Highest Correct Classification Rate Research Articles

Accurate Ab Initio and Template-Based Prediction of Short Intrinsically-Disordered Regions by Bidirectional Recurrent Neural Networks Trained on Large-Scale Datasets.

Intrinsically-disordered regions lack a well-defined 3D structure, but play key roles in determining the function of many proteins. Although predictors of disorder have been shown to achieve relatively high rates of correct classification of these segments, improvements over the the years have been slow, and accurate methods are needed that are capable of accommodating the ever-increasing amount of structurally-determined protein sequences to try to boost predictive performances. In this paper, we propose a predictor for short disordered regions based on bidirectional recurrent neural networks and tested by rigorous five-fold cross-validation on a large, non-redundant dataset collected from MobiDB, a new comprehensive source of protein disorder annotations. The system exploits sequence and structural information in the forms of frequency profiles, predicted secondary structure and solvent accessibility and direct disorder annotations from homologous protein structures (templates) deposited in the Protein Data Bank. The contributions of sequence, structure and homology information result in large improvements in predictive accuracy. Additionally, the large scale of the training set leads to low false positive rates, making our systems a robust and efficient way to address high-throughput disorder prediction.

Exploring the feasibility of smart phone microphone for measurement of acoustic voice parameters and voice pathology screening.

The objective of this study is to evaluate the reliability of acoustic voice parameters obtained using smart phone (SP) microphones and investigate the utility of use of SP voice recordings for voice screening. Voice samples of sustained vowel/a/obtained from 118 subjects (34 normal and 84 pathological voices) were recorded simultaneously through two microphones: oral AKG Perception 220 microphone and SP Samsung Galaxy Note3 microphone. Acoustic voice signal data were measured for fundamental frequency, jitter and shimmer, normalized noise energy (NNE), signal to noise ratio and harmonic to noise ratio using Dr. Speech software. Discriminant analysis-based Correct Classification Rate (CCR) and Random Forest Classifier (RFC) based Equal Error Rate (EER) were used to evaluate the feasibility of acoustic voice parameters classifying normal and pathological voice classes. Lithuanian version of Glottal Function Index (LT_GFI) questionnaire was utilized for self-assessment of the severity of voice disorder. The correlations of acoustic voice parameters obtained with two types of microphones were statistically significant and strong (r = 0.73-1.0) for the entire measurements. When classifying into normal/pathological voice classes, the Oral-NNE revealed the CCR of 73.7% and the pair of SP-NNE and SP-shimmer parameters revealed CCR of 79.5%. However, fusion of the results obtained from SP voice recordings and GFI data provided the CCR of 84.60% and RFC revealed the EER of 7.9%, respectively. In conclusion, measurements of acoustic voice parameters using SP microphone were shown to be reliable in clinical settings demonstrating high CCR and low EER when distinguishing normal and pathological voice classes, and validated the suitability of the SP microphone signal for the task of automatic voice analysis and screening.

Investigating the use of support vector machine classification on structural brain images of preterm-born teenagers as a biological marker.

Preterm birth has been shown to induce an altered developmental trajectory of brain structure and function. With the aid support vector machine (SVM) classification methods we aimed to investigate whether MRI data, collected in adolescence, could be used to predict whether an individual had been born preterm or at term. To this end we collected T1–weighted anatomical MRI data from 143 individuals (69 controls, mean age 14.6y). The inclusion criteria for those born preterm were birth weight ≤ 1500g and gestational age < 37w. A linear SVM was trained on the grey matter segment of MR images in two different ways. First, all the individuals were used for training and classification was performed by the leave–one–out method, yielding 93% correct classification (sensitivity = 0.905, specificity = 0.942). Separately, a random half of the available data were used for training twice and each time the other, unseen, half of the data was classified, resulting 86% and 91% accurate classifications. Both gestational age (R = –0.24, p<0.04) and birth weight (R = –0.51, p < 0.001) correlated with the distance to decision boundary within the group of individuals born preterm. Statistically significant correlations were also found between IQ (R = –0.30, p < 0.001) and the distance to decision boundary. Those born small for gestational age did not form a separate subgroup in these analyses. The high rate of correct classification by the SVM motivates further investigation. The long–term goal is to automatically and non–invasively predict the outcome of preterm–born individuals on an individual basis using as early a scan as possible.

Combined Use of Standard and Throat Microphones for Measurement of Acoustic Voice Parameters and Voice Categorization

The aim of the present study was to evaluate the reliability of the measurements of acoustic voice parameters obtained simultaneously using oral and contact (throat) microphones and to investigate utility of combined use of these microphones for voice categorization. Voice samples of sustained vowel /a/ obtained from 157 subjects (105 healthy and 52 pathological voices) were recorded in a soundproof booth simultaneously through two microphones: oral AKG Perception 220 microphone (AKG Acoustics, Vienna, Austria) and contact (throat) Triumph PC microphone (Clearer Communications, Inc, Burnaby, Canada) placed on the lamina of thyroid cartilage. Acoustic voice signal data were measured for fundamental frequency, percent of jitter and shimmer, normalized noise energy, signal-to-noise ratio, and harmonic-to-noise ratio using Dr. Speech software (Tiger Electronics, Seattle, WA). The correlations of acoustic voice parameters in vocal performance were statistically significant and strong (r = 0.71-1.0) for the entire functional measurements obtained for the two microphones. When classifying into healthy-pathological voice classes, the oral-shimmer revealed the correct classification rate (CCR) of 75.2% and the throat-jitter revealed CCR of 70.7%. However, combination of both throat and oral microphones allowed identifying a set of three voice parameters: throat-signal-to-noise ratio, oral-shimmer, and oral-normalized noise energy, which provided the CCR of 80.3%. The measurements of acoustic voice parameters using a combination of oral and throat microphones showed to be reliable in clinical settings and demonstrated high CCRs when distinguishing the healthy and pathological voice patient groups. Our study validates the suitability of the throat microphone signal for the task of automatic voice analysis for the purpose of voice screening.

Integration of classifiers analysis and hyperspectral imaging for rapid discrimination of fresh from cold-stored and frozen-thawed fish fillets

The investigation of visible and near infrared hyperspectral imaging (400–1000nm) coupled with classifiers and spectral pre-processing techniques was conducted to discriminate fresh from cold-stored (4°C for 7days) and frozen-thawed (−20°C and −40°C for 30days) grass carp fish fillets. Four classifiers with four spectral pretreatment methods were applied to establish the classification models. Compared with the original models established using the full wavelengths, the classification models with three classifiers of soft independent modeling of class analogy (SIMCA), least squares-support vector machine (LS-SVM) and probabilistic neural network (PNN) in tandem with the first derivative pretreatment showed the best classification performance and the highest correct classification rate (CCR) of 94.29%. In addition, in order to reduce the high dimensionality of hyperspectral images, seven optimal wavelengths were selected by successive projections algorithm (SPA) and used to simplify the classification models. The simplified model obtained by the LS-SVM classifier coupled with the first derivative pre-processing method also presented good prediction accuracy with the CCR of 91.43%. The results demonstrated that the integration of hyperspectral imaging and classifiers analysis had a great potential for on-line detection and was feasible to rapidly and non-invasively discriminate fresh and frozen-thawed fish fillets.

Using the Personality Assessment Inventory to Identify ADHD-Like Symptoms

Objective: The purpose of this study was to investigate the possibility of creating a new scale on the Personality Assessment Inventory (PAI) to assess for ADHD-like symptomatology. Method: Two approaches to scale development were compared: (a) stepwise regression using PAI scales and subscales as predictors and the CAARS-H index as the dependent variable and (b) a MANOVA using discharge diagnosis as the grouping variable and PAI scales and subscales as the dependent variables. Results: Stepwise regression identified six PAI scales and subscales (PIM, RXR, SOM-S, ARD-T, MAN-A, and SCZ-T) that were responsible for a large proportion of variance (39%) in the CAARS-H index, and the MANOVA also identified six subscales that significantly differed between those clients eventually diagnosed with ADHD and those not diagnosed (ANT-S, SCZ-T, MAN-A, ANX-C, BOR-S, and MAN-G). Conclusion: Although each set produced statistically significant findings, neither produced high correct classification rates in the eventual diagnosis of ADHD.

Assessing influential trade effects via high-frequency market reactions

In the literature, traders are often classified into informed and uninformed and the trades from informed traders have market impacts. We investigate these trades by first establishing a scheme to identify the influential trades from the ordinary trades under certain criteria. The differential properties between these two types of trades are examined via the four transaction states classified by the trade price, trade volume, quotes, and quoted depth. Marginal distribution of the four states and the transition probability between different states are shown to be distinct for informed trades and ordinary liquidity trades. Furthermore, four market reaction factors are introduced and logistic regression models of the influential trades are established based on these four factors. Empirical study on the high-frequency transaction data from the NYSE TAQ database show supportive evidence for high correct classification rates of the logistic regression models.

Evaluation of the Manchester Triage System in the Acute Coronary Syndrome

Background: Cardiovascular diseases are the leading cause of death worldwide. The search for safe solutions to ensure priority in the healthcare of patients with severe conditions, such as acute coronary syndrome, is a determining factor in the prognosis. Objective: To evaluate the healthcare of patients with acute coronary syndrome undergoing the Manchester triage system applied by nurses. Methods: Retrospective observational study based on the review of medical reports of patients admitted for acute coronary syndrome between October 2010 and April 2012 in a public hospital of cardiology in Joinville, Brazil. The following were assessed: health history, symptoms on admission, medical diagnosis, classification by the Manchester triage system, opening hours and procedures, information on admission, and the outcome of the case. The data were entered into an e-form and submitted to statistical analysis. Results: Of the 191 patients selected, 65.0% were men and 47.0% had a history of coronary artery disease. The Manchester triage system allowed correctly classifying 80.1% of patients. The flowchart “thorax pain” with the discriminator “chest pain” was the most used. The average waiting time for classification, medical care and first electrocardiogram were 12.2, 28.6 and 24.9 minutes, respectively. Conclusion: The Manchester triage system allowed a high rate of correct classification by the nurse of patients with acute coronary syndrome.

Sex estimation using the femur of Austrians born in the 19th to the middle of the 20th century.

Sex estimation of human bones or bone fragments is crucial for the identification process of unknown bodies. Regional differences and secular changes lead to ever changing osteometric standards for different populations. This study provides femoral dimensions of Austrians born between 1822 and 1949 and evaluates the discriminating power of sex estimation functions developed by discriminant analyses. Depending on the state of preservation up to 127 femora of adult individuals (72 females and 55 males) were measured. The following means and standard variations were obtained: Maximum length (male: 449.7 ± 16.7mm, female: 413.9 ± 16.3mm), head circumference (148.5 ± 7.8 vs. 133.8 ± 6.4mm), vertical (46.9 ± 2.4 vs. 41.9 ± 2.1mm), transverse (46.3 ± 2.4 vs. 41.5 ± 2.0mm), and maximum head diameter (47.0 ± 2.3 vs. 42.0 ± 2.1mm), as well as sagittal (28.3 ± 2.2 vs. 26.2 ± 1.8mm), transverse (27.9 ± 2.5 vs. 26.1 ± 2.0mm), and maximum midshaft diameter (29.2 ± 2.1 vs. 27.3 ± 1.6mm) and finally condylar width (79.6 ± 4.6 vs. 71.9 ± 3.8mm). The (cross-validated) discriminant analysis for single measurements showed that the best classification is obtained using femur head dimensions, with correct sexing rates between 84.8 % for the circumference and 87.8 % for the maximum diameter. The maximum length achieved a similarly high rate of correct classification with 86.5 %, whereas the rate for the condylar width (80.6 %) was somewhat lower. The potential of midshaft dimensions for sex estimation was substantially weaker (62.3 to 70.8 %). With regard to a multivariate analysis, a stepwise selection procedure favoured a combination of maximum length and vertical head diameter with 90.1 % being correctly classified. A combination of maximum length, maximum head diameter and maximum midshaft diameter resulted in an even higher rate of 91.5 %. These Austrian standards may provide additional possibilities for sexing unknown bodies and skeletal remains born in the 19(th) to the middle of the 20(th) century.

Developmental Changes in Morphology of the Middle and Posterior External Cranial Base in Modern Homo sapiens.

The basicranium has been described as phylogenetically informative, developmentally stable, and minimally affected by external factors and consequently plays an important role in cranial size and shape in subadult humans. Here basicranial variation of subadults from several modern human populations was investigated and the impact of genetic relatedness on basicranial morphological similarities was investigated. Three-dimensional landmark data were digitized from subadult basicrania from seven populations. Published molecular data on short tandem repeats were statistically compared to morphological data from three ontogenetic stages. Basicranial and temporal bone morphology both reflect genetic distances in childhood and adolescence (5–18 years), but not in infancy (<5 years). The occipital bone reflects genetic distances only in adolescence (13–18 years). The sphenoid bone does not reflect genetic distances at any ontogenetic stage but was the most diagnostic region evaluated, resulting in high rates of correct classification among populations. These results suggest that the ontogenetic processes driving basicranial development are complex and cannot be succinctly summarized across populations or basicranial regions. However, the fact that certain regions reflect genetic distances suggests that the morphology of these regions may be useful in reconstructing population history in specimens for which direct DNA evidence is unavailable, such as archaeological sites.

Resampling-based Classification Using Depth for Functional Curves

The depths, which have been used to detect outliers or to extract a representative subset, can be applied to classification. We propose a resampling-based classification method based on the fact that resampling techniques yield a consistent estimator of the distribution of a statistic. The performance of this method was evaluated with eight contaminated models in terms of Correct Classification Rates (CCRs), and the results were compared with other known methods. The proposed method consistently showed higher average CCRs and 4% higher CCR at the maximum compared to other methods. In addition, this method was applied to Berkeley data. The average CCRs were between 0.79 and 0.85.

Combination of spectra and texture data of hyperspectral imaging for differentiating between free-range and broiler chicken meats

This study investigated the potential of visible and near infrared (Vis/NIR) hyperspectral imaging (HSI) to differentiate between free-range and broiler chicken meats. 120 hyperspectral images of chicken fillets were acquired and then calibrated for reflectance. Spectral data were extracted from the region of interest (ROI), followed by multiple scatter correction (MSC) to reduce the noise. Successive projection algorithm (SPA) was used to select optimal wavelengths from the full spectra. On the other hand, principal component analysis (PCA) was applied to select optimum characteristic images, and the first two principal component (PC) images were selected because PC1 and PC2 explained over 95% of variances of all spectra. Then, gray-level gradient co-occurrence matrix (GLGCM) was implemented on PC1 and PC2 images to extract 30 textural variables in total. Based on data fusion, classification models were established, in which the radial basis function-support vector machine (RBF-SVM) model gave the best results with high correct classification rate (CCR) of 93.33% for the prediction samples, demonstrating that combining spectra with texture data were effective for differentiating between free-range and broiler chicken meats.

Application of Purkait's triangle method on a skeletal population from southern Europe

The existence of sexual dimorphism in individual elements of the human skeleton allows sex determination and significantly enhances the identification of fragmentary skeletal material. Traditionally, the pelvis and the skull are used for an accurate sex determination. Sometimes they are not available and alternative methods have to be used. The aim of this work is to study the sexual dimorphism of the proximal portion of the posterior femur by analyzing the biometric data of the Purkait's triangle. The studied sample comprises 203 individuals (112 males and 91 females) from the Modern Human Skeletal Reference Collection (The Athens Collection) of the Department of Animal & Human Physiology (National & Kapodistrian University of Athens). The biometric data were analyzed by Discriminant Analysis and equations were generated for sex determination, which gave an overall correct classification of 78.3%. For the right and the left femur the percentages of correct classification were 77.8% and 75.9% respectively. These accuracy rates were compared to those of an Indian population and were found to be lower (86.5% for males and 86.3% for females). The study of sexual dimorphism (SD) in Greek, Indian, European-American and African-American populations gave interesting results. The AB diameter in the Greek sample shows the lowest degree of SD while the AC diameter shows the highest. In the other three population samples, the BC diameter shows the highest degree of SD. We assume that the femur anatomy of the Greeks is the major cause that the Purkait method does not give high rates of correct classification. It is therefore advisable that practitioners exercise caution when using this method on skeletal material from Greece, especially in forensic contexts.

Predicción de la reprobación de cursos de matemática básicos en las carreras de Física, Meteorología, Matemática, Ciencias Actuariales y Farmacia

Recibido 5 de junio de 2013 • Corregido 4 de abril de 2014 • Aceptado 22 de agosto de 2014Este artículo sintetiza un estudio realizado en el 2013, con el objetivo de predecir la reprobación en cursos de matemática por parte de los estudiantes de las carreras de Farmacia, Matemática, Ciencias Actuariales, Física y Meteorología de la Universidad de Costa Rica (UCR). Mediante la técnica estadística de regresión logística, aplicada a la cohorte del 2010, se pronosticó la reprobación de los estudiantes de estas carreras en uno de sus cursos introductorios de matemática (Cálculo I para Física y Meteorología, Principios de matemática para Matemática y Ciencias actuariales y ecuaciones diferenciales aplicadas para Farmacia). Para estos modelos se utilizaron, como variables predictivas, los componentes del promedio de admisión a la UCR, el género del estudiante y el porcentaje de ítems correctos en la Prueba de Habilidades Cuantitativas (HC). La variable con mayor importancia en todos los modelos fue HC y los modelos con mejores porcentajes de clasificación correcta de casos fueron las regresiones logísticas. Para los casos utilizados en la estimación de los modelos de Física-Meteorología, Farmacia y Matemática-Ciencias Actuariales se obtuvo un 89.8%, 73.6% y 93.9% de clasificación correcta, respectivamente.

Cognitive diagnostic assessment with different weight for attribute: based on the Dina model.

An attribute weight calculation method which used a Bayesian network and the least squares distance method was proposed to assign different weights to different attributes in cognitive diagnosis. This method is independent of any specific cognitive diagnostic models, so it is practicable to consider attribute weight not only in the models with explicit expression but also in item response theory-based cognitive diagnostic methods. Simulation studies showed that the data fit for the least squares distance method was excellent and the weighted status can yield higher correct classification rates than the unweighted status. The weighted status had a promising performance in recognizing the knowledge states of examinees for various slippage probabilities under different attribute hierarchies. The numbers of items and attributes could also affect the examinees' classification accuracy.

Detecting cognitive impairment after concussion: sensitivity of change from baseline and normative data methods using the CogSport/Axon cognitive test battery.

Concussion-related cognitive impairments are typically evaluated with repeated neuropsychological assessments where post-injury performances are compared with pre-injury baseline data (baseline method). Many cases of concussions, however, are evaluated in the absence of baseline data by comparing post-injury performances with normative data (normative method). This study aimed to compare the sensitivity and specificity of these two methods using the CogSport/Axon test battery. Normative data and reliable change indices were computed from a non-injured athlete sample (n = 235). Test-retest data from non-injured (n = 260) and recently concussed (n = 29) athlete samples were then used to compare the two methods. The baseline method was found to be more sensitive than the normative method, and both methods had high specificity and overall correct classification rates. This suggests that while the normative method identifies most cases of recent concussions, the baseline method remains a more precise approach to assessing concussion-related cognitive impairments.

Would credit scoring work for Islamic finance? A neural network approach

Purpose – This paper aims to distinguish whether the decision-making process of the Islamic financial houses in the UK can be improved through the use of credit scoring modeling techniques as opposed to the currently used judgmental approaches. Subsidiary aims are to identify how scoring models can reclassify accepted applicants who later are considered as having bad credit and how many of the rejected applicants are later considered as having good credit, and highlight significant variables that are crucial in terms of accepting and rejecting applicants, which can further aid the decision-making process. Design/methodology/approach – A real data set of 487 applicants is used consisting of 336 accepted credit applications and 151 rejected credit applications made to an Islamic finance house in the UK. To build the proposed scoring models, the data set is divided into training and hold-out subsets. The training subset is used to build the scoring models, and the hold-out subset is used to test the predictive capabilities of the scoring models. Seventy per cent of the overall applicants will be used for the training subset, and 30 per cent will be used for the testing subset. Three statistical modeling techniques, namely, discriminant analysis, logistic regression (LR) and multilayer perceptron (MP) neural network, are used to build the proposed scoring models. Findings – The findings reveal that the LR model has the highest correct classification (CC) rate in the training subset, whereas MP outperforms other techniques and has the highest CC rate in the hold-out subset. MP also outperforms other techniques in terms of predicting the rejected credit applications and has the lowest misclassification cost above other techniques. In addition, results from MP models show that monthly expenses, age and marital status are identified as the key factors affecting the decision-making process. Originality/value – This contribution is the first to apply credit scoring modeling techniques in Islamic finance. Also in building a scoring model, the authors' application applies a different approach by using accepted and rejected credit applications instead of good and bad credit histories. This identifies opportunity costs of misclassifying credit applications as rejected.

Quantification ofγ- andα-tocopherol isomers in combination with pattern recognition model as a tool for differentiating dry-cured shoulders of Iberian pigs raised on different feeding systems

Quantification of γ- and α-tocopherol in dry-cured shoulders of Iberian pigs was evaluated as a tool for differentiating feeding backgrounds or regimens. Samples (n = 115) were obtained over two different seasons from the four categories of pigs described in the Industry Quality Policy, i.e. pigs fed in free-range conditions (FREE-RANGE), pigs fed in free-range conditions and provided feed supplements (FREE-FEED), pigs fed outdoors with feed and with access to grass (FEED-OUT) and pigs fed in intensive conditions with feed (FEED). Linear discriminant functions were calculated and validated. The validation results showed that 20% of the muscle samples were not correctly classified into the four feeding categories, giving an 80% success rate. The FEED group had the lowest proportion of errors, with 100% of samples correctly classified. For the FREE-RANGE group, 87% of samples were assigned to the correct feeding system by cross-validation; however, 13% were considered as FREE-FEED. A higher rate of correct classification can be obtained when using three categories or by calculating the weight gain in free-range conditions using regression equations. Taking into account the high variability of the samples and the high success in classification, these results are of interest and may be applied in practical situations.

Empowering difficult classes with a similarity-based aggregation in multi-class classification problems

One-vs-One strategy divides the original multi-class problem into as many binary classification problems as pairs of classes. Then, independent base classifiers are learned to face each problem, whose outputs are combined to predict a single class label. This way, the accuracy of the baseline classifiers without decomposition is usually enhanced, aside from enabling the usage of binary classifiers, i.e., Support Vector Machines, to solve multi-class problems. This paper analyzes the fact that existing aggregations favor easily recognizable classes; hence, the accuracy enhancement mainly comes from the higher correct classification rates over these classes. Using other evaluation criteria, the significant improvements of One-vs-One are diminished, showing a weakness due to the presence of difficult classes. Difficult classes can be defined as those obtaining a lower correct classification rate than that obtained by the other classes in the problem. After studying the problem of difficult classes in this framework and aiming to empower these classes, a novel similarity-based aggregation is presented, which generalizes the well-known weighted voting. The experimental analysis shows that the new methodology is able to increase the recognition of difficult classes, obtaining a more balanced performance over all classes, which is a desirable behavior. The methodology is tested within several Machine Learning paradigms and is compared with the state-of-the-art on aggregations for One-vs-One strategy. The results are contrasted by the proper statistical tests, as suggested in the literature.

Diffusion-weighted MRI, 11C-choline PET and 18F-fluorodeoxyglucose PET for predicting the Gleason score in prostate carcinoma

To evaluate the accuracy of transrectal ultrasound-guided (TRUS) biopsy, diffusion-weighted (DW) magnetic resonance imaging (MRI), (11)C-choline (CHOL) positron emission tomography (PET), and (18)F-fluorodeoxyglucose (FDG) PET in predicting the prostatectomy Gleason risk (GR). The study included 21 patients who underwent TRUS biopsy and multi-technique imaging before radical prostatectomy. Values from five different tests (TRUS biopsy, DW MRI, CHOL PET, FDG PET, and combined DW MRI/CHOL PET) were correlated with the prostatectomy GR using Spearman's ρ. Tests that were found to have significant correlations were used to classify patients into GR groups. The following tests had significant correlations with prostatectomy GR: TRUS biopsy (ρ = 0.617, P = 0.003), DW MRI (ρ = -0.601, P = 0.004), and combined DW MRI/CHOL PET (ρ = -0.623, P = 0.003). CHOL PET alone and FDG PET only had weak correlations. The correct GR classification rates were 67% with TRUS biopsy, 67% with DW MRI, and 76% with combined DW MRI/CHOL PET. DW MRI and combined DW MRI/CHOL PET have significant correlations and high rates of correct classification of the prostatectomy GR, the strength and accuracy of which are comparable with TRUS biopsy. • Accurate determination of the Gleason score is essential for prostate cancer management. • DW MRI ± CHOL PET correlated significantly with prostatectomy Gleason score. • These correlations are similar to that between TRUS biopsy and prostatectomy.