Nuisance Factors Research Articles

Semiparametric Additive Modeling of the Restricted Mean Survival Time.

Analysis of the restricted mean survival time (RMST) has become increasingly common in biomedical studies during the last decade as a means of estimating treatment or covariate effects on survival. Advantages of RMST over the hazard ratio (HR) include increased interpretability and lack of reliance on the often tenuous proportional hazards assumption. Some authors have argued that RMST regression should generally be the frontline analysis as opposed to methods based on counting process increments. However, in order for the use of the RMST to be more mainstream, it is necessary to broaden the range of data structures to which pertinent methods can be applied. In this report, we address this issue from two angles. First, most of existing methodological development for directly modeling RMST has focused on multiplicative models. An additive model may be preferred due to goodness of fit and/or parameter interpretation. Second, many settings encountered nowadays feature high-dimensional categorical (nuisance) covariates, for which parameter estimation is best avoided. Motivated by these considerations, we propose stratified additive models for direct RMST analysis. The proposed methods feature additive covariate effects. Moreover, nuisance factors can be factored out of the estimation, akin to stratification in Cox regression, such that focus can be appropriately awarded to the parameters of chief interest. Large-sample properties of the proposed estimators are derived, and a simulation study is performed to assess finite-sample performance. In addition, we provide techniques for evaluating a fitted model with respect to risk discrimination and predictive accuracy. The proposed methods are then applied to liver transplant data to estimate the effects of donor characteristics on posttransplant survivaltime.

Reducing risk of false positives in the in vivo comet assay and improving result reliability

The risk of generating false positive in vivo comet assay results can be increased when procedural bias and/or technical variability is poorly controlled. This has been an ongoing concern since comet was first introduced into regulatory safety testing. But the proprietary nature of regulated studies and the 3Rs have limited the ability to conduct and publish the comparative in vivo studies necessary to determine the effect these factors can have on comet assay results when substances other than well characterized positive control compounds are evaluated in multiple tissues. That changed when Helix3 was asked to repeat for regulatory submission three independent in vivo comet studies with positive results generated by three other laboratories evaluating the effects of three different test substances on the liver, duodenum, and stomach. We repeated each study using the same test substance and experimental design as the original labs but with our standard quality control methods implemented to reduce procedural bias and variability. In every case, we generated negative results that regulatory authorities accepted over the initial positive results due to evidence of high technical variability and procedural bias in the original labs and studies. Meanwhile, the International Workshop on Genotoxicity (IWGT) compared >14 years of Helix3 comet historical control data (HCD) to HCD from 6 other experienced comet laboratories and concluded that our data exhibited the highest overall background % tail DNA levels with the lowest inter-study variability resulting in the highest quality HCD of all the labs evaluated. These case studies and the IWGT report suggest that our enhanced quality control methods and higher (>2 % mean of slide median tail DNA) background levels can effectively mitigate the nuisance factors that can generate false positive in vivo comet assay results. To facilitate a better understanding of the technical parameters that can significantly influence the comet results, we describe our enhanced procedures with justifications and examples.

Analysis of the longitudinal stability of human plasma miRNAs and implications for disease biomarkers

There is great interest in developing clinical biomarker assays that can aid in non-invasive diagnosis and/or monitoring of human diseases, such as cancer, cardiovascular disease, and neurological diseases. Yet little is known about the longitudinal stability of miRNAs in human plasma. Here we assessed the intraindividual longitudinal stability of miRNAs in plasma from healthy human adults, and the impact of common factors (e.g., hemolysis, age) that may confound miRNA data. We collected blood by venipuncture biweekly over a 3-month period from 22 research participants who had fasted overnight, isolated total RNA, then performed miRNA qPCR. Filtering and normalization of the qPCR data revealed amplification of 134 miRNAs, 74 of which had high test–retest reliability and low percentage level drift, meaning they were stable in an individual over the 3-month time period. We also determined that, of nuisance factors, hemolysis and tobacco use have the greatest impact on miRNA levels and variance. These findings support that many miRNAs show intraindividual longitudinal stability in plasma from healthy human adults, including some reported as candidate biomarkers for Alzheimer’s disease.

OOD-CV-v2 : An extended Benchmark for Robustness to Out-of-Distribution Shifts of Individual Nuisances in Natural Images.

Enhancing the robustness of vision algorithms in real-world scenarios is challenging. One reason is that existing robustness benchmarks are limited, as they either rely on synthetic data or ignore the effects of individual nuisance factors. We introduce OOD-CV-v2, a benchmark dataset that includes out-of-distribution examples of 10 object categories in terms of pose, shape, texture, context and the weather conditions, and enables benchmarking of models for image classification, object detection, and 3D pose estimation. In addition to this novel dataset, we contribute extensive experiments using popular baseline methods, which reveal that: 1) Some nuisance factors have a much stronger negative effect on the performance compared to others, also depending on the vision task. 2) Current approaches to enhance robustness have only marginal effects, and can even reduce robustness. 3) We do not observe significant differences between convolutional and transformer architectures. We believe our dataset provides a rich test bed to study robustness and will help push forward research in this area. Our dataset is publically available online, https://genintel.mpi-inf.mpg.de/ood-cv-v2.html.

The association between calf birth weight and the postcalving performance of its dairy dam in the absence of dystocia

The objective of the study was to quantify the association between the birth weight of a calf and the subsequent performance of its dairy dam in the absence of any recorded calving assistance. A total of 11,592 lactation records from 4,549 spring-calving dairy cows were used. The association between a series of quantitative cow performance metrics (dependent variable) and calf birth weight (independent variable) was determined using linear mixed models; logistic regression was used where the dependent variable was binary. Nuisance factors in the models were calf sex, heterosis coefficient of both the cow and calf, dry period length immediately before the birth of the calf, cow age at calving relative to the median cow age per parity, breed proportion of the cow, cow live weight between 100 and 200 d of lactation relative to the mean cow weight per parity, and contemporary group. Calf birth weight was included in the model as either a continuous or a categorical variable. Primiparous and multiparous cows were analyzed separately. Mean (SD) calf birth weight was 36.2 (6.8) kg. In primiparous cows, calf birth weight was associated with milk yield in the first 60 d of lactation, calving to first service interval, calving body weight (BW), and both nadir BW and body condition score (BCS). In multiparous cows, calf birth weight was associated with total milk, fat, and protein yield in the first 60 and 305 d of lactation, peak milk yield, total milk solids, both calving and nadir BW, and BCS loss from calving to nadir. Relative to primiparous cows that gave birth to calves weighing 34 to 37 kg (i.e., population mean), their contemporaries who gave birth to calves that weighed 15 to 29 kg produced 9.82 kg more milk in the first 60 d of lactation, had a 2-d shorter interval to first service, and were 8.08 kg and 5.51 kg lighter at calving and nadir BW, respectively; the former was also 0.05 units lower in BCS (5-point scale, 1 = emaciated and 5 = obese) at nadir. Relative to multiparous cows that gave birth to calves that were 34 to 37 kg birth weight, multiparous cows that gave birth to calves that were 15 to 29 kg yielded 59.63 kg, 2.44 kg, and 1.76 kg less milk, fat, and protein, respectively, in the first 60 d of lactation; produced 17.69 kg less milk solids throughout the 305-d lactation; and were also 10.49 kg lighter at nadir and lost 0.01 units more BCS to nadir. In a separate series of analyses, sire breed was added to the model as a fixed effect with and without calf birth weight. When calf birth weight was not adjusted for, 60-d milk yield for multiparous cows who gave birth to calves sired by a traditional beef breed (i.e., Angus, Hereford) produced 59.63 kg more milk than multiparous cows who gave birth to calves sired by a Holstein-Friesian. Hence, calf birth weight is associated with some subsequent dam performance measures; however, where associations do exist, the effect is biologically small.

In vivo genotoxicity testing strategies: Report from the 8th International Workshop on Genotoxicity Testing (IWGT)

AbstractThe in vivo working group (WG) considered three topics: acceptable maximum doses for negative erythrocyte micronucleus (MN) tests, validation status of MN assays in non‐hematopoietic tissues, and nuisance factors in the comet assay. The WG reached agreement on many issues, including: negative erythrocyte MN studies should be acceptable if dosing is conducted to Organisation for Economic Co‐operation and Development (OECD) test guideline (TG) 474 recommendations and if sufficient bone marrow exposure is demonstrated; consensus on the evidence required to demonstrate “sufficient” exposure was not reached. The liver MN test using six‐week‐old rats is sufficiently validated to develop an OECD TG, but the impact of animal age warrants additional study. Ki‐67 is a reliable marker for cellular proliferation in hepatocytes. The gastrointestinal tract MN test is useful for detecting poorly absorbed or rapidly degraded aneugens, and for genotoxic metabolites formed in the colon. Although current validation data are insufficient to support the development of an OECD TG, the methodologies are sufficient to consider as an appendix to OECD TG474. Comparison of comet assay results to laboratory historical control data (HCD) should not be used in data evaluation, unless the HCD distribution is demonstrated to be stable and the predominant source of HCD variation is due to animal, not study, factors. No universally acceptable negative control limit for any tissue was identified. Methodological differences in comet studies can result in variable data interpretations; more data are required before best practice recommendations can be made. Hedgehogs alone are unreliable indicators of cytotoxicity and additional investigations into cytotoxicity markers are required.

Learning Foreground Information Bottleneck for few-shot semantic segmentation

Few-shot semantic segmentation aims to segment unseen classes with only a few annotated samples, which has great values for the real-world application in the wild. However, since the target class is treated as the background in the training, the network tends to extract much irrelevant nuisance factors, which results in the feature undermining problem for the target class. Consequently, it is difficult to produce an accurate segmentation map. To address this problem, in this paper, we apply the information bottleneck theory to few-shot semantic segmentation and propose the Foreground Information Bottleneck (FIB) module. Based on the support information, FIB module filters out the irrelevant information and promotes the foreground-related feature paradigms. Meanwhile, to solve the intractable mutual information and enable the end-to-end optimization of FIB module, we derive the Foreground Information Bottleneck Loss (FIBLoss) according to the inherent attribute of few-shot segmentation. Moreover, since there exists severe noise interference in the wild, we design a Target Information Refinement (TIR) block to further exploit discriminative cues of foreground. TIR block calculates the pairwise interaction and exploits the detailed information of the foreground object, which is beneficial to the feature refinement. Extensive experiments on two challenging datasets reflect the proposed FIB module significantly improves the performance of few-shot segmentation and delivers the state-of-the-art results.

Optimization of degree of conformance for multi‐response systems considering model imprecision: A data‐driven metaheuristic approach

AbstractQuality‐based design optimization is a process consisting of factor identification, finding an empirical/physical model between responses and factors, and finding a factor setting leading to the best quality levels. Response surface methodology is a well‐grounded data‐driven approach that applies a collection of statistical‐mathematical techniques to achieve the design requirements. Today's complex systems are usually affected by nuisance factors and have more than one output characteristic with some degree of correlation. This study aims to develop two‐stage stochastic programming for a multi‐response optimization model based on the degree of conformance as the most comprehensive metric for such design problems. The proposed approach can include the model imprecision in the optimization procedure using the probabilistic properties of the estimated parameters. This method would be applied in highly sensitive processes or costly operations that require a more accurate setting to avoid extra reworks or scraps. A simulated annealing algorithm with a boosted scenario checking modification has been applied to cope with the expensive computations for problem instances with a large sample size of uncertain parameters. The proposed model has also analyzed two cases from the literature, and the results support its superiority compared to the existing approaches.

Qualitatively Different Delay-Dependent Working Memory Distortions in People With Schizophrenia and Healthy Control Participants

BackgroundImpairments in working memory (WM) have been well documented in people with schizophrenia (PSZ). However, these quantitative WM impairments can often be explained by nonspecific factors, such as impaired goal maintenance. Here, we used a spatial orientation delayed response task to explore a qualitative difference in WM dynamics between PSZ and healthy control participants (HCs). More specifically, we took advantage of the discovery that WM representations may drift either toward or away from previous trial targets (serial dependence). We tested the hypothesis that WM representations would drift toward the previous trial target in HCs but away from the previous trial target in PSZ. MethodsWe assessed serial dependence in PSZ (n = 31) and HCs (n = 25) using orientation as the to-be-remembered feature and memory delays lasting from 0 to 8 seconds. Participants were asked to remember the orientation of a teardrop-shaped object and reproduce the orientation after a delay period of varying length. ResultsConsistent with prior studies, we found that current trial memory representations were less precise in PSZ than in HCs. We also found that WM for the current trial orientation drifted toward the previous trial orientation in HCs (representational attraction) but drifted away from the previous trial orientation in PSZ (representational repulsion). ConclusionsThese results demonstrate a qualitative difference in WM dynamics between PSZ and HCs that cannot be easily explained by nuisance factors such as reduced effort. Most computational neuroscience models also fail to explain these results because they maintain information solely by means of sustained neural firing, which does not extend across trials. The results suggest a fundamental difference between PSZ and HCs in longer-term memory mechanisms that persist across trials, such as short-term potentiation and neuronal adaptation.

Invariant Representations with Stochastically Quantized Neural Networks

Representation learning algorithms offer the opportunity to learn invariant representations of the input data with regard to nuisance factors. Many authors have leveraged such strategies to learn fair representations, i.e., vectors where information about sensitive attributes is removed. These methods are attractive as they may be interpreted as minimizing the mutual information between a neural layer's activations and a sensitive attribute. However, the theoretical grounding of such methods relies either on the computation of infinitely accurate adversaries or on minimizing a variational upper bound of a mutual information estimate. In this paper, we propose a methodology for direct computation of the mutual information between neurons in a layer and a sensitive attribute. We employ stochastically-activated binary neural networks, which lets us treat neurons as random variables. Our method is therefore able to minimize an upper bound on the mutual information between the neural representations and a sensitive attribute. We show that this method compares favorably with the state of the art in fair representation learning and that the learned representations display a higher level of invariance compared to full-precision neural networks.

Evaluation of the applicability of selected analytical techniques for determining the characteristics of humic substances sourced from by-products of the wastewater treatment process

Although humic substances (HSs) are among the most valuable compounds in the environment, they often constitute nuisance factors for wastewater treatment plants (WWTPs). However, their recovery from by-products of WWTPs opens up opportunities for their utilization. Therefore, this study aimed to evaluate the suitability of selected analytical methods for determining the structure, properties, and possible utilization of HSs originating from WWTPs based on model humic compounds (MHCs). As a consequence, the study proposed separate approaches for the initial and in-depth characterization of HSs. The results demonstrate that UV-Vis spectroscopy may be recognized as a cost-effective approach for the preliminary characterization of HSs. Such a method indeed gives similar information on the degree of complexity of MHCs as X-EDS and FTIR, and similarly to them allows for the differentiation of their particular fractions. In turn, X-EDS and FTIR techniques were recommended to be used for in-deep analysis of HSs due to their ability to detect heavy metals and biogenic elements in their structure. Contrarily to other studies, the presented research indicates that only selected absorbance coefficients - A253/A230, Q4/6, and ΔlogK may help to distinguish particular humic fractions and evaluate changes in their behavior, independently of their concentration (coefficient of variation < 20%). Herein, the fluorescence capacities of MHCs were found to be equally affected as their optical properties by the changes in their concentration. Focusing on the obtained results, this study recommends that the quantitative comparison of the properties of HSs should be proceeded by the standardization of their concentration. Herein, the stability of other spectroscopic parameters characterizing solutions MHCs was achieved within a concentration from 40 to 80 mg L−1. Among them, the analyzed MHCs were differentiated the most by the SUVA254 coefficient, which value was almost 4 times higher for SAHSs (8.69) than for ABFASs (2.01).

Professional stress and job satisfaction of nurses working in infectious diseases wards during the SARS-CoV-2 pandemic

The SARS-CoV-2 pandemic brings many challenges to the daily work of nurses who put their lives and health at risk for the general public every day. This causes them to be highly stressed at work, causes mental strain, and lowers the level of job satisfaction. The aim of the research was to assess the level of stress and job satisfaction of nurses working in infectious diseases wards. Materials and methods The survey was conducted among 169 nurses in the period from February to April 2022. The study was conducted using the diagnostic survey method, the technique was a questionnaire survey, and the survey tool was a survey questionnaire provided in a Google Form. The differences between the qualitative variables were analyzed using the χ2 test of independence. We use the χ2 test of independence (non-parametric significance test for two or more independent samples) to test whether the variables measured at the nominal level or higher in the comparison groups differ from each other. The significance level of the study was p &lt;0.05. The calculations were made with the SPSS 25 software. Results Work in direct contact with people infected with the coronavirus was performed by the majority of the respondents. Only about 18.0% of nurses used psychological assistance during the pandemic. Among the harmful and nuisance factors in the workplace, the respondents mentioned stress, contact with potentially infectious material, coronavirus infection despite appropriate personal protective equipment and cuts. Conclusions Stress in the workplace was felt especially due to the need to operate specialized equipment, not yet operated by the respondents. Nurses in the workplace were mainly exposed to stress related to performing difficult and responsible activities. Only a small part of the respondents was definitely satisfied with their work.

How robust is the relationship between neural processing speed and cognitive abilities?

Individual differences in processing speed are consistently related to individual differences in cognitive abilities, but the mechanisms through which a higher processing speed facilitates reasoning remain largely unknown. To identify these mechanisms, researchers have been using latencies of the event-related potential (ERP) to study how the speed of cognitive processes associated with specific ERP components is related to cognitive abilities. Although there is some evidence that latencies of ERP components associated with higher-order cognitive processes are related to intelligence, results are overall quite inconsistent. These inconsistencies likely result from variations in analytic procedures and little consideration of the psychometric properties of ERP latencies in relatively small sample studies. Here we used a multiverse approach to evaluate how different analytical choices regarding references, low-pass filter cutoffs, and latency measures affect the psychometric properties of P2, N2, and P3 latencies and their relations with cognitive abilities in a sample of 148 participants. Latent correlations between neural processing speed and cognitive abilities ranged from -.49 to -.78. ERP latency measures contained about equal parts of measurement error variance and systematic variance, and only about half of the systematic variance was related to cognitive abilities, whereas the other half reflected nuisance factors. We recommend addressing these problematic psychometric properties by recording EEG data from multiple tasks and modeling relations between ERP latencies and covariates in latent variable models. All in all, our results indicate that there is a substantial and robust relationship between neural processing speed and cognitive abilities when those issues are addressed.

Remote auditory assessment using Portable Automated Rapid Testing (PART) and participant-owned devices.

Remote testing of auditory function can be transformative to both basic research and hearing healthcare; however, historically, many obstacles have limited remote collection of reliable and valid auditory psychometric data. Here, we report performance on a battery of auditory processing tests using a remotely administered system, Portable Automatic Rapid Testing. We compare a previously reported dataset collected in a laboratory setting with the same measures using uncalibrated, participant-owned devices in remote settings (experiment 1, n = 40) remote with and without calibrated hardware (experiment 2, n = 36) and laboratory with and without calibrated hardware (experiment 3, n = 58). Results were well-matched across datasets and had similar reliability, but overall performance was slightly worse than published norms. Analyses of potential nuisance factors such as environmental noise, distraction, or lack of calibration failed to provide reliable evidence that these factors contributed to the observed variance in performance. These data indicate feasibility of remote testing of suprathreshold auditory processing using participants' own devices. Although the current investigation was limited to young participants without hearing difficulties, its outcomes demonstrate the potential for large-scale, remote hearing testing of more hearing-diverse populations both to advance basic science and to establish the clinical viability of auditory remote testing.

Deep Learning Approach for Radar-Based People Counting

With the development of deep learning (DL) frameworks in the field of pattern recognition, DL-based algorithms have outperformed handcrafted feature (HF)-based ones in various applications. However, there still exist several challenges in applying the DL framework to a radar-based people counting (RPC) task: The powerful representation capacity of a deep neural network (DNN) learns not only the desired human-induced components but also unwanted nuisance factors, and available data for RPC is usually insufficient to train a huge-sized DNN, leading to an increased possibility of overfitting. To tackle this problem, we propose novel solutions for the successful application of the DL framework to the RPC task from various perspectives. First, we newly formulate the preprocessing pipelines to transform the raw received radar echoes into a better-matched form for a DNN. Second, we devise a novel backbone architecture that reflects the spatiotemporal characteristics of the radar signals, while relieving the burden on training through a parameter efficient design. Finally, an unsupervised pretraining process and a newly defined loss function are proposed for further stabilized network convergence. Several experimental results using real measured data show that the proposed scheme enables an effective utilization of DL for RPC, achieving a significant performance improvement compared to conventional RPC methods.

Theoretical false positive psychology.

A fundamental goal of scientific research is to generate true positives (i.e., authentic discoveries). Statistically, a true positive is a significant finding for which the underlying effect size (δ) is greater than 0, whereas a false positive is a significant finding for which δ equals 0. However, the null hypothesis of no difference (δ = 0) may never be strictly true because innumerable nuisance factors can introduce small effects for theoretically uninteresting reasons. If δ never equals zero, then with sufficient power, every experiment would yield a significant result. Yet running studies with higher power by increasing sample size (N) is one of the most widely agreed upon reforms to increase replicability. Moreover, and perhaps not surprisingly, the idea that psychology should attach greater value to small effect sizes is gaining currency. Increasing N without limit makes sense for purely measurement-focused research, where the magnitude of δ itself is of interest, but it makes less sense for theory-focused research, where the truth status of the theory under investigation is of interest. Increasing power to enhance replicability will increase true positives at the level of the effect size (statistical true positives) while increasing false positives at the level of theory (theoretical false positives). With too much power, the cumulative foundation of psychological science would consist largely of nuisance effects masquerading as theoretically important discoveries. Positive predictive value at the level of theory is maximized by using an optimal N, one that is neither too small nor too large.

MbDenoise: microbiome data denoising using zero-inflated probabilistic principal components analysis

The analysis of microbiome data has several technical challenges. In particular, count matrices contain a large proportion of zeros, some of which are biological, whereas others are technical. Furthermore, the measurements suffer from unequal sequencing depth, overdispersion, and data redundancy. These nuisance factors introduce substantial noise. We propose an accurate and robust method, mbDenoise, for denoising microbiome data. Assuming a zero-inflated probabilistic PCA (ZIPPCA) model, mbDenoise uses variational approximation to learn the latent structure and recovers the true abundance levels using the posterior, borrowing information across samples and taxa. mbDenoise outperforms state-of-the-art methods to extract the signal for downstream analyses.

GKNet: Grasp keypoint network for grasp candidates detection

Contemporary grasp detection approaches employ deep learning to achieve robustness to sensor and object model uncertainty. The two dominant approaches design either grasp-quality scoring or anchor-based grasp recognition networks. This paper presents a different approach to grasp detection by treating it as keypoint detection in image-space. The deep network detects each grasp candidate as a pair of keypoints, convertible to the grasp representation g = {x,y,w,θ} T , rather than a triplet or quartet of corner points. Decreasing the detection difficulty by grouping keypoints into pairs boosts performance. To promote capturing dependencies between keypoints, a non-local module is incorporated into the network design. A final filtering strategy based on discrete and continuous orientation prediction removes false correspondences and further improves grasp detection performance. GKNet, the approach presented here, achieves a good balance between accuracy and speed on the Cornell and the abridged Jacquard datasets (96.9% and 98.39% at 41.67 and 23.26 fps). Follow-up experiments on a manipulator evaluate GKNet using four types of grasping experiments reflecting different nuisance sources: static grasping, dynamic grasping, grasping at varied camera angles, and bin picking. GKNet outperforms reference baselines in static and dynamic grasping experiments while showing robustness to varied camera viewpoints and moderate clutter. The results confirm the hypothesis that grasp keypoints are an effective output representation for deep grasp networks that provide robustness to expected nuisance factors.

Acoustic Modelling From Raw Source and Filter Components for Dysarthric Speech Recognition

Acoustic modelling for automatic dysarthric speech recognition (ADSR) is a challenging task. Data deficiency is a major problem and substantial differences between typical and dysarthric speech complicate the transfer learning. In this paper, we aim at building acoustic models using the raw magnitude spectra of the source and filter components for ADSR. The proposed multi-stream models consist of convolutional, recurrent and fully-connected layers allowing for pre-processing various information streams and fusing them at an optimal level of abstraction. We demonstrate that such a multi-stream processing leverages information encoded in the vocal tract and excitation components and leads to normalising nuisance factors such as speaker attributes and speaking style. This leads to a better handling of dysarthric speech that exhibits large inter- and intra-speaker variabilities and results in a notable performance gain. Furthermore, we analyse the learned convolutional filters and visualise the outputs of different layers after dimensionality reduction to demonstrate how the speaker-related attributes are normalised along the pipeline. We also compare the proposed multi-stream model with various systems based on MFCC, FBank, raw waveform and i-vector, and, study the training dynamics as well as usefulness of the feature normalisation and data augmentation via speed perturbation. On the widely used TORGO and UASpeech dysarthric speech corpora, the proposed approach leads to a competitive performance of up to 35.3% and 30.3% WERs for dysarthric speech, respectively.

The response to genetic merit for milk production in dairy cows differs by cow body weight

Attention is increasing on both cow size and body weight (BW) as energy sinks and thus as contributors to differences in production efficiency among cows. What is not currently clear, however, is how cow BW affects the increase in yield per cow per unit increase in genetic merit for milk production. This void in knowledge was filled in the present study using BW data from 20,470 lactations on 16,980 Holstein-Friesian dairy cows stratified into 4 groups on BW adjusted for differences in parity, days in milk, and body condition score. Using linear mixed models that adjusted for nuisance factors, cow phenotypic milk production variables were regressed on estimates of parental average genetic merit for the respective trait within each stratum of BW defined within contemporary group; estimates of genetic merit were from the national genetic evaluations. Both the intercept and linear regression coefficients on genetic merit were compared across BW strata. The intercepts representing the mean phenotypic yield at a genetic merit of zero differed among BW strata; irrespective of yield trait, the least squares means yield per BW stratum increased numerically as cows got heavier, although not every stepwise increase in BW stratum was associated with significantly greater yield compared with the previous (lighter) stratum. Nonetheless, the yield of the cows in the lightest of the 4 strata was always less than that of the heaviest 2 strata; relative to the lightest stratum, cows in the heaviest BW stratum produced only 3 to 4% more yield. Furthermore, the association between phenotypic yield and its respective measures of genetic merit differed by BW stratum; the response to selection for each of the yield traits was 15 to 23% greater for the heaviest stratum of cows compared with their contemporaries in the lightest stratum. Although BW stratum was associated with mean fat and protein concentration after adjusting for differences in genetic merit for fat and protein concentration, the association did not differ by BW stratum for either fat or protein concentration. The effect of BW on efficiency should consider the association between BW and not only mean phenotypic yield at a given genetic merit, but also how the differences in yield diverge as genetic merit increases.