Independent Validation Data Research Articles

Single-center versus multi-center data sets for molecular prognostic modeling: a simulation study

BackgroundPrognostic models based on high-dimensional omics data generated from clinical patient samples, such as tumor tissues or biopsies, are increasingly used for prognosis of radio-therapeutic success. The model development process requires two independent discovery and validation data sets. Each of them may contain samples collected in a single center or a collection of samples from multiple centers. Multi-center data tend to be more heterogeneous than single-center data but are less affected by potential site-specific biases. Optimal use of limited data resources for discovery and validation with respect to the expected success of a study requires dispassionate, objective decision-making. In this work, we addressed the impact of the choice of single-center and multi-center data as discovery and validation data sets, and assessed how this impact depends on the three data characteristics signal strength, number of informative features and sample size.MethodsWe set up a simulation study to quantify the predictive performance of a model trained and validated on different combinations of in silico single-center and multi-center data. The standard bioinformatical analysis workflow of batch correction, feature selection and parameter estimation was emulated. For the determination of model quality, four measures were used: false discovery rate, prediction error, chance of successful validation (significant correlation of predicted and true validation data outcome) and model calibration.ResultsIn agreement with literature about generalizability of signatures, prognostic models fitted to multi-center data consistently outperformed their single-center counterparts when the prediction error was the quality criterion of interest. However, for low signal strengths and small sample sizes, single-center discovery sets showed superior performance with respect to false discovery rate and chance of successful validation.ConclusionsWith regard to decision making, this simulation study underlines the importance of study aims being defined precisely a priori. Minimization of the prediction error requires multi-center discovery data, whereas single-center data are preferable with respect to false discovery rate and chance of successful validation when the expected signal or sample size is low. In contrast, the choice of validation data solely affects the quality of the estimator of the prediction error, which was more precise on multi-center validation data.

Automatic Triage of 12‐Lead ECGs Using Deep Convolutional Neural Networks

BACKGROUNDThe correct interpretation of the ECG is pivotal for the accurate diagnosis of many cardiac abnormalities, and conventional computerized interpretation has not been able to reach physician‐level accuracy in detecting (acute) cardiac abnormalities. This study aims to develop and validate a deep neural network for comprehensive automated ECG triage in daily practice.METHODS AND RESULTSWe developed a 37‐layer convolutional residual deep neural network on a data set of free‐text physician‐annotated 12‐lead ECGs. The deep neural network was trained on a data set with 336.835 recordings from 142.040 patients and validated on an independent validation data set (n=984), annotated by a panel of 5 cardiologists electrophysiologists. The 12‐lead ECGs were acquired in all noncardiology departments of the University Medical Center Utrecht. The algorithm learned to classify these ECGs into the following 4 triage categories: normal, abnormal not acute, subacute, and acute. Discriminative performance is presented with overall and category‐specific concordance statistics, polytomous discrimination indexes, sensitivities, specificities, and positive and negative predictive values. The patients in the validation data set had a mean age of 60.4 years and 54.3% were men. The deep neural network showed excellent overall discrimination with an overall concordance statistic of 0.93 (95% CI, 0.92–0.95) and a polytomous discriminatory index of 0.83 (95% CI, 0.79–0.87).CONCLUSIONSThis study demonstrates that an end‐to‐end deep neural network can be accurately trained on unstructured free‐text physician annotations and used to consistently triage 12‐lead ECGs. When further fine‐tuned with other clinical outcomes and externally validated in clinical practice, the demonstrated deep learning–based ECG interpretation can potentially improve time to treatment and decrease healthcare burden.

A comprehensive review and performance evaluation of bioinformatics tools for HLA class I peptide-binding prediction.

Human leukocyte antigen class I (HLA-I) molecules are encoded by major histocompatibility complex (MHC) class I loci in humans. The binding and interaction between HLA-I molecules and intracellular peptides derived from a variety of proteolytic mechanisms play a crucial role in subsequent T-cell recognition of target cells and the specificity of the immune response. In this context, tools that predict the likelihood for a peptide to bind to specific HLA class I allotypes are important for selecting the most promising antigenic targets for immunotherapy. In this article, we comprehensively review a variety of currently available tools for predicting the binding of peptides to a selection of HLA-I allomorphs. Specifically, we compare their calculation methods for the prediction score, employed algorithms, evaluation strategies and software functionalities. In addition, we have evaluated the prediction performance of the reviewed tools based on an independent validation data set, containing 21101 experimentally verified ligands across 19 HLA-I allotypes. The benchmarking results show that MixMHCpred 2.0.1 achieves the best performance for predicting peptides binding to most of the HLA-I allomorphs studied, while NetMHCpan 4.0 and NetMHCcons 1.1 outperform the other machine learning-based and consensus-based tools, respectively. Importantly, it should be noted that a peptide predicted with a higher binding score for a specific HLA allotype does not necessarily imply it will be immunogenic. That said, peptide-binding predictors are still very useful in that they can help to significantly reduce the large number of epitope candidates that need to be experimentally verified. Several other factors, including susceptibility to proteasome cleavage, peptide transport into the endoplasmic reticulum and T-cell receptor repertoire, also contribute to the immunogenicity of peptide antigens, and some of them can be considered by some predictors. Therefore, integrating features derived from these additional factors together with HLA-binding properties by using machine-learning algorithms may increase the prediction accuracy of immunogenic peptides. As such, we anticipate that this review and benchmarking survey will assist researchers in selecting appropriate prediction tools that best suit their purposes and provide useful guidelines for the development of improved antigen predictors in the future.

Lymphocyte-Monocyte-Neutrophil Index: A Predictor of Severity of Coronavirus Disease 2019 Patients Produced by Sparse Principal Component Analysis

Background: It is important to recognize severe ill coronavirus disease 2019 (COVID-19) patients from moderate ones in order to save more lives. We attempted to present an predictor for disease severity from clinical laboratory markers using sparse principal component analysis method. Methods: Forty-four clinical characteristics and laboratory markers of 82 COVID-19 patients (Hefei cohort) from the First Affiliated Hospital of University of Science and Technology of China (USTC) were analyzed retrospectively and sparse principal component analysis (SPCA) was performed to examine the correlation between the markers and extract relevant features. The controlling parameter alpha of SPCA was adjusted for better variable selection. Then the produced principal components (PCs) by SPCA were subjected to multivariate logistic regression for disease severity prediction, and the significant PCs were selected. Then, a Lymphocyte-Monocyte-Neutrophil index (LMN index) was deduced from the significant PCs and was used for disease severity prediction. Furthermore, an independent cohort including 169 COVID-19 patients (Nanchang Cohort) from the First Affiliated Hospital of NanChang University was used as a validation dataset and prediction efficiency of LMN index and classical clinical markers were also evaluated. Findings: Using SPCA, the first to thirteenth PCs accounted for 81·7% of the cumulative proportion variance of the original 44 clinical characteristics and laboratory markers. Multivariate logistic regression revealed the PC1 was significantly associated with disease severity with odds ratio of 74272·28 (623·83 - 178483250). When the controlling parameter alpha was adjusted to 0·001, the PC1 is only dependent on five laboratory markers: lymphocyte count (LYM), lymphocyte percentage (LYM%), neutrophil count (NEU), monocyte count (MONO) ,and serum phosphorus. LMN index determined by LYM, LYM%, NEU ,and MONO was deduced from the PC1 and significant relationships were investigated between LMN indices with age, comorbidity status and CD4+ ,and CD8 T lymphocyte counts. More important, during hospitalization, LMN indices decreased obviously as treatment takes effect, and they declined more sharply for mild ill COVID-19 patients compared with those of severe ill ones. When used to predict disease progression, the LMN index gave the accuracy of 0·780 and 0·760 in the training data (Hefei cohort) and the independent validation data (Nanchang Cohort) respectively, which was more efficient than classical clinical markers. Interpretation: Using SPCA method, the LMN index determined by four blood routine test markers was deduced. It showed robust disease severity prediction efficiency of COVID-19 patients and have the potential for clinical applications. Funding Statement: Fundamental Research Funds for the Central Universities of China(No. YD9110002001). Declaration of Interests: XLM reports grants from Fundamental Research Funds for the Central Universities of China. All other authors declare no competing interests. Ethics Approval Statement: This study was approved by the Ethics Committee of the First Affiliated Hospital of USTC and the Ethics Committee of the First Affiliated Hospital of NanChang University.

Augmenting Interpretation of Chest Radiographs With Deep Learning Probability Maps.

Pneumonia is a common clinical diagnosis for which chest radiographs are often an important part of the diagnostic workup. Deep learning has the potential to expedite and improve the clinical interpretation of chest radiographs. While earlier approaches have emphasized the feasibility of "binary classification" to accomplish this task, alternative strategies may be possible. We explore the feasibility of a "semantic segmentation" deep learning approach to highlight the potential foci of pneumonia on frontal chest radiographs. In this retrospective study, we trained a U-net convolutional neural network (CNN) to predict pixel-wise probability maps for pneumonia using a public data set provided by the Radiological Society of North America (RSNA) comprised of 22,000 radiographs and radiologist-defined bounding boxes. We reserved 3684 radiographs as an independent validation data set and assessed overall performance for localization using Dice overlap and classification performance using the area under the receiver-operator characteristic curve. For classification/detection of pneumonia, area under the receiver-operator characteristic curve on frontal radiographs was 0.854 with a sensitivity of 82.8% and specificity of 72.6%. Using this strategy of neural network training, probability maps localized pneumonia to lung parenchyma for essentially all validation cases. For segmentation of pneumonia for positive cases, predicted probability maps had a mean Dice score (±SD) of 0.603±0.204, and 60.0% of these had a Dice score >0.5. A "semantic segmentation" deep learning approach can provide a probabilistic map to assist in the diagnosis of pneumonia. In combination with the patient's history, clinical findings and other imaging, this strategy may help expedite and improve diagnosis.

A Validated Model to Predict Postoperative Symptom Severity After Mandibular Third Molar Removal

The individualized prediction of postoperative symptom severity is essential for selecting interventions after mandibular third molar (M3M) removal. The purpose of the present study was to develop and validate a nomogram for personal prediction of postoperative symptom severity. A prospective cohort study was performed in the Stomatology Hospital of Tianjin Medical University. The sample was divided into training and testing data sets by time. The demographic, anatomic, radiographic, and operative variables were recorded. The self-reported postoperative symptom severity was recorded and defined as the primary outcome variable. Stepwise forward algorithms were applied to informative predictors based on Akaike's information criterion. Multivariable logistic regression analysis was used to develop the nomogram. An independent testing data set was used to validate the nomogram. Receiver operating characteristic curves and the Hosmer-Lemeshow test were used to assess model performance. P<.05 was considered to indicate statistical significance. The sample included 321 subjects who had undergone M3M removal. An independent validation data set included 103 consecutive patients. The median operation time was 15.0minutes (interquartile range, 8.3 to 21.6minutes) in the training data set (n=218). Patients with serious postoperative symptoms accounted for 48.6 and 47.6% of the training and testing data sets, respectively. Gender, age, smoking status, operation time, Pell-Gregory ramus classification, and preoperative symptoms were identified as predictors and assembled into the nomogram. The area under curve demonstrated adequate discrimination in the validation data set (0.69; 95% confidence interval, 0.59 to 0.80). The nomogram was well calibrated, with a Hosmer-Lemeshow χ2 statistic of 6.33 (P=.78) in the testing data set. The confusion matrix was also summarized, and the accuracy was 63.3 and 65.1% in the training and testing data set, respectively. The present study has proposed an effective nomogram with potential application in facilitating the individualized prediction of postoperative symptom severity after M3M removal.

Establishment of Multiple Myeloma Diagnostic Model Based on Logistic Regression in Clinical Laboratory.

Due to the insidious onset of multiple myeloma (MM), missed diagnosis and misdiagnosis have a serious impact on the health of MM patients. Simple, rapid, and valid laboratory screening is critical for MM clinical diagnosis. We used routine laboratory tests to establish a simple, inexpensive, and non-invasive diagnostic model for MM based on logistic regression. In the retrospective analysis, a total of 273 newly diagnosed MM inpatients and 288 non-MM participants, from January 2016 to December 2018 in Beijing Chaoyang hospital, Capital Medical University, were divided into training set and validation set. Age, gender, and the related routine laboratory tests for MM, including albumin (ALB), globulin (GLB), lactate dehydrogenase (LDH), creatinine (Cr), calcium (Ca2+), hemoglobin (Hb) and platelet (PLT), were analyzed by multivariate logistic regression to develop a diagnostic model. A diagnostic model was calculated using the formula MM index=-((-18×gender-3×ALB-Hb)/10), based on the logistic regression. The MM index [22 (20 - 25)] of MM patients was significantly lower than that of non-MM [30 (29 - 31)] in the training set (p < 0.001). It showed an excellent diagnostic performance in diagnosing MM through a receiver operating characteristic (ROC) curve, and its corresponding sensitivity, specificity, and area under the curve (AUC) were 95.6%, 96.7%, and 0.982 (0.968, 0.997), respectively. At a diagnostic risk threshold of 28, the model identified MM with a sensitivity of 95.6% and a specificity of 98.1% by using independent validation data. There was a significant positive correlation (r = 0.845, p < 0.001) between the DS grading and the MM index among all the participants. The established diagnostic model of MM index can successfully identify newly diagnosed MM from healthy controls. The diagnostic model of MM index may also act as a predictor of the severity of MM without therapy.

Comprehensive analysis of gene expression and DNA methylation data identifies potential biomarkers and functional epigenetic modules for lung adenocarcinoma.

Lung cancer has one of the highest mortality rates of malignant neoplasms. Lung adenocarcinoma (LUAD) is one of the most common types of lung cancer. DNA methylation is more stable than gene expression and could be used as a biomarker for early tumor diagnosis. This study is aimed to screen potential DNA methylation signatures to facilitate the diagnosis and prognosis of LUAD and integrate gene expression and DNA methylation data of LUAD to identify functional epigenetic modules. We systematically integrated gene expression and DNA methylation data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), bioinformatic models and algorithms were implemented to identify signatures and functional modules for LUAD. Three promising diagnostic and five potential prognostic signatures for LUAD were screened by rigorous filtration, and our tumor-normal classifier and prognostic model were validated in two separate data sets. Additionally, we identified functional epigenetic modules in the TCGA LUAD dataset and GEO independent validation data set. Interestingly, the MUC1 module was identified in both datasets. The potential biomarkers for the diagnosis and prognosis of LUAD are expected to be further verified in clinical practice to aid in the diagnosis and treatment of LUAD.

Detection of Gait From Continuous Inertial Sensor Data Using Harmonic Frequencies.

Mobile gait analysis using wearable inertial measurement units (IMUs) provides valuable insights for the assessment of movement impairments in different neurological and musculoskeletal diseases, for example Parkinson's disease (PD). The increase in data volume due to arising long-term monitoring requires valid, robust and efficient analysis pipelines. In many studies an upstream detection of gait is therefore applied. However, current methods do not provide a robust way to successfully reject non-gait signals. Therefore, we developed a novel algorithm for the detection of gait from continuous inertial data of sensors worn at the feet. The algorithm is focused not only on a high sensitivity but also a high specificity for gait. Sliding windows of IMU signals recorded from the feet of PD patients were processed in the frequency domain. Gait was detected if the frequency spectrum contained specific patterns of harmonic frequencies. The approach was trained and evaluated on 150 clinical measurements containing standardized gait and cyclic movement tests. The detection reached a sensitivity of 0.98 and a specificity of 0.96 for the best sensor configuration (angular rate around the medio-lateral axis). On an independent validation data set including 203 unsupervised, semi-standardized gait tests, the algorithm achieved a sensitivity of 0.97. Our algorithm for the detection of gait from continuous IMU signals works reliably and showed promising results for the application in the context of free-living and non-standardized monitoring scenarios.

Urothelial Carcinoma Detection Based on Copy Number Profiles of Urinary Cell-Free DNA by Shallow Whole-Genome Sequencing.

Current noninvasive assays for urothelial carcinoma (UC) lack clinical sensitivity and specificity. Given the utility of plasma cell-free DNA (cfDNA) biomarkers, the development of urinary cfDNA biomarkers may improve the diagnostic sensitivity. We assessed copy number alterations (CNAs) by shallow genome-wide sequencing of urinary cfDNA in 95 cancer-free individuals and 65 patients with UC, 58 with kidney cancer, and 45 with prostate cancer. We used a support vector machine to develop a diagnostic classifier based on CNA profiles to detect UC (UCdetector). The model was further validated in an independent cohort (52 patients). Genome sequencing data of tumor specimens from 90 upper tract urothelial cancers (UTUCs) and CNA data for 410 urothelial carcinomas of bladder (UCBs) from The Cancer Genome Atlas were used to validate the classifier. Genome sequencing data for urine sediment from 32 patients with UC were compared with cfDNA. To monitor the treatment efficacy, we collected cfDNA from 7 posttreatment patients. Urinary cfDNA was a more sensitive alternative to urinary sediment. The UCdetector could detect UC at a median clinical sensitivity of 86.5% and specificity of 94.7%. UCdetector performed well in an independent validation data set. Notably, the CNA features selected by UCdetector were specific markers for both UTUC and UCB. Moreover, CNA changes in cfDNA were consistent with the treatment effects. Meanwhile, the same strategy could localize genitourinary cancers to tissue of origin in 70.1% of patients. Our findings underscore the potential utility of urinary cfDNA CNA profiles as a basis for noninvasive UC detection and surveillance.

Fine-Resolution Mapping of Soil Total Nitrogen across China Based on Weighted Model Averaging

Accurate estimates of the spatial distribution of total nitrogen (TN) in soil are fundamental for soil quality assessment, decision making in land management, and global nitrogen cycle modeling. In China, current maps are limited to individual regions or are of coarse resolution. In this study, we compiled a new 90-m resolution map of soil TN in China by the weighted summation of random forest and extreme gradient boosting. After harmonizing soil data from 4022 soil profiles into a fixed soil depth (0–20 cm) by equal area spline, 18 environmental covariates were employed to characterize the spatial pattern of soil TN in topsoil across China. The accuracy assessments from independent validation data showed that the weighted model averaging gave the best predictions with an acceptable R2 (0.41). The prediction map showed that high-value areas of soil TN were mainly distributed in the eastern Tibetan Plateau, central Qilian Mountains and the north of the Greater Khingan Range. Climate factors had a considerable influence on the variation of the soil TN, and land-use types played a pivotal part in each climate zone. This high-resolution and high-quality soil TN data set in China can be very useful for future inventories of soil nitrogen, assessments of soil nutrient status, and management of arable land.

Agricultural cropland extent and areas of South Asia derived using Landsat satellite 30-m time-series big-data using random forest machine learning algorithms on the Google Earth Engine cloud

ABSTRACT The South Asia (India, Pakistan, Bangladesh, Nepal, Sri Lanka and Bhutan) has a staggering 900 million people (~43% of the population) who face food insecurity or severe food insecurity as per United Nations, Food and Agriculture Organization’s (FAO) the Food Insecurity Experience Scale (FIES). The existing coarse-resolution (≥250-m) cropland maps lack precision in geo-location of individual farms and have low map accuracies. This also results in uncertainties in cropland areas calculated from such products. Thereby, the overarching goal of this study was to develop a high spatial resolution (30-m or better) baseline cropland extent product of South Asia for the year 2015 using Landsat satellite time-series big-data and machine learning algorithms (MLAs) on the Google Earth Engine (GEE) cloud computing platform. To eliminate the impact of clouds, 10 time-composited Landsat bands (blue, green, red, NIR, SWIR1, SWIR2, Thermal, EVI, NDVI, NDWI) were derived for each of the three time-periods over 12 months (monsoon: Days of the Year (DOY) 151–300; winter: DOY 301–365 plus 1–60; and summer: DOY 61–150), taking the every 8-day data from Landsat-8 and 7 for the years 2013–2015, for a total of 30-bands plus global digital elevation model (GDEM) derived slope band. This 31-band mega-file big data-cube was composed for each of the five agro-ecological zones (AEZ’s) of South Asia and formed a baseline data for image classification and analysis. Knowledge-base for the Random Forest (RF) MLAs were developed using spatially well spread-out reference training data (N = 2179) in five AEZs. The classification was performed on GEE for each of the five AEZs using well-established knowledge-base and RF MLAs on the cloud. Map accuracies were measured using independent validation data (N = 1185). The survey showed that the South Asia cropland product had a producer’s accuracy of 89.9% (errors of omissions of 10.1%), user’s accuracy of 95.3% (errors of commission of 4.7%) and an overall accuracy of 88.7%. The National and sub-national (districts) areas computed from this cropland extent product explained 80-96% variability when compared with the National statistics of the South Asian Countries. The full-resolution imagery can be viewed at full-resolution, by zooming-in to any location in South Asia or the world, at www.croplands.org and the cropland products of South Asia downloaded from The Land Processes Distributed Active Archive Center (LP DAAC) of National Aeronautics and Space Administration (NASA) and the United States Geological Survey (USGS): https://lpdaac.usgs.gov/products/gfsad30saafgircev001/.

Assessing the impact of an invasive bryophyte on plant species richness using high resolution imaging spectroscopy

Abstract Remote sensing is commonly used to map the presence of invasive alien plant species across large areas. However, remote sensing also has potential to predict fractional cover, which is more closely linked to the potential ecosystem impact of invasive alien plant species. Here, we mapped the fractional cover of the invasive bryophyte Campylopus introflexus using high resolution imaging spectroscopy (233 bands, 490–2430 nm, 3 m × 3 m pixel size) in different habitat types within a coastal dune ecosystem. Maps were predicted from generalized partial least squares (gPLS) regression models trained with cover information from 266 field plots (3 m × 3 m). Models were calibrated using different data subsets, following a spatially blocked subsampling design. In addition, the relationship between C. introflexus fractional cover and plant species richness was evaluated using a subset of plot data. Fractional cover estimates from gPLS models resulted in R2 values of 0.64 ± 0.17 and an RMSE of 0.14 ± 0.02 based on independent validation data. Field observations showed a negative relationship between C. introflexus fractional cover and local plant species richness for pixels with a fractional cover above 44%. This threshold was exceeded in 4.3–7.1% of the studied area depending on the habitat type. Our study demonstrates that remote sensing can be used to map the fractional cover of an invasive bryophyte in a coastal dune ecosystem. Combining cover maps with a threshold value derived from an abundance-impact relationship we were able to highlight areas with potential negative impact on local plant species richness. This method can be implemented on other species and other ecosystem properties affected by plant invasions, allowing to highlight potential invasion impacts in a spatially explicit manner.

Predicting forest understory habitat for Canada lynx using LIDAR data

ABSTRACTCanada lynx (Lynx canadensis) is a federally threatened species in the contiguous United States. Within National Forests covered by the Northern Rockies Lynx Management Direction, Federal land managers must consider the effect of management activities on Canada lynx habitat. A common method to assess Canada lynx habitat used by the U.S. Forest Service is to measure horizontal cover using a cover board. We used field measurements and airborne Light Detection and Ranging (LIDAR) metrics to test beta regression models that predict estimates of horizontal cover on the Nez Perce–Clearwater National Forest, Idaho, USA, 2009–2015. We also investigated the effect on model predictions when the cover board was blocked by the main stem of a tree. Model fit statistics for normalized root mean square errors (RMSE%) were 30.8–33.7% and pseudo‐R2 ranged from 0.64 to 0.71. Using independent validation data, model performance statistics for RMSE% were 24.6–33.5% and R2 ranged from 0.51 to 0.69. We found that removing cover board measurements where the main stem of a tree blocked &gt;75% of the cover board produced the best model statistics. These models can be applied across LIDAR extents resulting in maps of horizontal cover estimates, which may be used in assessing effects of management activities on Canada lynx habitat. © 2019 The Wildlife Society.

Comparison of Fixed- and Mixed-Effects Approaches to Taper Modeling for Scots Pine in West Poland

Diameter measurements along the stem, which are the basis for taper models, usually have a hierarchical structure. Mixed-effects models, where fixed and random effects are distinguished, are a possible solution for this type of data. However, in order to fully absorb the potential of this method, random effects prediction, which requires additional measurements (diameter along stem), is recommended. This article presents a comparison of various fitting methods (mixed- and fixed-effects model approaches) of the variable-exponent taper model created by Kozak for determining the outside bark diameter along the stem and predicting the tree volume of Scots pine trees in west Poland. During the analysis, it was assumed that no additional measured data were available for practical use; therefore, for the mixed-effects model approach, fixed effects prediction without random effects was applied. Both fitting strategies were compared based on modeling and an independent validation data set. The comparison of mixed- and fixed-effects fitting strategies for the diameter along the stem indicated that the taper model fitted using the mixed-effects model approach better fit the data. Moreover, the error rate for the total tree volume prediction for the independent data set was lower for the mixed-effects model solution than for the fixed-effects one.

The Influence of Temperature and Community Structure on Light Absorption by Phytoplankton in the North Atlantic.

We present a model that estimates the spectral phytoplankton absorption coefficient () of four phytoplankton groups (picophytoplankton, nanophytoplankton, dinoflagellates, and diatoms) as a function of the total chlorophyll-a concentration (C) and sea surface temperature (SST). Concurrent data on (at 12 visible wavelengths), C and SST, from the surface layer (<20 m depth) of the North Atlantic Ocean, were partitioned into training and independent validation data, the validation data being matched with satellite ocean-colour observations. Model parameters (the chlorophyll-specific phytoplankton absorption coefficients of the four groups) were tuned using the training data and found to compare favourably (in magnitude and shape) with results of earlier studies. Using the independent validation data, the new model was found to retrieve total with a similar performance to two earlier models, using either in situ or satellite data as input. Although more complex, the new model has the advantage of being able to determine for four phytoplankton groups and of incorporating the influence of SST on the composition of the four groups. We integrate the new four-population absorption model into a simple model of ocean colour, to illustrate the influence of changes in SST on phytoplankton community structure, and consequently, the blue-to-green ratio of remote-sensing reflectance. We also present a method of propagating error through the model and illustrate the technique by mapping errors in group-specific using a satellite image. We envisage the model will be useful for ecosystem model validation and assimilation exercises and for investigating the influence of temperature change on ocean colour.

Three Novel Players: PTK2B, SYK, and TNFRSF21 Were Identified to Be Involved in the Regulation of Bovine Mastitis Susceptibility via GWAS and Post-transcriptional Analysis.

Bovine mastitis is a common inflammatory disease caused by multiple factors in early lactation or dry period. Genome wide association studies (GWAS) can provide a convenient and effective strategy for understanding the biological basis of mastitis and better prevention. 2b-RADseq is a high-throughput sequencing technique that offers a powerful method for genome-wide genetic marker development and genotyping. In this study, single nucleotide polymorphisms (SNPs) of the immune-regulated gene correlative with mastitis were screened and identified by two stage association analysis via GWAS-2b-RADseq in Chinese Holstein cows. We have screened 10,058 high quality SNPs from 7,957,920 tags and calculated their allele frequencies. Twenty-seven significant SNPs were co-labeled in two GWAS analysis models [Bayesian (P < 0.001) and Logistic regression (P < 0.01)], and only three SNPs (rs75762330, C > T, PIC = 0.2999; rs88640083, A > G, PIC = 0.1676; rs20438858, G > A, PIC = 0.3366) were annotated to immune-regulated genes (PTK2B, SYK, and TNFRSF21). Identified three SNPs are located in non-coding regions with low or moderate genetic polymorphisms. However, independent sample population validation (Case-control study) data showed that three important SNPs (rs75762330, P < 0.025, OR > 1; rs88640083, P < 0.005, OR > 1; rs20438858, P < 0.001, OR < 1) were significantly associated with clinical mastitis trait. Importantly, PTK2B and SYK expression was down-regulated in both peripheral blood leukocytes (PBLs) of clinical mastitis cows and in vitro LPS (E. coli)–stimulated bovine mammary epithelial cells, while TNFRSF21 was up-regulated. Under the same conditions, expression of Toll-like receptor 4 (TLR4), AKT1, and pro-inflammatory factors (IL-1β and IL-8) were also up-regulated. Interestingly, network analysis indicated that PTK2B and SYK are co-expressed in innate immune signaling pathway of Chinese Holstein. Taken together, these results provided strong evidence for the study of SNPs in bovine mastitis, and revealed the role of SYK, PTK2B, and TNFRSF21 in bovine mastitis susceptibility/tolerance.

Remote Sensing to Detect Nests of the Leaf-Cutting Ant Atta sexdens (Hymenoptera: Formicidae) in Teak Plantations

Leaf-cutting ants of the genus Atta are an important insect pest in forest plantations in many countries of South America. The objective of this work was to evaluate the potential for using Landsat-8 images, with medium spatial resolution and distributed free of charge, to detect leaf-cutting ant nests in Tectona grandis plantations in Brazil, using partial least squares discriminant analysis (PLS-DA). The regression model adjusted by PLS-DA selected three principal components with a cross-validation error of 0.275 to map and predict the presence of leaf-cutting ant nests in these plantations. The most important bands and vegetation indices were selected using the main variables in the projection (VIP) and predicted pixels with the presence or absence of leaf-cutting ant nests with an accuracy of 72.3% on an independent validation data set. The study indicates that Landsat-8 OLI images have the potential to detect and map leaf-cutting ant nests in T. grandis plantations.

A hybrid random forest to predict soccer matches in international tournaments

Abstract In this work, we propose a new hybrid modeling approach for the scores of international soccer matches which combines random forests with Poisson ranking methods. While the random forest is based on the competing teams’ covariate information, the latter method estimates ability parameters on historical match data that adequately reflect the current strength of the teams. We compare the new hybrid random forest model to its separate building blocks as well as to conventional Poisson regression models with regard to their predictive performance on all matches from the four FIFA World Cups 2002–2014. It turns out that by combining the random forest with the team ability parameters from the ranking methods as an additional covariate the predictive power can be improved substantially. Finally, the hybrid random forest is used (in advance of the tournament) to predict the FIFA World Cup 2018. To complete our analysis on the previous World Cup data, the corresponding 64 matches serve as an independent validation data set and we are able to confirm the compelling predictive potential of the hybrid random forest which clearly outperforms all other methods including the betting odds.

Enhancing VGI application semantics by accounting for spatial bias

ABSTRACTVolunteered geographic information (VGI) is becoming an important source of geospatial big data that support many applications. The application semantics of VGI, i.e. how well VGI reflects the real-world geographic phenomena of interest to the application, is essential for any VGI applications. VGI observations often are spatially biased (e.g. spatially clustered). Spatial bias poses challenges on VGI application semantics because it may impede the quality of inferences made from VGI. Using species distribution modeling (SDM) as an example application, this article argues that spatial bias impedes VGI application semantics, as gauged by SDM model performance, and accounting for bias enhances application semantics. VGI observations from eBird were used in a case study for modeling the distribution of the American Robin (Turdus migratorius) in U.S. T. migratorius observations from the North American Breeding Bird Survey were used as independent validation data for model performance evaluation. A grid-based strategy was adopted to filter eBird species observations to reduce spatial bias. Evaluations show that spatial bias in species observations degrades SDM model performance and filtering species observations improves model performance. This study demonstrates that VGI application semantics can be enhanced by accounting for the spatial bias in VGI observations.