Likelihood Ratio Test Criterion Research Articles

Take Two: Second Systolic Blood Pressure Provides Prognostic Information in Trauma

Abstract Introduction: An initial systolic blood pressure (SBP1) of <110 mmHg has been shown to predict mortality. However, SBP1 may not reflect hemodynamic changes during initial resuscitation. We aimed to determine if the second recorded blood pressure (SBP2) could provide additional prognostic value. Methods: An 8-year retrospective chart review was performed including patients who underwent trauma activations at a single institution. The initial systolic blood pressure (SBP 1) and second systolic blood pressure (SBP 2) were analyzed. Difference between the first and second systolic blood pressure (ΔSBP) was defined as SBP2 − SBP1. The primary outcome was inhospital mortality, and the secondary outcome was receipt of ≥2 units of blood in the first 4 h of admission. Univariable analysis and logistic regression analysis were performed to assess the relationship of SBP2 and ΔSBP with the study outcomes. Regression model fit was assessed by the likelihood ratio test and Akaike information criterion. Results: Eight thousand seven hundred and ninety-eight patients were included with 12% and 13% presenting with SBP1 <110 mmHg and SBP2 <110 mmHg, respectively. Four hundred and six (5%) died and 327 (4%) received ≥2 units of blood in the first 4 h. The addition of ΔSBP to regression models improved model fit in explaining the primary and secondary outcomes. Subgroup analysis found that the addition of ΔSBP improved model fit for those with penetrating mechanism of injury and those with Injury Severity Score of ≥9, but not for those with blunt moderate or severe traumatic brain injury. Conclusion: SBP2 offers additional prognostic value in predicting trauma outcomes. Incorporating subsequent hemodynamic data during resuscitation beyond the initial SBP in trauma databases should be considered.

Subspace‐based distributed target detection method with small training data samples

AbstractDetecting distributed targets precisely in homogeneous environments has been a hot topic in radar signal processing. Generally, distributed targets are often modelled with subspace models of unknown coordinates, and clutter is modelled as the complex Gaussian distribution with zero mean and unknown covariance matrix, while covariance matrix is estimated with a set of training data without the target signal. However, in practice, the complexity of the external environment makes the training data that satisfy the condition of independent homogeneous distribution less available. Therefore, it is assumed that the covariance matrix of the clutter is persymmetric structure and the approach of dimensionality reduction using subspace transformations is introduced, two detectors based upon generalised likelihood ratio test criterion and Wald test criterion in homogeneous environments are proposed. Theoretical analyses indicate the constant false alarm rate characteristics of the two proposed detectors for unknown clutter covariance matrices. Simulation analyses indicate that the proposed detector works well even with fewer training data samples, and its detection performance outperforms that of existing contrast detectors.

An Adaptive Radar Target Detection Method Based on Alternate Estimation in Power Heterogeneous Clutter

Multichannel radars generally need to utilize a certain amount of training samples to estimate the covariance matrix of clutter for target detection. Due to factors such as severe terrain fluctuations and complex electromagnetic environments, the training samples usually have different statistical characteristics from the data to be detected. One of the most common scenarios is that all data have the same clutter covariance matrix structure, while different data have different power mismatches, called power heterogeneous characteristics. For detection problems in the power heterogeneous clutter environments, we propose detectors based on alternate estimation, using the generalized likelihood ratio test (GLRT) criterion, Rao criterion, Wald criterion, Gradient criterion, and Durbin criterion. Monte Carlo simulation experiments and real data indicate that the detector based on the Rao criterion has the highest probability of detection (PD). Furthermore, when signal mismatch occurs, the detector based on the GLRT criterion has the best selectivity, while the detector based on the Durbin criterion has the most robust detection performance.

A new cure rate model with discrete and multiple exposures

Cure rate models are mostly used to study data arising from cancer clinical trials. Its use in the context of infectious diseases has not been explored well. In 2008, Tournoud and Ecochard first proposed a mechanistic formulation of cure rate model in the context of infectious diseases with multiple exposures to infection. However, they assumed a simple Poisson distribution to capture the unobserved pathogens at each exposure time. In this paper, we propose a new cure rate model to study infectious diseases with discrete multiple exposures to infection. Our formulation captures both over-dispersion and under-dispersion with respect to the count on pathogens at each time of exposure. We also propose a new estimation method based on the expectation maximization algorithm to calculate the maximum likelihood estimates of the model parameters. We carry out a detailed Monte Carlo simulation study to demonstrate the performance of the proposed model and estimation algorithm. The flexibility of our proposed model also allows us to carry out a model discrimination. For this purpose, we use both likelihood ratio test and information-based criteria. Finally, we illustrate our proposed model using a recently collected data on COVID-19.

Combining plasma p‐tau231 and glial fibrillary acidic protein produces higher discriminative accuracy for amyloid positivity than other blood‐based biomarker combinations

AbstractBackgroundPlasma phosphorylated tau (p‐tau)231 is an emerging blood‐based biomarker that has shown great promise for early identification of individuals at risk of Alzheimer’s disease (AD). However, it is unclear if this biomarker is sufficient as a standalone test or if diagnostic classification improves when simultaneously considering other plasma biomarkers. We investigated various plasma biomarkers, individually and in combination, in relation to cerebrospinal fluid (CSF) amyloid positivity.MethodVanderbilt Memory and Aging Project participants free of clinical dementia and stroke (n = 155, 72±6 years, 33% female) underwent fasting lumbar puncture and blood draw. CSF β‐amyloid42 (Aβ42) and β‐amyloid40 (Ab40) and plasma p‐tau231, glial fibrillary acidic protein (GFAP), Aβ42, Aβ40, and neurofilament light chain (NfL) were analyzed in batch. The Aβ42/40 ratio was used for all analyses and CSF amyloid positivity was defined as Aβ42/40 ratio <0.072. Logistic regression models related each biomarker to amyloid positivity, adjusting for age, sex, race/ethnicity, apolipoprotein E‐ε4 status, and cognitive status. Subsequent models included p‐tau231 as the predictor in a base model with other biomarkers being added individually and then in combination with other biomarkers to the base model. Models were compared using likelihood ratio test (LRTest) and Akaike Information Criterion (AIC) values.ResultIn single predictor models, p‐tau231 (C‐index = 0.87, p = 0.005) and GFAP (C‐index = 0.87, p = 0.01) were associated with amyloid positivity, while other biomarkers were not (p‐values>0.35). In models including p‐tau231 and various biomarker combinations, p‐tau231 remained significant in each model (p‐values<0.02). In comparison to the base model with p‐tau231 only, models adding GFAP (AIC = 143.8, LRTest p = 0.008), GFAP and NfL (AIC = 145.5, LRTest p = 0.02), and GFAP and Aβ42/40 (AIC = 145.8, LRTest p = 0.03) improved prediction of amyloid positivity. The model including plasma p‐tau231 and GFAP was the best fitting model (C‐index = 0.89).ConclusionIn this sample of community‐dwelling older adults free of clinical dementia, we found plasma p‐tau231 and GFAP were associated with amyloid positivity. Models including both p‐tau231 and GFAP performed best, and adding additional plasma biomarkers to this combination did not produce a better fitting model. Taken together, these findings highlight the utility of plasma p‐tau231 for screening for amyloid status and suggest that including plasma GFAP will improve discriminative accuracy.

Abstract 11396: Circulating Biomarkers Improve ChA 2 DS 2 -VASC Risk Prediction in Atrial Fibrillation: The CHA 2 DS 2 -VASc-Biomarkers Model

Introduction: Anticoagulation reduces stroke risk in atrial fibrillation (AF) but increases bleeding. Risk calculators are used to inform decision making for anticoagulation in AF. Yet, existing tools have substantial limitations and improvement efforts failed to change practice. Hypotheses: Biomarkers of ischemic stroke risk for general populations are associated with stroke risk in AF and improve performance of a recommended risk score (CHA 2 DS 2 -VASc). Methods: REGARDS is a prospective cohort of 30,239 Black or White adults aged ≥45 monitored for stroke. Nine blood biomarkers were measured at baseline in participants with AF, no prior stroke, and not taking anticoagulation at baseline. Hazard ratios of ischemic stroke were estimated by Cox models adjusted for demographics and stroke risk factors. Models with CHA 2 DS 2 -VASc alone and with biomarkers (CHA 2 DS 2 -VASc-B) were compared using tests of fit (likelihood ratio test and Aikake information criterion [AIC]) and predictive ability (continuous net reclassification index [NRI]). Results: Among 2,411 participants with baseline AF (median age 69, 55% female, 36% Black), 163 (7%) developed stroke over 13 years. After adjustments, higher cystatin C, growth differentiation factor 15 (GDF-15), interleukin 6, lipoprotein (a), and N-terminal pro-B type natriuretic peptide (NTproBNP) were associated with higher stroke risk (Figure A). Including all biomarkers substantially improved CHA 2 DS 2 -VASc model fit (Figure B; p <0.001; ΔAIC -12.6; <-10 indicates marked improvement) and predictive ability (Figure C; 5-year NRI 0.42). Adding only GDF-15 and NTproBNP resulted in the best model fit and a similar NRI, compared to all biomarkers. Conclusions: Higher levels of five blood biomarkers were associated with ischemic stroke risk in a large general population cohort with AF. Moreover, the CHA 2 DS 2 -VASc-B model was superior to CHA 2 DS 2 -VASc. Findings suggest we can better tailor anticoagulation to reduce stroke in AF.

Response Model Selection in Small Area Estimation Under not Missing at Random Nonresponse

Sverchkov and Pfeffermann[ 16 ] consider Small Area Estimation (SAE) under informative probability sampling of areas and within the sampled areas, and not missing at random (NMAR) nonresponse. To account for the nonresponse, the authors assume a given response model, which contains the outcome values as one of the covariates and estimate the corresponding response probabilities by application of the Missing Information Principle, which consists of defining the likelihood as if there was complete response and then integrating out the unobserved outcomes from the likelihood by employing the relationship between the distributions of the observed and the missing data. A key condition for the success of this approach is the ‘correct’ specification of the response model. In this article, we consider the likelihood ratio test and information criteria based on the appropriate likelihood and show how they can be used for the selection of the response model. We illustrate the approach by a small simulation study. AMS Subject Classification: 62D05, 62D10, 62F10

190. PROGNOSTIC VALUE OF COMBINATION OF LYMPHOCYTE-TO-MONOCYTE-RATIO AND CYFRA 21-1 IN ESOPHAGEAL SQUAMOUS CELL CARCINOMA

Abstract Background Previous studies have shown that serum tumor markers and systematic inflammatory and nutritional indicators, including Cytokeratin fragments (CYFRA) 21–1 and the lymphocyte-to-monocyte ratio (LMR), are associated with the prognosis of esophageal malignancies. However, their individual prognostic sensitivity and specificity are not yet sufficient. This study aims to investigate the potential of a combined score (CA-LMR) based on CYFRA 21-1 levels and LMR as a novel prognostic predictor for patients with esophageal squamous cell carcinoma (ESCC). Methods A total of 460 consecutive patients underwent oesophagectomy were enrolled, and three groups were established based on CA-LMR score. Overall survival (OS) and recurrence-free survival (RFS) were evaluated using the Kaplan–Meier analysis, and associated factors were analyzed by multivariate Cox analysis. Additionally, modified-pStage (mpStage) was developed based on CA-LMR score, the discriminatory ability, monotonicity, homogeneity and prognosis prediction ability were measured by log-rank χ2, linear-trend χ2, likelihood-ratio χ2 test and Akaike Information Criterion. Results The allocation of CA-LMR score of 0, 1 and 2 were 107 (23.3%), 280 (60.9%), and 73 (15.9%). There was a significant association between CA-LMR and male gender (P = 0.001), smaller BMI (P = 0.035), longer tumor lesions (P = 0.002), and high pT, pN, pStage (P < 0.001, P = 0.011, P = 0.001). The 5-year OS for CA-LMR scores of 0, 1, and 2 were 75.4%, 60.2%, and 32.8%, respectively (P < 0.001) (Fig. 1A,B). Multivariate analysis indicated CA-LMR score (P = 0.011) was an independent prognostic factor for OS (Fig. 1C-E). Furthermore, the proposed mpStage system demonstrated superior discriminatory ability, monotonicity, homogeneity and prognosis prediction ability over AJCC 8th pStage system. Conclusions The CA-LMR score, combined with tumor marker and inflammatory index, could use as a potential prognostic indicator, moreover, our modified pStage system exhibited superior stratification and prognostic accuracy for patients with ESCC, making it possible to serve as novel reference for clinical management.

Optimization of Gamma–Power–Log–logistic distribution and its applications in modelling volume of oil spillage

Oil spillage is one of the major problems in the Niger Delta region of Nigeria, which have affected human health and aquaculture. There is a need to estimate the volume of spill per incidence. Estimating the volume of crude oil after spill is a problem in environmental sciences. In this research, a newly-convoluted gamma–power generalized linear model (GLM), which was derived from Gamma–Power–Log–logistic distribution was developed. The proposed model was used to fit a regression model to estimate the spill volume of crude oil in the South–south geopolitical zone of Nigeria. Some statistical properties of the distribution such as, the transformation to a gamma distribution and the likelihood ratio test were derived. The Maximum Likelihood Estimation (MLE) technique, using generalized Newton Approximation with Expectation Maximization (EM) algorithm was used to obtain the model parameters and the consistency of the parameters estimated was justified by a simulation study. The simulation study result showed that the parameters estimated are consistent. The result of the likelihood ratio test (LRT) and other selection criteria adopted showed that the novel GPGLM outperformed other models. The LRT showed how much better the proposed model could be when compared to gamma and normal models as a good fit to bimodal skewed leptokurtic dependent response variable in a regression settings. The new model is an extension of the gamma model and would be very useful for modelling bimodal, skewed and leptokurtic dependent variable in a regression model. The study recommended that clean-up should be done promptly in order to achieve the Sustainable Development Goal of access to clean water and improved quality of life.

Polarization-space-time domain adaptive detection for the heterogeneous array

In this paper, to investigate the effect of different antenna attitudes on the problem of polarimetric target detection, we take the heterogeneous array (HA), which is a uniform linear array (ULA) composed of antennas with different attitudes, as the research object, and use the generalized likelihood ratio test (GLRT) criterion to derive an adaptive detection algorithm. Meanwhile, the probability of detection, as well as the probability of false alarm of this algorithm is presented. Through numerical simulations, we analyze the detection ability of the proposed algorithm in various scenarios, and the detection performance of the HA is compared with the ULA and the uniform linear array consisting of the orthogonal pairs of dipoles (ULAP). The simulation results indicate that the HA can improve its detection performance by exploiting the information in the polarization domain, and therefore, the polarimetric target detection ability of the array, which is composed of antennas with different attitudes, is proved.

Discrimination of Target and Mainlobe Jammers With FDA-MIMO Radar

This letter focuses on discriminating true target from mainlobe jammers using a frequency diverse array-multiple-input multiple-output (FDA-MIMO) radar. To this end, a three-stage detection and discrimination architecture is proposed. Specifically, mainlobe jammers caused by large time delays can be discriminated from the true target based on the range degrees-of-freedom (DOFs) of FDA-MIMO radar, while the true target and jammers generated with small time delays are distinguishable in the fast time domain. In particular, a binary hypothesis test is formulated for discrimination, where the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$H_{1}$</tex-math></inline-formula> hypothesis contains the true target, while the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$H_{0}$</tex-math></inline-formula> hypothesis contains the unknown mainlobe jammer. The mainlobe jammer can be formulated as either a known subspace model or an unknown rank-one model. The former leads to a subspace-based selective detector (SSD) and the latter generates a rank-one-based SD (ROSD). Then, the formulated tests are handled using the generalized likelihood ratio test (GLRT) criterion. At the analysis stage, the selectivity and detection performance are compared with the conventional GLRT and selective receivers. Simulation results validate that the proposed detectors achieve satisfactory detection performance while providing the selectivity to mainlobe jammers.

A GLRT-Based Multi-Pixel Target Detector in Hyperspectral Imagery

In hyperspectral imagery, target detection algorithms are usually based on the spectral signature information. Due to the advance of the spatial resolution of hyperspectral sensors, the ground sample distance may be much smaller than the size of targets. As a result, targets often occupy multiple consecutive pixels, which are referred to as multi-pixel targets. In this paper, we investigate the target detection problem for multi-pixel targets in hyperspectral imagery, when the target spectral signature is known. Jointly exploiting the pixels occupied by a target of interest, we propose a multi-pixel target detector resorting to the generalized likelihood ratio test criterion. Closed-form expressions for the probabilities of the false alarm and detection are derived, which are verified using Monte Carlo simulations. Experimental results on four real hyperspectral datasets show that the proposed detector outperforms its counterparts.

Adaptive Radar Detection in the Clutter and Noise Cover Pulse Jamming Environment

This paper deals with the adaptive detection of the point-like target under clutter and noise cover pulse (NCP) jamming. In the design stage, we assume that the jamming is orthogonal to the signal subspace. Simultaneously, two training datasets are considered available. According to generalized likelihood ratio test (GLRT), Wald, and Rao criteria, NCP-GLRT, NCP-Wald, and NCP-Rao detectors are designed respectively. Then, considering the degraded detection performance of the NCP-Rao, a modified detector called NCP-MRao is proposed. Moreover, the performance of the newly presented detectors is analyzed, and the probability of false alarm and the detection probability for the NCP-GLRT are derived. Finally, numerical simulation results show that compared with the analogous counterparts, the proposed detector can improve detection performance.

Using Path Analysis and Linear Regression to Test for Gender and Participation: Effects in a Culturally Tailored Diabetes Intervention for Latino Adults.

While the incidence and prevalence of type 2 diabetes is higher among Latino/as, Latino men are disproportionately affected and have poorer outcomes. We aimed to determine whether gender impacted any outcomes in a culturally tailored type 2 diabetes (T2D) intervention and to evaluate the effects of gender and intervention participation intensity on outcomes at 6-month follow-up. Nested path and regression models were compared with the likelihood ratio test and information criteria in a sample of Latino/a adults with T2D (n = 222) participating in a T2D community health worker (CHW)-led intervention. Path analysis showed that the effect of the intervention did not vary by gender. The intervention was associated with significant improvements in knowledge of T2D management 0.24 (0.10); p = 0.014, diabetes distress, −0.26 (0.12); p = 0.023, and self-efficacy, 0.61 (0.21); p = 0.005. At 6-month follow-up, improved self-management was associated with greater self-efficacy and Hemoglobin A1c (HbA1c) was lower by −0.18 (0.08); p = 0.021 for each unit of self-management behavior. Linear regressions showed that class attendance and home visits contributed to positive intervention results, while gender was non-significant. Pathways of change in a CHW-led culturally tailored T2D intervention can have a significant effect on participant behaviors and health status outcomes, regardless of gender.

Adaptive Distributed Target Detection for FDA-MIMO Radar in Gaussian Clutter Without Training Data

Since frequency diverse array multiple-input multiple-output (FDA-MIMO) radar possesses additional target range information for potential performance improvement, this article studies adaptive distributed targets detection for FDA-MIMO radar, where the targets are embedded in Gaussian clutter with unknown covariance matrix. The proposed FDA-MIMO radar detection model considers also the distributed targets occupying several secondary range cells, which is different from the classic detection models in multiple-input multiple-output (MIMO) and/or phase array (PA) radars that discuss only point-like targets. By exploiting the FDA-MIMO radar framework for distributed target detection, we propose the detector through a two-step generalized likelihood ratio test criteria without the need of training data and/or a priori covariance matrix. Moreover, closed-form expressions for the probability of false alarm and detection probability are derived, respectively. The proposed detector adheres to the property of a constant false alarm rate because its probability of false alarm is not restricted by the covariance matrix. The proposed method together with all theoretical analysis are verified by numerical results.

Parameter estimation and selection efficiency under Bayesian and frequentist approaches in peach trials

Identification of stable and high-yielding genotypes is a real challenge in peach breeding, since genotype-by-environment interaction (GE) masks the performance of the materials. The aim of this work was to evaluate the effectiveness of parameter estimation and genotype selection solving the linear mixed models (LMM) under frequentist and Bayesian approaches. Fruit yield of 308 peach genotypes were assessed under different seasons and replication numbers arranged in a completely randomized design. Under the frequentist framework the restricted maximum likelihood method to estimate variance component and genotypic prediction was used. Different models considering environment, genotype and GE effects according to the likelihood ratio test and Akaike information criteria were compared. In the Bayesian approach, the mean and the variance components were assumed to be random variables having a priori non-informative distributions with known parameters. According the deviance information criteria the most suitable Bayesian model was selected. The full model was the most appropriate to calculate parameters and genotypic predictions, which were very similar in both approaches. Due to imbalance data, Cullis’s method was the most appropriate to estimate heritability. It was calculated at 0.80, and selecting above 5% of the genotypes, the realized gain of 14.80 kg tree1 was attained. Genotypic frequentist and Bayesian predictions showed a positive correlation (r = 0.9991; P = 0.0001). Since the Bayesian method incorporates the credible interval for genetic parameters, genotypic Bayesian prediction would be a more useful tool than the frequentist approach and allowed the selection of 17 high-yielding and stable genotypes.

Persymmetric subspace detector for distributed target in partially homogeneous environment

This paper deals with the problem of distributed target detection in partially homogeneous Gaussian clutter whose covariance matrix is unknown but persymmetric. It is assumed that primary data and training data share the same clutter covariance matrix structure but different power levels. The target signal is supposed to lie in a multi-rank subspace with unknown coordinates. A persymmetric subspace detector is designed based on the generalised likelihood ratio test criteria. It is theoretically demonstrated that the proposed detector possesses constant false alarm rate property with respect to the unknown clutter covariance matrix as well as the power level. Experimental results illustrate the performance advantage of the proposed detector over the existing competitors, especially in training-limited scenarios.

On a reparameterization of a flexible family of cure models

The existence of items not susceptible to the event of interest is of both theoretical and practical importance. Although researchers may provide, for example, biological, medical, or sociological evidence for the presence of such items (cured), statistical models performing well under the existence or not of a cured proportion, frequently offer a necessary flexibility. This work introduces a new reparameterization of a flexible family of cure models, which not only includes among its special cases, the most studied cure models (such as the mixture, bounded cumulative hazard, and negative binomial cure model) but also classical survival models (ie, without cured items). One of the main properties of the proposed family, apart from its computationally tractable closed form, is that the case of zero cured proportion is not found at the boundary of the parameter space, as it typically happens to other families. A simulation study examines the (finite) performance of the suggested methodology, focusing to the estimation through EM algorithm and model discrimination, by the aid of the likelihood ratio test and Akaike information criterion; for illustrative purposes, analysis of two real life datasets (on recidivism and cutaneous melanoma) is also carried out.

Adaptive Subspace Signal Detection in Structured Interference Plus Compound Gaussian Sea Clutter

This paper discusses the problem of detecting subspace signals in structured interference plus compound Gaussian sea clutter with persymmetric structure. The sea clutter is represented by a compound Gaussian process wherein the texture obeys the inverse Gaussian distribution. The structured interference lies in a known subspace, which is independent with the target signal subspace. By resorting to the two-step generalized likelihood ratio test, two-step Rao, and two-step Wald design criteria, three adaptive subspace signal detectors are proposed. Moreover, the constant false-alarm rate property of the proposed detectors is proved. The experimental results based on IPIX real sea clutter data and simulated data illustrate that the proposed detectors outperform their counterparts.

Bayesian Approaches for Detecting Differential Item Functioning Using the Generalized Graded Unfolding Model.

Differential item functioning (DIF) analysis is one of the most important applications of item response theory (IRT) in psychological assessment. This study examined the performance of two Bayesian DIF methods, Bayes factor (BF) and deviance information criterion (DIC), with the generalized graded unfolding model (GGUM). The Type I error and power were investigated in a Monte Carlo simulation that manipulated sample size, DIF source, DIF size, DIF location, subpopulation trait distribution, and type of baseline model. We also examined the performance of two likelihood-based methods, the likelihood ratio (LR) test and Akaike information criterion (AIC), using marginal maximum likelihood (MML) estimation for comparison with past DIF research. The results indicated that the proposed BF and DIC methods provided well-controlled Type I error and high power using a free-baseline model implementation, their performance was superior to LR and AIC in terms of Type I error rates when the reference and focal group trait distributions differed. The implications and recommendations for applied research are discussed.