Statistical Prediction Research Articles

ПРОГНОЗУВАННЯ РОЗВИТКУ ЗАХВОРЮВАНЬ ТКАНИН ПАРОДОНТА

Introduction. Periodontal tissue disease is one of the most common dental pathologies, which among young people occurs with a frequency of 60% to 99%. Therefore, the problem of finding new links in the pathogenesis, the reasons for the growing prevalence of periodontal disease, as well as effective methods for its early diagnosis and prevention, is relevant. Objectives. Establish the possibility of using individual stomatological and psychophysiological features to predict the development of periodontal disease. Materials and methods. 156 students aged 18-23 years old without systemic diseases were surveyed for some features of oral hygiene and nutrition. Also the study subjects underwent a dental examination, psychological testing and the assessment of individual typological features of higher nervous activity and autonomous regulation. The model for statistical prediction were designed using neural networks. Results. Two neural networks were designed with the best predictors among dental history and examination, psychological testing, parameters of higher nervous activity and heart rate variability analysis. The diagnostic sensitivity of the first prognostic model was 83.33 % and the specificity was 92.31 %. The second model was characterized by 90.00 % sensitivity and 78.57 % specificity. Conclusion. The method of modeling using neural networks based on the index assessment of the condition of teeth hard tissues, the level of oral hygiene and the evaluation of psychophysiological features can effectively predict the risk of periodontal disease development in young people

Supersymmetric particle and Higgs boson masses from the landscape: Dynamical versus spontaneous supersymmetry breaking

Perturbative supersymmetry breaking on the landscape of string vacua is expected to favor large soft terms as a power-law or log distribution, but tempered by an anthropic veto of inappropriate vacua or vacua leading to too large a value for the derived weak scale -- a violation of the atomic principle. Indeed, scans of such vacua yield a statistical prediction for light Higgs boson mass m_h~ 125 GeV with sparticles (save possibly light higgsinos) typically beyond LHC reach. In contrast, models of dynamical SUSY breaking (DSB) -- with a hidden sector gauge coupling g^2 scanned uniformly -- lead to gaugino condensation and a uniform distribution of soft parameters on a log scale. Then soft terms are expected to be distributed as $m_{\rm soft}^{-1}$ favoring small values. A scan of DSB soft terms generally leads to $m_h\ll 125$ GeV and sparticle masses usually below LHC limits. Thus, the DSB landscape scenario seems excluded from LHC search results. An alternative is that the exponential suppression of the weak scale is set anthropically on the landscape via the atomic principle.

The Potential Value of Monocyte to Lymphocyte Ratio, Platelet to Mean Platelet Volume Ratio in the Diagnosis of Periprosthetic Joint Infections.

ObjectiveTo explore the possibility of obtaining more accurate information from routine blood tests for the diagnosis of periprosthetic joint infection (PJI).MethodsThis is a retrospective study. Between 2017 and 2018, a total of 246 patients who underwent total hip or knee revision surgery were included in this study. There were 146 females and 100 males, and the mean age of the patients was 62.1 ± 12.75 years. Laboratory parameters erythrocyte sedimentation rate (ESR), C‐reactive protein (CRP), D‐dimer, plasma fibrinogen, serum white blood cell (WBC), and calculable ratio markers were collected. Based on leukocytes (monocyte count, neutrophil count, lymphocyte count), platelet count, and mean platelet volume Inflammation‐related ratio markers were calculated, which including monocyte to lymphocyte ratio (MLR), neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and platelet to mean platelet volume ratio (PMR). Follow‐up of all studied cases for at least 1 year. The diagnostic value of the markers based on the receiver operating characteristic (ROC) analysis. The most optimal combinations of blood markers were selected by the prediction models. Statistical analyses and prediction models were performed using R software.ResultsOf the 246 patients, 125 were diagnosed with PJI and 121 with aseptic loosening. A higher rate of patients underwent revision surgery due to hip prosthesis loosening in the aseptic loosening group (74.4%) compared to the PJI group (45.6%, P < 0.001). ROC curves showed that the area under the curve (AUC) for classical markers, fibrinogen was 0.853 (95% confidence interval [CI], 0.805–0.901), ESR was 0.836 (95% CI, 0.785–0.887) and CRP was 0.825 (95% CI, 0.773–0.878). Followed by the PMR, PLR, NLR and MLR, which showed promising diagnostic performance with AUCs of 0.791, 0.785, 0.736, and 0.733. The AUCs of the ratio markers were higher than those of D‐dimer (0.691;95% CI, 0.6243–0.7584) and serum WBC (0.622; 95% CI, 0.552–0.691). After the predictive model calculation, AUC was up to 0.923 (95% CI, 0.891–0.951) when plasma fibrinogen combined with MLR and PMR and interpreted excellent discriminatory capacity with a sensitivity of 86.40% and a specificity of 84.17%. The new combination significantly increases the accuracy and reliability of the diagnosis of PJI (P < 0.001). The AUC increased to 0.899 (95% CI, 0.861–0.931; P = 0.007) and 0.916 (95% CI, 0.880–0.946; P < 0.001), followed by CRP and ESR, respectively. All plasma fibrinogen, ESR, and CRP combined with both PMR and MLR achieved the highest specificity (89.17%) and PPV (85.34%).ConclusionThe diagnostic performance greatly improved when plasma fibrinogen, ESR, and CRP combined with ratio markers.

A time series projection model of online seasonal demand for American wine and potential disruption in the supply channels due to COVID-19

PurposeThis study statistically examines the shifting distribution channels in the American wine industry based on the growth trajectory of sales, seasonality and disruption due to consumers switching to online platforms. The purpose of this paper is to design a model that will have general applicability beyond the wine industry.Design/methodology/approachThe research uses regression-based additive decomposition of time series data to predict the trajectory of the market share for the digital distribution channel. The study develops a statistical prediction model using time series data between 2007 and 2020, inclusive, sourced from US Annual Wine Reports and Bureau of Alcohol, Tobacco and Firearms databases.FindingsThe results show an increasing trajectory of wine sales through the online distribution channel with predictable seasonality. The disruptive effects of consumer switching behavior point to a steady increase in sales due both to increasing demand and accelerating switching. Nevertheless, the model shows that bricks and mortar purchases will remain strong and continue to account for the bulk of wine sales. COVID-19 has caused a step function increase in online sales but this should moderate after the crisis subsides and can be tested further.Originality/valueThis study is original in developing a model for an industry where bricks and mortar sales are growing and are expected to remain strong while there is still identifiable switching to online sales. The wine industry presents a classic case of accelerating switching behavior where there is still a strong franchise for in-store purchases. The model should have general applicability to distribution channels beyond the wine industry where steady growth, marked seasonality and disruptive consumer switching are in evidence.

Evolution of microstructure and hardness in shell moulded impeller casting

ABSTRACT To remain competitive in this growing field of the manufacturing industry, there is a need to reduce the lead times in the development of a component. The virtual simulation environment tool can be utilised for minimising production time as well as assure defect-free casting. This technique is based on heat transfer and volumetric calculations for prediction of possible defects. In this paper, analysis of automobile impeller casting is explored. The modelling of the casting and feeding system is done using the CATIA software. The CAD model of the impeller is imported into the simulation environment for solidification simulation. The simulation is performed using SolidCAST software which is based on the Finite Difference Method. The simulations are performed to ensure proper feeding for achieving sound casting. The melting trials are performed in jobbing foundry. The hardness and microstructure analysis of simulated and experimental casting is performed so as to correlate. The section thickness and cooling rates in casting were analysed and used for formulating hardness and microstructure in the experimental castings. The statistical predictions based on simulated and experimental results are presented for formulating Brinell hardness and amount of pearlite phase in particular section of shell moulded ductile iron casting.

Biomolecular Binding under Confinement: Statistical Predictions of Steric Influence in Absence of Long-Distance Interactions.

We propose a theoretical model for the influence of confinement on biomolecular binding at the single-molecule scale at equilibrium, based on the change of the number of microstates (localization and orientation) upon reaction. Three cases are discussed: DNA sequences shorter and longer than the single strain DNA Kuhn length and spherical proteins, confined into a spherical container (liposome, droplet, etc.). The influence of confinement is found to be highly dependent on the molecular structure and significant for large molecules (relative to container size).

Supervised Statistical Learning Prediction of Soybean Varieties and Cultivation Sites Using Rapid UPLC-MS Separation, Method Validation, and Targeted Metabolomic Analysis of 31 Phenolic Compounds in the Leaves.

Soybean (Glycine max; SB) leaf (SL) is an abundant non-conventional edible resource that possesses value-adding bioactive compounds. We predicted the attributes of SB based on the metabolomes of an SL using targeted metabolomics. The SB was planted in two cities, and SLs were regularly obtained from the SB plant. Nine flavonol glycosides were purified from SLs, and a validated simultaneous quantification method was used to establish rapid separation by ultrahigh-performance liquid chromatography-mass detection. Changes in 31 targeted compounds were monitored, and the compounds were discriminated by various supervised machine learning (ML) models. Isoflavones, quercetin derivatives, and flavonol derivatives were discriminators for cultivation days, varieties, and cultivation sites, respectively, using the combined criteria of supervised ML models. The neural model exhibited higher prediction power of the factors with high fitness and low misclassification rates while other models showed lower. We propose that a set of phytochemicals of SL is a useful predictor for discriminating characteristics of edible plants.

Development of an Algorithm for Prediction of the Wind Speed in Renewable Energy Environments

The aim of this work is to develop an algorithm that is able to provide predictions of wind speed statistics (WSS) in renewable energy environments. The subject is clearly interesting, as predictions of storms and extreme winds are important for decision makers and emergency response teams in renewable energy environments, e.g., in places where wind turbines could be located, including cities. The goal of the work is achieved through two phases: (a) During the preparation phase, the construction of a big WSS database based on computational fluid dynamics (CFD) is carried out, which includes flow fields of different wind directions in all grid numerical points; (b) In the second phase, the algorithm is used to find the records in the WSS database with the closest meteorological conditions to the meteorological conditions of interest. The evaluation of the CFD model (including both RANS and LES turbulence methodologies) is performed using the experimental data of the MUST (Mock Urban Setting Test) wind tunnel experiment.

Comparative analysis of machine learning approaches to analyze and predict the COVID-19 outbreak.

BackgroundForecasting the time of forthcoming pandemic reduces the impact of diseases by taking precautionary steps such as public health messaging and raising the consciousness of doctors. With the continuous and rapid increase in the cumulative incidence of COVID-19, statistical and outbreak prediction models including various machine learning (ML) models are being used by the research community to track and predict the trend of the epidemic, and also in developing appropriate strategies to combat and manage its spread.MethodsIn this paper, we present a comparative analysis of various ML approaches including Support Vector Machine, Random Forest, K-Nearest Neighbor and Artificial Neural Network in predicting the COVID-19 outbreak in the epidemiological domain. We first apply the autoregressive distributed lag (ARDL) method to identify and model the short and long-run relationships of the time-series COVID-19 datasets. That is, we determine the lags between a response variable and its respective explanatory time series variables as independent variables. Then, the resulting significant variables concerning their lags are used in the regression model selected by the ARDL for predicting and forecasting the trend of the epidemic.ResultsStatistical measures—Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Symmetric Mean Absolute Percentage Error (SMAPE)—are used for model accuracy. The values of MAPE for the best-selected models for confirmed, recovered and deaths cases are 0.003, 0.006 and 0.115, respectively, which falls under the category of highly accurate forecasts. In addition, we computed 15 days ahead forecast for the daily deaths, recovered, and confirm patients and the cases fluctuated across time in all aspects. Besides, the results reveal the advantages of ML algorithms for supporting the decision-making of evolving short-term policies.

LILITA_N21: Updated version of the Monte Carlo fusion–evaporation code

Since the code LILITA was presented in 1981 by Gomez Del Campo and Stockstad (1981)[1] and used to simulate the emission of light particles in the decay of a compound nucleus, relevant improvements have been made. Revisions leading to LILITA_N97 were begun by G. La Rana (Naples), M. Kaplan (CMU, Pittsburgh, PA), J.M. Alexander (Stony Brook, NY) and continued by the Naples Group (Moro et al., 2012) [2]. In this work new key features of the code consisting of the implementation of transmission coefficients based on global parametrization of the optical model potential, a description of the nuclear shape based on the nuclear stratosphere model, the parallelization of the code and a graphical user interface are presented. We expect that the new code version, LILITA_N21, makes it possible a better reproduction of wide data set from literature. In order to illustrate how this could be achieved, the standard statistical model predictions have been compared with those obtained with the new code features. A single set of parameters of the new code simultaneously well reproduces the neutron, proton and α-particle energy spectra, angular distributions, and multiplicities in the fusion–evaporation channel. By way of example, here the calculations are compared with experimental data of the reaction 60Ni+100Mo at the excitation energy of about 280MeV (Charity et al., 2003) [3]. The capability to reduce the execution time by exploiting the code parallelization is also discussed.

Statistical Modeling for PM10, PM2.5 and PM1 at Gangneung Affected by Local Meteorological Variables and PM10 and PM2.5 at Beijing for Non- and Dust Periods

Multiple statistical prediction modeling of PM10, PM2.5 and PM1 at Gangneung city, Korea, was performed in association with local meteorological parameters (air temperature, wind speed and relative humidity) and PM10 and PM2.5 concentrations of an upwind site in Beijing, China, in the transport route of Chinese yellow dusts which originated from the Gobi Desert and passed through Beijing to the city from 18 March to 27 March 2015. Before and after the dust periods, the PM10, PM2.5 and PM1 concentrations showed as being very high at 09:00 LST (the morning rush hour) by the increasing emitted pollutants from vehicles and flying dust from the road and their maxima occurred at 20:00 to 22:00 LST (the evening departure time) from the additional pollutants from resident heating boilers. During the dust period, these peak trends were not found due to the persistent accumulation of dust in the city from the Gobi Desert through Beijing, China, as shown in real-time COMS-AI satellite images. Multiple correlation coefficients among PM10, PM2.5 and PM1 at Gangneung were in the range of 0.916 to 0.998. Multiple statistical models were devised to predict each PM concentration, and the significant levels through multi-regression analyses were p &lt; 0.001, showing all the coefficients to be significant. The observed and calculated PM concentrations were compared, and new linear regression models were sequentially suggested to reproduce the original observed PM values with improved correlation coefficients, to some extent.

A statistical fatigue life prediction model applicable for fibre-reinforced roller-compacted concrete pavement

This study was conducted to predict the laboratory-scale fatigue life of the fibre-reinforced roller-compacted concrete pavement by determining how macro-synthetic and steel fibres affected its mechanical and the fracture properties. Accordingly, compressive, flexural and indirect tensile strength tests were carried out first and then fatigue tests were done at different stress levels and the related fatigue life was predicted using a multiple linear regression model based on the effective mechanical properties. Results showed that the stress level, fracture energy, and fibre content, respectively, were significant influential factors in the fatigue life prediction process. In the fatigue test, the introduced failure point parameter revealed that the more was the amount of the failure point, the greater was the energy absorbed by the fibres at the cracking zone through the stress transfer mechanism. Finally, the fatigue prediction model was validated through misspecification, heteroscedasticity, and multicollinearity tests which confirmed the regression model’s multilinearity assumption.

Echocardiographic reference intervals for right ventricular indices, including 3-dimensional volume and 2-dimensional strain measurements in healthy dogs.

BackgroundThere is currently a lack of reference intervals (RIs) for the novel measures like 3‐dimensional (3D) echocardiography or speckle‐tracking strain for assessment of right ventricular (RV) structure and function.ObjectivesTo generate RIs and to determine the influence of age, heart rate, and body weight (BW) on various RV function indices using a dedicated RV software for 3D RV end‐diastolic volume (EDV), end‐systolic volume (ESV), ejection fraction (EF), global and free wall RV longitudinal strain (RVLS), end‐diastolic area (RVEDA), end‐systolic area (RVESA), fractional area change (FAC), tricuspid annular plane systolic excursion (TAPSE), and tissue Doppler imaging (TVI)‐derived systolic myocardial velocity of the lateral tricuspid annulus (S′).AnimalsHealthy adult client‐owned dogs (n = 211) of various breeds and ages.MethodsProspective study. Reference intervals were estimated as statistical prediction intervals using allometric scaling for BW‐dependent variables. Right‐sided (upper limit) or left‐sided (lower limit) 95% RIs were calculated for every variable. Inter‐ and intraobserver variability was determined.ResultsMost variables showed clinically acceptable repeatability with coefficient of variation less than 10. Upper or respectively lower RI after allometric scaling to normalize for different BWs were: EDVn ≤ 2.5 mL/kg0.942, ESVn ≤ 1.2 mL/kg0.962, TAPSEn ≥ 4.5 mm0.285, RVEDAn ≤ 1.4 cm2/kg0.665, RVESAn ≤ 0.8 cm2/kg0.695, and TVI S′n ≥ 5.6 cm/s/kg0.186. The calculated limits for indices without allometric normalization were: EF > 42.1%, FAC > 30.0%, free wall RVLS < −20.8%, and global RVLS < −18.3%.ConclusionsEchocardiographic RIs for RV structure and function are provided.

Approach to a new forecasting method of hydrological variables in tropical basins

In this paper, an approach to a new method for the spatio-temporal forecasting of hydrological variables in basins is proposed termed as CIHAM-UC-SPF-HV. This was demonstrated using as hydrological variables to the precipitation and evaporation measured in 174 and 82 monitoring stations respectively, during period 1960–2000 located in the UTM zone 19 North of Venezuela and the indirect variable estimated by a water balance. This comprises the stages: 1) Collection of hydrological data, 2) Estimation of hydrological variable's spatial prediction, 3) Temporal forecasting of parameters derived from statistical spatial prediction model for hydrological variables, 4) Spatio-Temporal forecasting of hydrological variables, and 5) Validation of the results in the forecasted hydrological variables. The main advantages are: 1) the combination of deterministic and spatially correlated random components, both of these supported by historical records associated to time series of hydrological variables, and 2) multiple mathematical structures contribute to predict the parameters of statistical spatial forecasting model of the hydrological variables, selecting that by which the seasonal pattern and trend to be the closest to the observed values. The method is suitable to reproduce the spatio-temporal pattern between observed and forecasted values below two standard deviation of the mean values.

A comparison of two neural network architectures for fast structural response prediction

AbstractIn this contribution, we compare two different neural network architectures to predict the response statistics of structures. The overall goal is a significant speed‐up of the numerically expensive Monte Carlo simulation. The first approach is based on a convolutional neural network that learns from the whole excitation history, whereas the second approach is based on a feed‐forward network architecture learning from hand‐designed features. Both procedures use supervised learning: The neural networks learn from an initial subset before the prediction of the response statistics of the Monte Carlo simulation is possible.

GIS-Based bivariate statistical Model prediction of groundwater potential mapping For Sustainable developments in Suez Governorate, Egypt

GIS-Based bivariate statistical Model prediction of groundwater potential mapping For Sustainable developments in Suez Governorate, Egypt

GIS-Based Bivariate Statistical Model Prediction of Groundwater Potential Mapping for Sustainable Developments in Suez Governorate, Egypt Document Type : Original research articles

GIS-Based Bivariate Statistical Model Prediction of Groundwater Potential Mapping for Sustainable Developments in Suez Governorate, Egypt Document Type : Original research articles

Solar Flare Prediction Based on the Fusion of Multiple Deep-learning Models

Abstract Solar flare formation mechanisms and their corresponding predictions have commonly been difficult topics in solar physics for decades. The traditional forecasting method manually constructs a statistical relationship between the measured values of solar active regions and solar flares that cannot fully utilize the information related to solar flares contained in observational data. In this article, we first used neural-network methods driven by the measured magnetogram and magnetic characteristic parameters of the sunspot group to learn the prediction model and predict solar flares. The prediction fusion model is based on a deep neural network, convolutional neural network, and bidirectional long short-term memory neural network and can predict whether a sunspot group will have a flare event above class M or class C in the next 24 or 48 hr. The real skill statistics (TSS) and F1 scores were used to evaluate the performances of our fusion model. The test results clearly show that this fusion model can make full use of the information related to solar flares and combine the advantages of each independent model to capture the evolution characteristics of solar flares, which is a much better performance than traditional statistical prediction models or any single machine-learning method. We also proposed two frameworks, namely F1_FFM and TSS_FFM, which optimize the F1 score and TSS score, respectively. The cross validation results show that they have their respective advantages in the F1 score and TSS score.

Automated Machine Learning Model Development for Intracranial Aneurysm Treatment Outcome Prediction: A Feasibility Study.

Background: The prediction of aneurysm treatment outcomes can help to optimize the treatment strategies. Machine learning (ML) has shown positive results in many clinical areas. However, the development of such models requires expertise in ML, which is not an easy task for surgeons.Objectives: The recently emerged automated machine learning (AutoML) has shown promise in making ML more accessible to non-computer experts. We aimed to evaluate the feasibility of applying AutoML to develop the ML models for treatment outcome prediction.Methods: The patients with aneurysms treated by endovascular treatment were prospectively recruited from 2016 to 2020. Treatment was considered successful if angiographic complete occlusion was achieved at follow-up. A statistical prediction model was developed using multivariate logistic regression. In addition, two ML models were developed. One was developed manually and the other was developed by AutoML. Three models were compared based on their area under the precision-recall curve (AUPRC) and area under the receiver operating characteristic curve (AUROC).Results: The aneurysm size, stent-assisted coiling (SAC), and posterior circulation were the three significant and independent variables associated with treatment outcome. The statistical model showed an AUPRC of 0.432 and AUROC of 0.745. The conventional manually trained ML model showed an improved AUPRC of 0.545 and AUROC of 0.781. The AutoML derived ML model showed the best performance with AUPRC of 0.632 and AUROC of 0.832, significantly better than the other two models.Conclusions: This study demonstrated the feasibility of using AutoML to develop a high-quality ML model, which may outperform the statistical model and manually derived ML models. AutoML could be a useful tool that makes ML more accessible to the clinical researchers.

Medication control of flunixin in racing horses: Possible detection times using Monte Carlo simulations.

SummaryBackgroundFor medication control in several jurisdictions, withdrawal time is the period of refrain from racing after drug administration. It is set by adding a safety period to an experimental detection time. However, there are no reports of statistical analyses of detection time for the determination of withdrawal time in flunixin meglumine‐treated horses.ObjectiveTo analyse the population pharmacokinetics of flunixin in horses through the generation of a dataset for detection time statistical analysis and predictions via Monte Carlo simulation.Study designExperimental study.MethodsDrug plasma and urine concentrations following single intravenous administration of flunixin 1.1 mg/kg bodyweight (BW) in 10 horses and multiple administration of q 24 hours for 5 days in 10 horses were measured using liquid chromatography with tandem mass spectrometry (LC‐MS/MS). Data were modelled using a nonlinear mixed effect model followed by Monte Carlo simulation. Irrelevant plasma concentration (IPC) and irrelevant urine concentration (IUC) were calculated using the Toutain approach. Detection times were obtained considering the time after the last administration for selected quantiles of 5000 hypothetical horses under the international screening limit (ISL) proposed by the International Federation of Horseracing Authorities (plasma: 1 ng/mL, urine; 100 ng/mL).ResultsFor a regimen of 1.1 mg/kg BW q 24 hours, the IPC and IUC values were 2.0 and 73.0 ng/mL respectively. Detection times in plasma above the ISL for 90% of simulated horses were estimated as 74 hours after a single 1.1 mg/kg dose administration, 149 and 199 hours after multiple doses over 5 days at either 24‐ or 12‐hour intervals respectively. Corresponding detection times in urine were 46, 68 and 104 hours respectively.Main limitationOnly female horses were investigated.ConclusionsStatistical detection times for different flunixin meglumine regimens indicated a delay of detection time in plasma after multiple administrations under ISL.