interpretability. We investigate the cross-purchase effects of household segments derived from this random coefficient model. As additional interpretation aid we cluster categories and category pairs by integer programming. We demonstrate what the best performing sparse model implies for cross-selling by product recommendations and store layout. The sparse model leads to managerial implications with respect to the effects of advertising in local newspapers and flyers that are as a rule close to those implied by its non-sparse counterpart.

Forward Selection of Relevant Factors by Means of MDR-EFE Method

The suboptimal procedure under consideration, based on the MDR-EFE algorithm, provides sequential selection of relevant (in a sense) factors affecting the studied, in general, non-binary random response. The model is not assumed linear, the joint distribution of the factors vector and response is unknown. A set of relevant factors has specified cardinality. It is proved that under certain conditions the mentioned forward selection procedure gives a random set of factors that asymptotically (with probability tending to one as the number of observations grows to infinity) coincides with the “oracle” one. The latter means that the random set, obtained with this algorithm, approximates the features collection that would be identified, if the joint distribution of the features vector and response were known. For this purpose the statistical estimators of the prediction error functional of the studied response are proposed. They involve a new version of regularization. This permits to guarantee not only the central limit theorem for normalized estimators, but also to find the convergence rate of their first two moments to the corresponding moments of the limiting Gaussian variable.

The causality measure of partial mutual information from mixed embedding (PMIME) revisited.

The measure of partial mutual information from mixed embedding (PMIME) is an information theory-based measure to accurately identify the direct and directional coupling, termed Granger causality or simply causality, between the observed variables or subsystems of a high-dimensional dynamical and complex system, without any a priori assumptions about the nature of the coupling relationship. In its core, it is a forward selection procedure that aims to iteratively identify the lag-dependence structure of a given observed variable (response) to all the other observed variables (candidate drivers). This model-free approach is capable of detecting nonlinear interactions, abundantly present in real-world complex systems, and it was shown to perform well on multivariate time series of moderately high dimension. However, the PMIME presents some inefficiencies in its performance mainly when applied on strongly stochastic (linear or nonlinear) systems as it may falsely detect non-existent relationships. Moreover, and by construction, the measure cannot extract purely synergetic relationships present in a system. In the current work, the issue of false detections is addressed by introducing an improved resampling significance test and a procedure of rechecking the identified drivers (backward revision). Regarding the inability to detect synergetic relationships, the PMIME is further enhanced by checking pairs as candidate drivers for the response variable after having considered all drivers individually. The effects of these modifications are investigated in a systematic simulation study on properly designed systems involving strong stochasticity, regressor terms with synergetic effects, and a system dimension ranging from 3 to 30. The overall results of the simulations indicate that these modifications indeed improve the performance of PMIME and alleviate to a significant degree the issues of the original algorithm. Guidelines for balancing between accuracy and computational efficiency are also given, particularly relevant for real-world applications. Finally, the measure performance is investigated in the study of futures of various government bonds and stock market indices in the period around COVID-19 pandemic.

A meta-regression analysis to evaluate the effects of narasin on grow-finish pig performance.

Ionophores are feed additives that decrease gram-positive microbial populations by disrupting the ion transfer across cell membranes resulting in improved growth performance. Narasin (Skycis; Elanco Animal Health, Greenfield, IN) is an FDA-approved ionophore utilized for increased rate of weight gain and improved feed efficiency in growing-finishing pigs. A meta-regression analysis was conducted to evaluate the effects of added narasin in growing-finishing pig diets to predict its influence on average daily gain (ADG), feed efficiency (G:F), and carcass yield. A database was developed containing 21 technical reports, abstracts, and refereed papers from 2012 to 2021 representing 35 observations for growth performance data in studies ranging from 35 to 116 d in length (overall data). In addition, within these 35 observations, individual period data were evaluated (143 observations) using weekly, bi-weekly, or monthly performance intervals (period data). Regression model equations were developed, and predictor variables were assessed with a stepwise manual forward selection procedure. The ADG model using the overall data included ADG, ADFI, and G:F of the control group, added narasin dose, and narasin feeding duration categorized as longer or shorter than 65 d. Predictor variables included in the G:F model using overall data were ADG, ADFI, and G:F of the control group and added narasin dose. For carcass yield, the final model included ADFI and G:F of the control group, added narasin dose, and narasin feeding duration of longer than 65 d. In the period model for ADG, the predictors included ADG, ADFI, and G:F of the control group, added narasin dose, and average BW of the control group categorized into greater than or less than 105kg. For period data for G:F, the model selected ADG, ADFI, and G:F of the control group and added narasin dose. Based on the results, the overall response to added narasin for ADG and G:F was quadratic and tended to decrease as ADG and G:F increased. A similar quadratic response was observed for the individual period data. In summary, using median values from the database for predictor variables, this meta-analysis demonstrated narasin would be expected to improve ADG between 1.06% and 1.65%, G:F between 0.71% and 1.71%, and carcass yield by 0.31% when fed continuously for longer than 65 d.

The Association of Physiological and Pharmacological Anesthetic Parameters With Motor-Evoked Potentials: A Multivariable Longitudinal Mixed Model Analysis.

During spinal surgery, the motor tracts can be monitored using muscle-recorded transcranial electrical stimulation motor-evoked potentials (mTc-MEPs). We aimed to investigate the association of anesthetic and physiological parameters with mTc-MEPs. Intraoperative mTc-MEP amplitudes, mTc-MEP area under the curves (AUC), and anesthetic and physiological measurements were collected retrospectively from the records of 108 consecutive patients undergoing elective spinal surgery. Pharmacological parameters of interest included propofol and opioid concentration, ketamine and noradrenaline infusion rates. Physiological parameters recorded included mean arterial pressure (MAP), bispectral index (BIS), heart rate, hemoglobin O 2 saturation, temperature, and Et co2 . A forward selection procedure was performed using multivariable mixed model analysis. Data from 75 (69.4%) patients were included. MAP and BIS were significantly associated with mTc-MEP amplitude ( P < .001). mTc-MEP amplitudes increased by 6.6% (95% confidence interval [CI], 2.7%-10.4%) per 10 mm Hg increase in MAP and by 2.79% (CI, 2.26%-3.32%) for every unit increase in BIS. MAP ( P < .001), BIS ( P < .001), heart rate ( P = .01), and temperature ( P = .02) were significantly associated with mTc-MEP AUC. The AUC increased by 7.5% (CI, 3.3%-11.7%) per 10 mm Hg increase of MAP, by 2.98% (CI, 2.41%-3.54%) per unit increase in BIS, and by 0.68% (CI, 0.13%-1.23%) per beat per minute increase in heart rate. mTc-MEP AUC decreased by 21.4% (CI, -38.11% to -3.98%) per degree increase in temperature. MAP, BIS, heart rate, and temperature were significantly associated with mTc-MEP amplitude and/or AUC. Maintenance of BIS and MAP at the high normal values may attenuate anesthetic effects on mTc-MEPs.

Thermal modeling for control applications of 60,000 homes in North America using smart thermostat data

As smart thermostats become increasingly available in residential buildings, there is an opportunity to use measured building data to calibrate models for community and district applications, instead of relying on high-fidelity simulations. This study used smart thermostat data from 60,000 houses in North America to create single-zone models. The model structure was defined with an automated forward selection procedure. 61% of the final models were classified as good fits and the structure of 80% of them was of 5th-order (5 thermal capacitances). An investigation of the 24-hour prediction error of the models showed that the ones classified as good fits are accurate enough for day-ahead predictions and Model Predictive Control (MPC) applications. An analysis of the model parameters suggested no strong correlation between them and the available metadata. The time constants of the houses were estimated, providing valuable information about the houses thermal inertia. Building models that can accurately capture and leverage building thermal inertia are ideal candidates for MPC and energy flexibility applications.

Respiratory Parameters as Predictors of Balance and Gait Ability in Patients with Stroke at Discharge

Pulmonary complications are frequent in stroke, contributing to both mortality and morbidity rates. Respiratory parameters in such patients encompass both pulmonary function and respiratory muscle strength. Identifying respiratory function variables that influence the balance and gait ability of patients with stroke is crucial for enhancing their recovery in these aspects. However, no study has assessed predictions for a comprehensive array of balance and gait abilities in such patients. We aimed to examine whether initial respiratory muscle strength and pulmonary function can predict balance and gait ability at discharge from a rehabilitation program. Thirty-one patients with stroke were included in this prospective observational study. Multiple regression models with a forward selection procedure were employed to identify respiratory parameters (including peak expiratory flow and maximal expiratory pressure) that contributed to the results of balance assessments and gait evaluations at the time of discharge. The peak expiratory flow (PEF) served as a predictor explaining 42.0% of the variance. Similarly, the maximal expiratory pressure (MEP) was a predictor variable explaining 32.0% of the variance. PEF and MEP assessments at the initial stage as predictive factors for both balance and gait ability are important in stroke management.

Robust function-on-function interaction regression

A function-on-function regression model with quadratic and interaction effects of the covariates provides a more flexible model. Despite several attempts to estimate the model’s parameters, almost all existing estimation strategies are non-robust against outliers. Outliers in the quadratic and interaction effects may deteriorate the model structure more severely than their effects in the main effect. We propose a robust estimation strategy based on the robust functional principal component decomposition of the function-valued variables and [Formula: see text]-estimator. The performance of the proposed method relies on the truncation parameters in the robust functional principal component decomposition of the function-valued variables. A robust Bayesian information criterion is used to determine the optimum truncation constants. A forward stepwise variable selection procedure is employed to determine relevant main, quadratic, and interaction effects to address a possible model misspecification. The finite-sample performance of the proposed method is investigated via a series of Monte-Carlo experiments. The proposed method’s asymptotic consistency and influence function are also studied in the supplement, and its empirical performance is further investigated using a U.S. COVID-19 dataset.

Modeling climate change impacts on vector-borne disease using machine learning models: Case study of Visceral leishmaniasis (Kala-azar) from Indian state of Bihar

Visceral leishmaniasis or Kala-azar (KA) is a Vector-Borne Disease (VBD) that remains the second-largest parasitic killer across the globe (mortality rate: 75–95%). More than 60% of KA cases originate in South Asia, wherein India accounts for 2/3rd of the cases, and Bihar, a state in India, alone accounts for more than 50% of the Indian cases. Past studies suspected climate change vulnerabilities as a driving cause of KA outbreaks. The VBDs-based epidemic prediction systems have been developed to mitigate recurrent outbreaks; however, Machine Learning (ML) based approaches still need to be explored for modeling changing climate impacts on KA cases. This study, for the first time, develops a Radial Basis Function (RBF) kernel-based Support Vector Regression (SVR), hereinafter RBF-kernel-based-SVR model for the most-affected endemic districts of Bihar (northern-India), using the data from 2016 and 2021. Forward selection, backward elimination, and stepwise regression procedures were adopted while selecting influential climatic variables, followed by the k-fold cross-validation technique and, then, the RBF-kernel-based-SVR algorithm for classification. Results suggested that temperature, wind speed, rainfall, and population density significantly contributed to the KA outbreaks. This study also developed Multiple Linear Regression (MLR) and Multilayer Perceptron (MLP) models to compare SVR with other classification models. Findings indicated that the proposed RBF-kernel-based-SVR model [Correlation Coefficient (CC) = 0.82, Root-Mean-Square Error (RMSE) = 12.20, and Nash–Sutcliffe Efficiency (NSE) = 0.66] outperformed MLR (0.81, 14.20, 0.48) and MLP (0.81, 12.95, 0.61). Study recommends using the RBF-kernel-based-SVR model as a quick and efficient model capable of detecting KA cases with high predictability even under limited data availability. Such models can assist public health authorities, given monitoring KA spread, learning the climate impacts of outbreaks, and ensuring timelier health services.

Assessment of positive selection across SARS-CoV-2 variants via maximum likelihood.

Study of the genome of the SARS-CoV-2 virus, particularly with regard to understanding evolution of the virus, is crucial for managing the COVID-19 pandemic. To this end, we sample viral genomes from the GISAID repository and use several of the maximum likelihood approaches implemented in PAML, a collection of open source programs for phylogenetic analyses of DNA and protein sequences, to assess evidence for positive selection in the protein-coding regions of the SARS-CoV-2 genome. Across all major variants identified by June 2021, we find limited evidence for positive selection. In particular, we identify positive selection in a small proportion of sites (5-15%) in the protein-coding region of the spike protein across variants. Most other variants did not show a strong signal for positive selection overall, though there were indications of positive selection in the Delta and Kappa variants for the nucleocapsid protein. We additionally use a forward selection procedure to fit a model that allows branch-specific estimates of selection along a phylogeny relating the variants, and find that there is variation in the selective pressure across variants for the spike protein. Our results highlight the utility of computational approaches for identifying genomic regions under selection.

Modeling Gastrointestinal Tract Wet Pool Size in Small Ruminants.

The gastrointestinal tract (GIT) wet pool size (GITwps) refers to the total amount of wet contents in GIT, which in small ruminants can reach up to 19% of their body weight (BW). This study aimed to develop models to comprehensively predict GITwps in small ruminants using a meta-regression approach. A dataset was created based on 21 studies, comprising 750 individual records of sheep and goats. Various predictor variables, including BW, sex, breed, species, intake level, physiological states, stages and types of pregnancy, dry matter intake, and neutral detergent fiber intake (NDFI), were initially analyzed through simple linear regression. Subsequently, the variables were fitted using natural logarithm transformations, considering the random effect of the study and residual error, employing a supervised forward selection procedure. Overall, no significant relationship between GITwps and BW (p = 0.326) was observed for animals fed a milk-based diet. However, a strong negative linear relationship (p < 0.001) was found for animals on a solid diet, with the level of restriction influencing GITwps only at the intercept. Furthermore, the prediction of GITwps was independent of sex and influenced by species in cases where individuals were fed ad libitum. Pregnant females showed a noticeable reduction in GITwps, which was more pronounced in cases of multiple pregnancies, regardless of species (p < 0.01). The composition of the diet was found to be the primary factor affecting the modulation of GITwps, with NDFI able to override the species effect (p < 0.0001). Overall, this study sheds light on the factors influencing GITwps in small ruminants, providing valuable insights into their digestive processes and nutritional requirements.

The association of serum adiponectin level with activities of daily living in hospitalized elderly patients with heart failure

BackgroundSeveral studies have reported a relationship between elevated serum adiponectin levels and poor outcomes in patients with heart failure (HF). However, data on the activities of daily living (ADL) in elderly patients with HF are limited. MethodsWe evaluated 218 hospitalized elderly (≥65 years) patients with HF who underwent a comprehensive cardiac rehabilitation (CR) program during hospitalization. Serum adiponectin levels were measured before discharge. The Barthel index (BI) score was evaluated at discharge. Low ADL was defined as a BI score < 85. ResultsSerum adiponectin levels were significantly associated with low ADL [p = 0.03; odds ratio (OR), 1.024, per 1.0 μg/mL increase]. In logistic or regression analyses adjusted for age, sex, body mass index, and estimated glomerular filtration rate, high adiponectin levels (≥16.2 μg/mL) were significantly associated with low ADL (p = 0.04; OR, 2.53), malnutrition (p < 0.01; OR, 2.88), and 6-min walk distance (p = 0.04; β = −17.5). In the multivariate analysis adjusted for conventional risk factors of low ADL, high adiponectin levels were also significantly associated with low ADL (p = 0.03; OR, 2.68). In the stepwise forward selection procedure, a high adiponectin level was an independent determinant of low ADL (p = 0.02; R2 = 0.0262). Both net reclassification improvement (0.53; p < 0.01) and integrated discrimination improvement (0.02; p = 0.01) improved significantly after the addition of high adiponectin level to conventional risk factors. In the regression analysis adjusted for age and sex, serum adiponectin levels were significantly (p < 0.0025) negatively associated with abdominal visceral and subcutaneous adipose tissue areas, body weight, body mass index, and serum triglyceride levels. ConclusionsHigh serum adiponectin levels were not only significantly associated with an increased risk of low ADL, but also with an increased risk of malnutrition and low physical activity in elderly patients with HF after the in-hospital CR program.

Risk Factors Associated With Suicidal Ideation in Individuals With Spinal Cord Injury

ObjectiveTo assess the frequency of suicidal ideation (SI) among individuals with chronic spinal cord injury (SCI) and to identify risk factors associated with SI. DesignCross-sectional. SettingCommunity setting. ParticipantsTwo hundred and forty-six individuals with chronic SCI participating in the Spinal Cord Injury Model Systems at a Level 1 Trauma center. InterventionsNot applicable. Main Outcome MeasureSI, as assessed by question 9 of the Patient Health Questionnaire-9. ResultsTwenty-seven (11.0%) participants endorsed SI; 6 (22.2%) of whom endorsed active SI with a plan of self-harm. Participants who endorsed SI had significantly higher depressive symptoms, lower resilience, and lower satisfaction with life (all Ps<.001). They also had lower perceived health (P<.001), Craig Handicap Assessment & Reporting Technique Short Form (CHART-SF) physical independence (P=.013), and Spinal Cord Injury – Functional Index with Assistive Technology domains of basic mobility (P=.003), self-care (P=.042), and fine motor skills (P=.035). However, participants who endorsed SI were not significantly different in re-hospitalization rates and in other domains of CHART-SF and SCI-AT. Logistic regression, with a forward selection procedure, was used to identify significant predictors of endorsing SI in the context of multiple associated variables. Depressive symptoms (odds ratio [OR]=1.18, P=.020), resilience (OR=0.85, P=.003), and physical independence (OR=0.98, P=.019) remained significant predictors of SI. ConclusionStudy findings suggest higher levels of SI among people with SCI, a substantial proportion of whom have active SI. Individuals with SCI who endorse SI have greater burden of poor physical and mental health, as well as poorer functional status and adaptation. Interventions targeting multiple dimensions of quality of life may help reduce risk of SI and suicide among individuals with SCI.

A Spatio-Temporal Analysis of OECD Member Countries’ Health Care Systems: Effects of Data Missingness and Geographically and Temporally Weighted Regression on Inference

Determinants of health care quality and efficiency are of importance to researchers, policy-makers, and public health officials as they allow for improved human capital and resource allocation as well as long-term fiscal planning. Statistical analyses used to understand determinants have neglected to explicitly discuss how missing data are handled, and consequently, previous research has been limited in inferential capability. We study OECD health care data and highlight the importance of transparency in the assumptions grounding the treatment of data missingness. Attention is drawn to the variation in ordinary least squares coefficient estimates and performance resulting from different imputation methods, and how this variation can undermine statistical inference. We also suggest that parametric regression models used previously are limited and potentially ill-suited for analysis of OECD data due to the inability to deal with both spatial and temporal autocorrelation. We propose the use of an alternative method in geographically and temporally weighted regression. A spatio-temporal analysis of health care system efficiency and quality of care across OECD member countries is performed using four proxy variables. Through a forward selection procedure, medical imaging equipment in a country is identified as a key determinant of quality of care and health outcomes, while government and compulsory health insurance expenditure per capita is identified as a key determinant of health care system efficiency.

Patient Factors in the Dose Selection of Oral Sumatriptan for Acute Migraine: A Post Hoc Analysis of Two Randomized Controlled Studies.

Patients are seeking greater involvement in their healthcare. It therefore may be beneficial to provide guidance on initial oral sumatriptan dose selection for the treatment of acute migraine in nontraditional settings, such as telehealth and other remote forms of medical care. We sought to determine whether clinical or demographic factors are predictive of oral sumatriptan dose preference. This was a post hoc analysis of two clinical studies designed to determine preference for 25, 50, or 100mg oral sumatriptan. Patients were aged 18-65years with at least a 1year history of migraine and experienced, on average, between one and six severe or moderately severe migraine attacks per month, with or without aura. Predictive factors were demographic measures, medical history, and migraine characteristics. Possibly predictive factors were identified using three analyses: classification and regression tree analysis, marginal significance (P < 0.1) within a full-model logistic regression, and/or selection within a forward-selection procedure in a logistic regression. A reduced model containing the variables identified in the preliminary analyses was developed. Due to differences in study design, data were not combined. A dose preference was expressed by 167 patients in Study 1 and 222 patients in Study 2. Gender and medical history of urologic and/or psychological conditions in Study 1 and duration of migraine history, height, and medical history of endocrine or neurologic disease and headache severity in Study 2 were identified as possibly predictive. The predictive model showed low positive predictive value (PPV; 23.8%) and low sensitivity (21.7%) for Study 1. For Study 2, the model showed moderate PPV (60.0%) but low sensitivity (10.9%). No clinical or demographic characteristics alone or in combination were consistently or strongly associated with preference for oral sumatriptan dose level. The studies on which this paper is based were conducted before trial registration indexes were introduced.

Atrial fibrillation classification based on the 2D representation of minimal subset ECG and a non-deep neural network.

Atrial fibrillation (AF) is the most common cardiac arrhythmia, and its early detection is critical for preventing complications and optimizing treatment. In this study, a novel AF prediction method is proposed, which is based on investigating a subset of the 12-lead ECG data using a recurrent plot and ParNet-adv model. The minimal subset of ECG leads (II &V1) is determined via a forward stepwise selection procedure, and the selected 1D ECG data is transformed into 2D recurrence plot (RP) images as an input to train a shallow ParNet-adv Network for AF prediction. In this study, the proposed method achieved F1 score of 0.9763, Precision of 0.9654, Recall of 0.9875, Specificity of 0.9646, and Accuracy of 0.9760, which significantly outperformed solutions based on single leads and complete 12 leads. When studying several ECG datasets, including the CPSC and Georgia ECG databases of the PhysioNet/Computing in Cardiology Challenge 2020, the new method achieved F1 score of 0.9693 and 0.8660, respectively. The results suggested a good generalization of the proposed method. Compared with several state-of-art frameworks, the proposed model with a shallow network of only 12 depths and asymmetric convolutions achieved the highest average F1 score. Extensive experimental studies proved that the proposed method has a high potential for AF prediction in clinical and particularly wearable applications.

Factors Associated With the Development of Tinnitus and With the Degree of Annoyance Caused by Newly Developed Tinnitus.

Objectives:Tinnitus is highly prevalent, but only a few risk factors for developing tinnitus are known and little is known about factors associated with the degree of annoyance of new-onset tinnitus. Longitudinal analysis can reveal risk factors associated with the development of tinnitus and might lead to targeted prevention. The aim of this study is twofold. (1) To identify risk factors that are longitudinally associated with the odds of developing tinnitus 5 years later. (2) To identify factors that are cross-sectionally associated with tinnitus annoyance in adults with new-onset tinnitus.Methods:Baseline, 5-year, and 10-year follow-up data of participants in the Netherlands Longitudinal Study on Hearing (NL-SH) were used. The NL-SH is a web-based prospective cohort study, which started in 2006 and includes both normal hearing and hearing-impaired adults aged 18 to 70 years at baseline. The NL-SH uses an online digit-triplet speech-in-noise test to asses speech recognition ability in noise, and online questionnaires on various aspects of life. At follow-up, participants are asked (1) if they suffer from tinnitus and (2) to rate tinnitus annoyance on a 0 to 100 numeric rating scale. We investigated whether demographic (age, sex, living arrangement, educational level), lifestyle (history of tobacco smoking, alcohol use), health (asthma, severe heart disease, hypertension, history of stroke, osteoarthritis, rheumatoid arthritis, epilepsy, multiple sclerosis, and migraine), hearing (speech recognition ability in noise, hyperacusis, and occupational noise exposure), and psychological variables (distress, somatization, depression, and anxiety) were potential risk factors for new-onset tinnitus, or associated with annoyance caused by new-onset tinnitus. Generalized estimating equations were used to longitudinally analyze the association between potential risk factors and new-onset tinnitus measured 5 years later. A multivariable association model was constructed using a forward selection procedure with p < 0.05 for model entry. Linear regression analysis was used to cross-sectionally analyze the association between potential factors and tinnitus annoyance in new-onset tinnitus. For this purpose, a multivariable association model was constructed using a forward selection procedure with p <0.05 for model entry.Results:In total, 734 participants without tinnitus at baseline were included, from which 137 participants reported to suffer from new-onset tinnitus 5 or 10 years later. Risk factors for new-onset tinnitus were history of smoking (odds ratio 1.5, 95% confidence interval [CI] 1.0 to 2.2, p = 0.027) and higher levels of somatization (odds ratio 2.0, 95% CI 1.2 to 3.3, overall p = 0.024). Factors associated with the degree of tinnitus annoyance were increased levels of anxiety (β = 11.6, 95% CI 2.3-20.8, overall p = 0.035) and poor speech recognition ability in noise (β = 13.5, 95% CI, 4.4 to 22.6, overall p = 0.014).Conclusions:Higher levels of somatization and a history of smoking were found to be risk factors for new-onset tinnitus 5 years later. Anxiety and poor speech recognition ability in noise were associated with higher degrees of tinnitus annoyance in new-onset tinnitus. Somatization deserves to be addressed in future research and clinical practice as it might provide part of a model for the development of chronic tinnitus.

Optimizing Analog Ensembles for Sub-Daily Precipitation Forecasts

This study systematically explores existing and new optimization techniques for analog ensemble (AnEn) post-processing of hourly to daily precipitation forecasts over the complex terrain of southwest British Columbia, Canada. An AnEn bias-corrects a target model forecast by searching for past dates with similar model forecasts (i.e., analogs), and using the verifying observations as ensemble members. The weather variables (i.e., predictors) that select the best past analogs vary among stations and seasons. First, different predictor selection techniques are evaluated and we propose an adjustment in the forward selection procedure that considerably improves computational efficiency while preserving optimization skill. Second, temporal trends of predictors are used to further enhance predictive skill, especially at shorter accumulation windows and longer forecast horizons. Finally, this study introduces a modification in the analog search that allows for selection of analogs within a time window surrounding the target lead time. These supplemental lead times effectively expand the training sample size, which significantly improves all performance metrics—even more than the predictor weighting and temporal-trend optimization steps combined. This study optimizes AnEns for moderate precipitation intensities but also shows good performance for the ensemble median and heavier precipitation rates. Precipitation is most challenging to predict at finer temporal resolutions and longer lead times, yet those forecasts see the largest enhancement in predictive skill from AnEn post-processing. This study shows that optimization of AnEn post-processing, including new techniques developed herein, can significantly improve computational efficiency and forecast performance.