Number Of Available Data Points Research Articles

Fractional Flow Reserve-Based Patient Risk Classification.

Cardiovascular diseases (CVDs) are a leading cause of death. If not treated in a timely manner, cardiovascular diseases can cause a plethora of major life complications that can include disability and a loss of the ability to work. Globally, acute myocardial infarction (AMI) is responsible for about 3 million deaths a year. The development of strategies for prevention, but also the early detection of cardiovascular risks, is of great importance. The fractional flow reserve (FFR) is a measurement used for an assessment of the severity of coronary artery stenosis. The goal of this research was to develop a technique that can be used for patient fractional flow reserve evaluation, as well as for the assessment of the risk of death via gathered demographic and clinical data. A classification ensemble model was built using the random forest machine learning algorithm for the purposes of risk prediction. Referent patient classes were identified by the observed fractional flow reserve value, where patients with an FFR higher than 0.8 were viewed as low risk, while those with an FFR lower than 0.8 were identified as high risk. The final classification ensemble achieved a 76.21% value of estimated prediction accuracy, thus achieving a mean prediction accuracy of 74.1%, 77.3%, 78.1% and 83.6% over the models tested with 5%, 10%, 15% and 20% of the test samples, respectively. Along with the machine learning approach, a numerical approach was implemented through a 3D reconstruction of the coronary arteries for the purposes of stenosis monitoring. Even with a small number of available data points, the proposed methodology achieved satisfying results. However, these results can be improved in the future through the introduction of additional data, which will, in turn, allow for the utilization of different machine learning algorithms.

Systematic Handling of Environmental Fate Data for Model Development-Illustrated for the Case of Biodegradation Half-Life Data.

The assessment of environmental hazard indicators such as persistence, mobility, toxicity, or bioaccumulation of chemicals often results in highly variable experimental outcomes. Persistence is particularly affected due to a multitude of influencing environmental factors, with biodegradation experiments resulting in half-lives spanning several orders of magnitude. Also, half-lives may lie beyond the limits of reliable half-life quantification, and the number of available data points per substance may vary considerably, requiring a statistically robust approach for the characterization of data. Here, we apply Bayesian inference to address these challenges and characterize the distributions of reported soil half-lives. Our model estimates the mean, standard deviation, and corresponding uncertainties from a set of reported half-lives experimentally obtained for a single substance. We apply our inference model to 893 pesticides and pesticide transformation products with experimental soil half-lives of varying data quantity and quality, and we infer the half-life distribution for each compound. By estimating average half-lives, their experimental variability, and the uncertainty of the estimations, we provide a reliable data source for building predictive models, which are urgently needed by regulatory authorities to manage existing chemicals and by industry to design benign, nonpersistent chemicals. Our approach can be readily adapted for other environmental hazard indicators.

Rigorous Mapping of Data to Qualitative Properties of Parameter Values and Dynamics: A Case Study on a Two-Variable Lotka-Volterra System.

In this work, we describe mostly analytical work related to a novel approach to parameter identification for a two-variable Lotka-Volterra (LV) system. Specifically, this approach is qualitative, in that we aim not to determine precise values of model parameters but rather to establish relationships among these parameter values and properties of the trajectories that they generate, based on a small number of available data points. In this vein, we prove a variety of results about the existence, uniqueness, and signs of model parameters for which the trajectory of the system passes exactly through a set of three given data points, representing the smallest possible data set needed for identification of model parameter values. We find that in most situations such a data set determines these values uniquely; we also thoroughly investigate the alternative cases, which result in nonuniqueness or even nonexistence of model parameter values that fit the data. In addition to results about identifiability, our analysis provides information about the long-term dynamics of solutions of the LV system directly from the data without the necessity of estimating specific parameter values.

Coupling covariance matrix adaptation with continuum modeling for determination of kinetic parameters associated with electrochemical CO2 reduction

Coupling covariance matrix adaptation with continuum modeling for determination of kinetic parameters associated with electrochemical CO2 reduction

A Markov Chain Monte Carlo approach for Pseudo-Input selection in Sparse Gaussian Processes

The effectiveness of non-parametric methods for regression comes at the price of high computational complexity. In fact, these methods scale as O(N3), where N is the number of available data points. One possible option to address this issue consists in introducing a set of fictitious (pseudo-) inputs of size M ≪ N such that the computational effort is reduced to O(M2N). To estimate hyper-parameters and pseudo-inputs, a non-convex optimization problem needs to be solved. As opposed to the conventional gradient-based approach used in the literature, this paper proposes a stochastic scheme leveraging Markov Chain Monte Carlo methods. Numerical comparisons show that the latter returns a more efficient set of pseudo-inputs, leading to a superior performance in terms of mean squared error.

LOS/NLOS Estimators for mmWave Cellular Systems With Blockages

Designers of millimeter wave (mmWave) cellular systems need to evaluate line-of-sight (LOS) maps to provide good service to users in urban scenarios. In this letter, we derive estimators to obtain LOS maps in scenarios with potential blocking elements. Applying previous stochastic geometry results, we formulate the optimal Bayesian estimator of the LOS map using a limited number of actual measurements at different locations. The computational cost of the optimal estimator is derived and is proven to be exponential in the number of available data points. An approximation is discussed, which brings the computational complexity from exponential to quasilinear and allows the implementation of a practical estimator. Finally, we compare numerically the optimal estimator and the approximation with other estimators from the literature and also with an original heuristic estimator with good performance and low computational cost. For the comparison, both synthetic layouts and a real layout of Chicago have been used.

Time Correction of Ocean-Bottom Seismometers Using Improved Ambient Noise Cross Correlation of Multicomponents and Dual-Frequency Bands

Abstract An effective approach was developed for identifying and correcting ocean-bottom seismometer (OBS) time errors through improving ambient noise cross-correlation function (NCCF) analysis and combination with other methods. Significant improvements were illustrated through analyzing data from a passive-source seismic experiment in the southwestern sub-basin of the South China Sea. A novel method was first developed that can effectively identify errors in the sampling frequency of the OBS instruments. The traditional NCCF method was then expanded by increasing the analyzed data spectrum from a single-frequency band to dual-frequency band pairs, thus doubling the number of available data points and substantially improving the time correction quality. For data with relatively low signal-to-noise ratios, the average time errors were reduced from the original average values of 60–80 ms by the conventional methods to <40 ms using the improved approaches. The new multistep procedure developed in this study has general applicability to analysis of other OBS experiments. The demonstrated significant improvements in the data quality are critical for advancing seismic tomography and other modern marine geophysical studies that require high accuracy in the OBS data.

Trend estimation and layer boundary detection in depth-dependent soil data using sparse Bayesian lasso

This paper proposes a method for estimating trends and detecting layer boundaries in depth-dependent soil data based on a least absolute shrinkage statistical operator (lasso). Although the lasso appears to be promising for subsurface modeling because no predetermined basis functions or stratification models are required, it does not provide information on the uncertainty of its estimated solution, i.e., a point estimate. In subsurface modeling, however, characterization of uncertainty is pivotal because soil data can be (spatially) sparse and noisy. A lasso-based method that can quantify its estimation accuracy while preserving its attractive sparsity feature is proposed. The performance of this sparse Bayesian lasso (SBLasso) is demonstrated through numerical tests and an actual case study of its accuracy of trend estimation and layer boundary detection. The degree of accuracy or inaccuracy of estimation provided by the SBLasso clearly corresponds to data quality, such as the number of available data points, noise level, and noise correlation. A method of soil stratification based on SBLasso was also proposed, and the stratification results by SBLasso were compared with those produced by existing methods for validation.

Smooth operator: Modifying the Anhøj rules to improve runs analysis in statistical process control.

The run chart is one form of statistical process control chart that is particularly useful for detecting persistent shifts in data over time. The Anhøj rules test for shifts by looking for unusually long runs (L) of data points on the same side of the process centre (mean or median) and unusually few crossings (C) of the centre depending on the number of available data points (N). Critical values for C and L have mainly been studied in isolation. But what is really of interest is the joint distribution of C and L, which has so far only been studied using simulated data series. We recently released an R package, crossrun that calculates exact values for the joint probabilities of C and L that allowed us to study the diagnostic properties of the Anhøj rules in detail and to suggest minor adjustments to improve their diagnostic value. Based on the crossrun R package we calculated exact values for the joint distribution of C and L for N = 10-100. Furthermore, we developed two functions, bestbox() and cutbox() that automatically seek to adjust the critical values for C and L to balance between sensitivity and specificity requirements. Based on exact values for the joint distribution of C and L for N = 10-100 we present measures of the diagnostic value of the Anhøj rules. The best box and cut box procedures improved the diagnostic value of the Anhøj rules by keeping the specificity and sensitivity close to pre-specified target values. Based on exact values for the joint distribution of longest run and number of crossings in random data series this study demonstrates that it is possible to obtain better diagnostic properties of run charts by making minor adjustment to the critical values for C and L.

Extraction of neutrino mixing parameters from experiments with multiple identical detectors

The statistical method used in the analyses of measurements of neutrino oscillation mixing angle [Formula: see text] by the Daya Bay Collaboration[Formula: see text] and RENO Collaboration3 is based on variational minimization of a [Formula: see text] function defined in terms of quantities of interest and pull factors which are introduced to deal with effects of systematic uncertainties. For both experiments, the number of parameters that need to be determined is greater than the number of available data points (20 versus 6 for the Daya Bay and 12 versus 2 for the RENO). While the results for the mixing angle and the normalization factor were reported, results for the other parameters (pull factors) were omitted in their publications.[Formula: see text] There exist multiple sets of parameters from the minimization of the [Formula: see text] function. We investigate the sensitivity of the extracted mixing angle on this nonuniqueness of minimization results for the Daya Bay data using two methods of minimization. We report results for all parameters, including those of physics interest and pull factors. The obtained results for the mixing angle and the normalization factor are in agreement with those reported by the Daya Bay Collaboration.

Parametric Study of Amorphous High-Entropy Alloys formation from two New Perspectives: Atomic Radius Modification and Crystalline Structure of Alloying Elements

Chemical and topological parameters have been widely used for predicting the phase selection in high-entropy alloys (HEAs). Nevertheless, previous studies could be faulted due to the small number of available data points, the negligence of kinetic effects, and the insensitivity to small compositional changes. Here in this work, 92 TiZrHfM, TiZrHfMM, TiZrHfMMM (M = Fe, Cr, V, Nb, Al, Ag, Cu, Ni) HEAs were prepared by melt spinning, to build a reliable and sufficiently large material database to inspect the robustness of previously established parameters. Modification of atomic radii by considering the change of local electronic environment in alloys, was critically found out to be superior in distinguishing the formation of amorphous and crystalline alloys, when compared to using atomic radii of pure elements in topological parameters. Moreover, crystal structures of alloying element were found to play an important role in the amorphous phase formation, which was then attributed to how alloying hexagonal-close-packed elements and face-centered-cubic or body-centered-cubic elements can affect the mixing enthalpy. Findings from this work not only provide parametric studies for HEAs with new and important perspectives, but also reveal possibly a hidden connection among some important concepts in various fields.

EMPIRICAL ESTIMATION OF NON-LINEAR INPUT–OUTPUT MODELS: AN ENTROPY ECONOMETRICS APPROACH

Despite theoretical advances, non-linear input–output models have been empirically applied only to a limited extent. This is mainly due to the fact that the number of parameters to be estimated is much higher than the number of available data points. Taking advantage of the recent proliferation of input–output databases and by applying an estimation strategy that relies on entropy econometrics, this paper suggests a way to estimate the parameters that characterize non-linear relationships between inputs and output. This non-linear modelling allows for considering time-specific input coefficients, instead of fixed ones. Several types of multipliers can be derived from this non-linear model, and the proposed generalized maximum entropy (GME) estimator allows estimating them from time series or cross-sectional datasets of input–output tables. The proposed GME technique is illustrated by means of an empirical application that estimates the parameters that characterize a non-linear input–output model for the Spanish economy over the period 1995–2011.

Root Cause Analysis by a Combined Sparse Classification and Monte Carlo Approach

Classification methods with embedded feature selection capability are very appealing for the analysis of complex processes since they allow the analysis of root causes even when the number of input variables is high. In this work, we investigate the performance of three techniques for classification within a Monte Carlo strategy with the aim of root cause analysis. We consider the naïve bayes classifier and the logistic regression model with two different implementations for controlling model complexity, namely, a LASSO-like implementation with a ℓ1 norm regularization and a fully Bayesian implementation of the logistic model, the so called relevance vector machine. Several challenges can arise when estimating such models mainly linked to the characteristics of the data: a large number of input variables, high correlation among subsets of variables, the situation where the number of variables is higher than the number of available data points and the case of unbalanced datasets. Using an ecological and a semiconductor manufacturing dataset, we show advantages and drawbacks of each method, highlighting the superior performance in term of classification accuracy for the relevance vector machine with respect to the other classifiers. Moreover, we show how the combination of the proposed techniques and the Monte Carlo approach can be used to get more robust insights into the problem under analysis when faced with challenging modelling conditions.

Methods and numerical aspects of nanoscopic contact area estimation in atomistic tribological simulations

We show how data obtained from molecular dynamics (MD) simulations of nanoscale friction should be treated for producing constitutive system parameters with a proper error estimation. A visualisation scheme for discrete atomistic geometries based on the smooth particle method (SPM) was parametrised and validated to yield an accurate and computationally robust estimation of the contact area between two touching nanoscopic asperities. We present some thoughts on the error estimation of the contact forces occurring due to the load and the shearing motion. The variance in the friction force constitutes the main source of error for the fitting of the constitutive system parameters. The dependence of the constitutive system parameters on the number of available data points was also studied. It was shown that an equal spacing (by load) of the data points can result in better values for the system parameters than the convergence trend suggests.

Automated alternation flicker for the detection of optic disc haemorrhages

Optic disc haemorrhages are associated with active glaucomatous neurodegeneration and ongoing visual field loss. We sought to determine whether automated alternation flicker enhances the detection of disc haemorrhages in serial images from patients with glaucoma when compared to side-by-side photographic evaluation and single-image display. Serial sets of optic nerve photographs of 394 eyes from 234 patients followed for glaucoma at the authors' institutions were included in this study. Eyes with disc haemorrhages were graded for difficulty level and randomized along with nondisc haemorrhage control images into one of three presentation groups (automated alternation flicker, side-by-side or single image). Seven graders viewed all images and assessed for the presence or absence of disc haemorrhages. The sensitivity of automated alternation flicker for disc haemorrhage detection (0.878) was higher than side-by-side (0.705; p = 0.002) and single photographs (0.757; p = 0.01). There was no specificity difference between pairs of presentation groups (all p ≥ 0.7). Automated alternation flicker was a more sensitive method for disc haemorrhage detection than the current clinical standards and may have an important role in the management of glaucoma.

Estimating carbon stocks at a regional level using soil information and easily accessible auxiliary variables

One of the most important challenges of digital soil mapping is the development of methods that allow the characterisation of large areas with a high-resolution. Surface soils, forming the largest pool of terrestrial organic carbon, may be able to sequester atmospheric carbon and thus mitigate climate change. However, this remains controversial, largely due to insufficient information on SOC stocks worldwide. One reason for this is the generally limited number of available data points, especially when large areas are considered, while another reason lies on the accuracy of interpolation techniques used for SOC mapping. The study was performed in Laos, a 230,566 km 2 area mostly forested and with steep slopes, and where soil data from 2806 pits is available. Our objective was to estimate SOC stocks to a depth of 1 m over the whole country while improving regional digital soil mapping (RDSM). SOC mapping by using purely spatial approaches of ordinary kriging (OK), inverse distance weighting (IDW) and regularized spline with tension (RST) was compared with the use of additional information on relief, climate and soils through co-kriging (OCK). The generation and validation data sets were composed of 2665 and 141 data points respectively. Overall, OCK using a multiple correlation with elevation above sea level, compound topographic index, mean slope gradient, average annual rainfall, and soil clay content ( R 2 = 0.42; P level < 0.001) as covariate, yielded the most accurate predictions (19.7 kg C m − 2 with standard error of ± 3.2 kg C m − 2 ; and 4.54 ± 0.74 billion tons of SOC for Laos). The pure interpolation techniques were less accurate with 4.51 ± 1.02 billion tons of SOC for OK and 4.88 ± 0.94 billion tons of SOC for RST. Besides providing nationwide estimates of SOC stocks these results indicate that using collectively soil punctual information on SOC stocks and their inter-relationships with controlling factors which are easy to gather might be an efficient way to improve RDSM.

Modelling intercrop management impact on runoff and erosion in a continuous maize cropping system: Part II. Model Pareto multi‐objective calibration and long‐term scenario analysis using disaggregated rainfall

SummaryIn order to evaluate the impact of intercrop management on runoff and erosion in a continuous maize cropping system, the plot scale, continuous and process‐based erosion model (CREHDYS) developed previously must be calibrated accounting for its two main outputs: runoff and sediment loss. To do that, a global Pareto multi‐objective calibration was applied to these two potentially conflicting objectives, considering daily runoff and periodical erosion rates, for two sites with different slopes and soil textures. This revealed a trade‐off between both objectives. The large resulting Pareto uncertainty regarding parameters did not translate into a large predictive uncertainty of daily runoff but resulted into a large uncertainty on erosion prediction. Globally, model results were satisfactory with regard to daily runoff prediction (Nash‐Sutcliffe index varying within the Pareto solution set from 0.65 to 0.91 for calibration and 0.64 to 0.77 for validation period) and relatively satisfactory for periodical erosion. However, the small number of available data points (three) for model validation in terms of periodical erosion prediction was not sufficient to ensure a proper validation. The calibrated model was in turn used to perform a scenario analysis of the long‐term hydrological and erosive impact of inter‐cropping period management in a continuous maize cropping system, using disaggregated rainfall. The long‐term simulations mainly revealed that, with regard to the erosion prevention during the inter‐cropping period, planting a winter cover crop is a better option than reduced tillage with a cultivator (0–12 cm), even if the cover is destroyed early (1 January). As compared with the situation of a bare heavily crusted soil with two semi‐permanent wheel tracks, reduced tillage led to an erosion reduction from 90 to 97%, an early cover destruction (1 January) to an erosion decrease from 92 to 98% and a cover destroyed on 1 March or later to an average soil loss reduction from 96 to 99%.

Assessment of data availability influence on integrated urban drainage modelling uncertainty

In urban water quality management, several models are connected and integrated for analysing the fate of pollutants from the sources in the urban catchment to the final recipient; classical problems connected with the selection and calibration of parameters are amplified by the complexity of the modelling approach increasing their uncertainty. The present paper aims at studying the influence of reductions in available data on the modelling response uncertainty with respect to the different integrated modelling outputs (both considering quantity and quality variables). At this scope, a parsimonious integrated home-made model has been used allowing for analysing the combinative effect of data availability regarding the different parts of the integrated urban drainage system; the uncertainty analysis approach has been applied to an experimental catchment in Bologna (Italy). The number of available data points has been fictitiously reduced obtaining data sets ranging between 25% and 100% of the actually measured data. For each of the data sets, uncertainty has been evaluated and its propagation from the upstream sub-model to the downstream ones has been assessed. The present study demonstrates that model calibration and modelling efficiency assessment may induce the operator to be excessively confident in the model results when available data are scarce. Quite the opposite is indeed true, that limited data availability increases modelling uncertainty. A conclusion of this article is that uncertainty analysis should always be conducted in order to effectively evaluate model reliability.

Unexploded ordnance location and identification using borehole magnetometery

It is estimated that 10% of war-time bombs did not explode and can be found either lying on the ground or buried at depths of up to 20 m depending on the soil type. These unexploded bombs also known as ordnance pose a real danger to construction workers and properties. Ground-surface-based detection methods are unable to detect objects at such depths due to the rapidly diminishing nature of the anomalous field and interference from metal debris on the ground surface. A borehole magnetometer equipped with a drilling system can penetrate to the depths of interest and collect high-resolution field data. However, conventional inversion techniques are incapable of handling one-dimensional borehole data due to the insufficient number of available data points. To overcome these difficulties, a borehole detection method is proposed that has an inversion algorithm specifically developed to deal with the high non-uniqueness of the borehole data. It has the advantages of a fast convergence and maximisation of information extracted from individual borehole measurements. This method is more reliable than traditional methods since it examines all the possibilities by taking a number of estimations. The information from each individual borehole is fully interpreted, allowing suggestions to be made for the positioning of the next drilling point allowing the minimisation of the number of holes required to obtain a definative solution. This results in a reduction in both hazard levels and cost. A method is proposed based on the recovered information that allows the identification and discrimination of unexploded ordnance from other objects in the search environment, such as steel pipes and barrels. It is not sensitive to interference in the data once the moment signature is recovered. The results obtained from a test site are extremely promising and demonstrate the proposed techniques capability to handle real-world inversion problems despite having only a small number of available data points.

Application of the Random Forest Method in Studies of Local Lymph Node Assay Based Skin Sensitization Data.

The random forest and classification tree modeling methods are used to build predictive models of the skin sensitization activity of a chemical. A new two-stage backward elimination algorithm for descriptor selection in the random forest method is introduced. The predictive performance of the random forest model was maximized by tuning voting thresholds to reflect the unbalanced size of classification groups in available data. Our results show that random forest with a proposed backward elimination procedure outperforms a single classification tree and the standard random forest method in predicting Local Lymph Node Assay based skin sensitization activity. The proximity measure obtained from the random forest is a natural similarity measure that can be used for clustering of chemicals. Based on this measure, the clustering analysis partitioned the chemicals into several groups sharing similar molecular patterns. The improved random forest method demonstrates the potential for future QSAR studies based on a large number of descriptors or when the number of available data points is limited.