Automatic Classification of Protein Sequences into Structure/Function Groups via Parallel Cascade Identification: A Feasibility Study

A recent paper introduced the approach of using nonlinear system identification as a means for automatically classifying protein sequences into their structure/function families. The particular technique utilized, known as parallel cascade identification (PCI), could train classifiers on a very limited set of exemplars from the protein families to be distinguished and still achieve impressively good two-way classifications. For the nonlinear system classifiers to have numerical inputs, each amino acid in the protein was mapped into a corresponding hydrophobicity value, and the resulting hydrophobicity profile was used in place of the primary amino acid sequence. While the ensuing classification accuracy was gratifying, the use of (Rose scale) hydrophobicity values had some disadvantages. These included representing multiple amino acids by the same value, weighting some amino acids more heavily than others, and covering a narrow numerical range, resulting in a poor input for system identification. This paper introduces binary and multilevel sequence codes to represent amino acids, for use in protein classification. The new binary and multilevel sequences, which are still able to encode information such as hydrophobicity, polarity, and charge, avoid the above disadvantages and increase classification accuracy. Indeed, over a much larger test set than in the original study, parallel cascade models using numerical profiles constructed with the new codes achieved slightly higher two-way classification rates than did hidden Markov models (HMMs) using the primary amino acid sequences, and combining PCI and HMM approaches increased accuracy. © 2000 Biomedical Engineering Society. PAC00: 8714Ee, 8715Cc, 3620Fz, 8715Aa     

Parallel Cascade Recognition of Exon and Intron DNA Sequences

Many of the current procedures for detecting coding regions on human DNA sequences combine a number of individual techniques such as discriminant analysis and neural net methods. Recent papers have used techniques from nonlinear systems identification, in particular, parallel cascade identification (PCI), as one means for classifying protein sequences into their structure/function groups. In the present paper, PCI is used in a pilot study to distinguish exon (coding) from intron (noncoding; interspersed within genes) human DNA sequences. Only the first exon and first intron sequences with known boundaries in genomic DNA from the T-cell receptor locus were used for training. Then, the parallel cascade classifiers were able to achieve classification rates of about 89% on novel sequences in a test set, and averaged about 82% when results of a blind test were included. In testing over a much wider range of human nucleotide sequences, PCI classifiers averaged 83.6% correct classifications. These results indicate that parallel cascade classifiers may be useful components in future coding region detection programs. © 2002 Biomedical Engineering Society. PAC2002: 8715Cc, 8714Gg, 8715Aa     

On the Advantages of Multi-Input Single-Output Parallel Cascade Classifiers

Parallel Cascade Identification (PCI) has been successfully applied to build dynamic nonlinear systems that address diverse challenges in the field of bioinformatics. PCI may be used to identify either single-input single-output (SISO) or multi-input single-output (MISO) models. Although SISO PCI models have typically sufficed, it has been suggested that MISO PCI systems could also be used to form bioinformatics classifiers, and indeed they were successfully applied in one study. This paper reports on the first systematic comparison of MISO and SISO PCI classifiers. Motivation for using the MISO structure is given. The construction of MISO parallel cascade models is also briefly reviewed. In order to compare the accuracy of SISO and MISO PCI classifiers, genetic algorithms are applied to optimize the model architecture on a number of equivalent single-input and multi-input biological training datasets. Through evaluation of both model structures on independent test datasets, we establish that MISO PCI is capable of building classifiers of equal accuracy to those resulting from SISO PCI models. Moreover, we discuss and illustrate the benefits of the MISO approach, including significant reduction in training and testing times, and the ability to adjust automatically the weighting of individual inputs according to information content.     

Post-revascularization Ejection Fraction Prediction for Patients Undergoing Percutaneous Coronary Intervention Based on Myocardial Perfusion SPECT Imaging Radiomics: a Preliminary Machine Learning Study

In this study, the ability of radiomics features extracted from myocardial perfusion imaging with SPECT (MPI-SPECT) was investigated for the prediction of ejection fraction (EF) post-percutaneous coronary intervention (PCI) treatment. A total of 52 patients who had undergone pre-PCI MPI-SPECT were enrolled in this study. After normalization of the images, features were extracted from the left ventricle, initially automatically segmented by k-means and active contour methods, and finally edited and approved by an expert radiologist. More than 1700 2D and 3D radiomics features were extracted from each patient’s scan. A cross-combination of three feature selections and seven classifier methods was implemented. Three classes of no or dis-improvement (class 1), improved EF from 0 to 5% (class 2), and improved EF over 5% (class 3) were predicted by using tenfold cross-validation. Lastly, the models were evaluated based on accuracy, AUC, sensitivity, specificity, precision, and F-score. Neighborhood component analysis (NCA) selected the most predictive feature signatures, including Gabor, first-order, and NGTDM features. Among the classifiers, the best performance was achieved by the fine KNN classifier, which yielded mean accuracy, AUC, sensitivity, specificity, precision, and F-score of 0.84, 0.83, 0.75, 0.87, 0.78, and 0.76, respectively, in 100 iterations of classification, within the 52 patients with 10-fold cross-validation. The MPI-SPECT-based radiomic features are well suited for predicting post-revascularization EF and therefore provide a helpful approach for deciding on the most appropriate treatment.

     

Effect of Altered Matrix Proteins on Quasilinear Viscoelastic Properties in Transgenic Mouse Tail Tendons

Tendons have complex mechanical behaviors that are viscoelastic, nonlinear, and anisotropic. It is widely held that these behaviors are provided for by the tissue's composition and structure. However, little data are available to quantify such structure–function relationships. This study quantified tendon mechanical behaviors, including viscoelasticity and nonlinearity, for groups of mice that were genetically engineered for altered extracellular matrix proteins. Uniaxial tensile stress-relaxation experiments were performed on tail tendon fascicles from the following groups: eight week old decorin knockout, eight week old reduced type I collagen, three week old control, and eight week old control. Data were fit using Fung's quasilinear viscoelastic model, where the model parameters represent the linear viscoelastic and nonlinear elastic response. The viscoelastic properties demonstrated a larger and faster stress relaxation for the decorin knockout and a smaller and slower stress relaxation for the three week control. The elastic parameter, A, in the eight week control group was significantly greater than in the collagen reduction and three week control groups. This study provides quantitative evidence for structure–function relationships in tendon, including the role of proteoglycan in viscoelasticity. Future studies should directly correlate composition and structure with tendon mechanics for the design and evaluation of tissue-engineered constructs or tendon repairs. © 2003 Biomedical Engineering Society. PAC2003: 8719Rr, 8715La, 8714Ee, 8710+e     

Electrical and Morphological Changes of Human Erythrocytes under High Hydrostatic Pressure Followed by Dielectric Spectroscopy

The electrical and morphological properties of human erythrocytes under high hydrostatic pressure up to 500 MPa have been studied by dielectric spectroscopy. The pressure-induced changes in the dielectric behavior of erythrocyte suspensions indicate that hydrostatic pressure causes the change in cell shape from discoidal to spherical and hemolysis at 200–300 MPa, the formation of buds and spicular processes followed by vesiculation at 300–400 MPa, and the increase in the membrane capacitance at 400–500 MPa. © 1999 Biomedical Engineering Society. PAC99: 8716Dg, 8716Uv, 8380Lz, 8715La     

Effect of Stenosis Asymmetry on Blood Flow and Artery Compression: A Three-Dimensional Fluid-Structure Interaction Model

A nonlinear three-dimensional thick-wall model with fluid-structure interactions is introduced to simulate blood flow in carotid arteries with an asymmetric stenosis to quantify the effects of stenosis severity, eccentricity, and pressure conditions on blood flow and artery compression (compressive stress in the wall). Mechanical properties of the tube wall are measured using a thick-wall stenosis model made of polyvinyl alcohal hydrogel whose mechanical properties are close to that of carotid arteries. A hyperelastic Mooney–Rivlin model is used to implement the experimentally measured nonlinear elastic properties of the tube wall. A 36.5% pre-axial stretch is applied to make the simulation physiological. The Navier–Stokes equations in curvilinear form are used for the fluid model. Our results indicate that severe stenosis causes critical flow conditions, high tensile stress, and considerable compressive stress in the stenosis plaque which may be related to artery compression and plaque cap rupture. Stenosis asymmetry leads to higher artery compression, higher shear stress and a larger flow separation region. Computational results are verified by available experimental data. © 2003 Biomedical Engineering Society. PAC2003: 8719Uv, 8710+e     

Assessment of four statistical pattern recognition techniques to assist in obstructive sleep apnoea diagnosis from nocturnal oximetry

The aim of this study is to assess the utility of traditional statistical pattern recognition techniques to help in obstructive sleep apnoea (OSA) diagnosis. Classifiers based on quadratic (QDA) and linear (LDA) discriminant analysis, K-nearest neighbours (KNN) and logistic regression (LR) were evaluated. Spectral and nonlinear input features from oxygen saturation (SaO₂) signals were applied. A total of 187 recordings from patients suspected of suffering from OSA were available. This initial dataset was divided into training set (74 subjects) and test set (113 subjects). Twelve classification algorithms were developed by applying QDA, LDA, KNN and LR with spectral features, nonlinear features and combination of both groups. The performance of each algorithm was measured on the test set by means of classification accuracy and receiver operating characteristic (ROC) analysis. QDA, LDA and LR showed better classification capability than KNN. The classifier based on LDA with spectral features provided the best diagnostic ability with an accuracy of 87.61% (91.05% sensitivity and 82.61% specificity) and an area under the ROC curve (AROC) of 0.925. The proposed statistical pattern recognition techniques could be applied as an OSA screening tool.     

Serial Short-Term Outcomes of Very-Low-Birth-Weight Infants in the Korean Neonatal Network From 2013 to 2020

BackgroundWe aimed to determine the current survival rate and short-term outcomes of very-low-birth-weight infants (VLBWIs) in Korea, as well as whether the survival rate and short-term outcomes have improved over time since 2013, which was when the Korean Neonatal Network (KNN) was launched.MethodsThis study used data from the annual reports of the KNN from 2013 to 2020. A total of 16,351 VLBWIs born at gestational age (GA) ≥ 22 weeks between January 1, 2013, and December 31, 2020, and who were registered in the KNN were enrolled. Serial outcomes were analyzed according to era (2013–14, 2015–16, 2017–18, and 2019–20).ResultsMore mothers delivered by cesarean section, had diabetes or hypertension during their pregnancy, and received antenatal steroids when analyzed by era. Fewer infants were intubated at birth and had air leaks when analyzed by era. The overall survival rate of VLBWIs between 2013 and 2020 was 87%. The rate of respiratory distress syndrome was 77% and that of bronchopulmonary dysplasia was 32% between 2013 and 2020. The rates of intraventricular hemorrhage (grade ≥ 3), periventricular leukomalacia, and sepsis decreased over time. The survival rate of infants with a GA of 26 weeks has improved serially according to era.ConclusionSince the launch of the KNN in 2013, the survival rates of infants with GA 26 weeks and short-term outcomes have improved, which implies a quality improvement in antenatal and delivery room care. Further studies on the long-term neurodevelopmental outcomes of these KNN registrants are warranted.     

Resting muscle glucose metabolites and related compounds in hypercapnia

Summary Glucose metabolites (lactate, pyruvate, citrate, malate), alanine, glutamate and adenosine triphosphate (ATP) were determined in the resting anterior tibial muscle of dogs. The muscle was sampled in anesthetized animals first breathing air, and secondly after an hour of breathing a hypercapnic mixture, FICO₂=0.10 (experimental subjects n=6) or air (control subjects n=6). A decrease in concentration of glucose metabolites (lactate: –34%; pyruvate: –24%; Citrate: –34%; malate: –54%), glutamate (–43%), alanine (–35%) and ATP (–8%) was observed in the resting muscle during acute hypercapnic acidosis. This was not the case in control animals breathing air.     

Homotypic interactions of soluble and immobilized osteopontin

Osteopontin (OPN) is an extracellular matrix protein found in bones and teeth, where it accumulates at matrix–matrix interfaces. We postulate that OPN interacts homotypically and heterotypically in the adhesion of apposing matrices. Using suspensions of OPN-coupled aldehyde/sulfate latex spheres, we measured the strength of homotypic OPN–OPN binding in vitro. Doublets formed through shear-induced collisions in a cone and plate rheoscope were subjected to shear stresses >0.6 N m^-2 and the fraction broken up determined over 60 s. Rapid initial breakup of 35% of doublets was followed by very slow breakup of the remaining 65%. Monte Carlo simulation of the breakup kinetics pointed to the existence of low and high bond strength populations of doublets. Dynamic light scattering spectroscopy of soluble OPN showed that 27% by mass existed as dimers. We postulate that OPN dimers binding to monomers account for the low strength bonds since a strong bond has already formed between the molecules of the dimer. In contrast, OPN–OPN monomer bonds had higher tensile strength than bonds between the high-affinity interaction of IgG and protein G, previously studied. Antibody blocking studies showed that the self-binding region of OPN resides in the C-terminus. These data suggest that homotypic OPN–OPN bonds have physiologically significant strength, supporting the hypothesis that OPN–OPN binding and self-assembly participate in adhesion within mineralized tissues. © 2002 Biomedical Engineering Society. PAC2002: 8714Ee, 8719Tt, 0250Ng, 8715Mi     

Using Uniaxial Pseudorandom Stress Stimuli to Develop Soft Tissue Constitutive Equations

A nonlinear systems identification method was used to develop constitutive equations for soft tissue specimens under uniaxial tension. The constitutive equations are developed from a single test by applying a pseudorandom Gaussian (PGN) stress input to the specimen, measuring the resulting strain, and calculating the Volterra–Wiener kernels. First and second order kernels were developed for two tissues with widely different properties, rat medial collateral knee ligaments, and rat skin. These kernels were used to predict the strain response to a variety of sinusoidal stress inputs. These predicted strains were compared with the measured strain response using the normalized mean squared error (NMSE). Results showed NMSEs in the range of 0.01–0.08 provided that the magnitudes of the applied stresses were present in the original PGN stress input. Overall, the method provides a means to develop soft tissue constitutive equations that can predict both nonlinear and viscoelastic behavior over a wide range of stress inputs. © 2002 Biomedical Engineering Society. PAC02: 8719Rr     

Radial basis function classifiers to help in the diagnosis of the obstructive sleep apnoea syndrome from nocturnal oximetry

The aim of this study is to assess the ability of radial basis function (RBF) classifiers as an assistant tool for the diagnosis of the obstructive sleep apnoea syndrome (OSAS). A total of 187 subjects suspected of suffering from OSAS were available for our research. The initial population was divided into training, validation and test sets for deriving and testing our neural classifiers. We used nonlinear features from nocturnal oxygen saturation (SaO₂) to perform patients’ classification. We evaluated three different RBF construction techniques based on the following algorithms: k-means (KM), fuzzy c-means (FCM) and orthogonal least squares (OLS). A diagnostic accuracy of 86.1, 84.7 and 85.5% was provided by the networks developed with KM, FCM and OLS, respectively. The three proposed networks achieved an area under the receiver operating characteristic (ROC) curve over 0.90. Our results showed that a useful non-invasive method could be applied to diagnose OSAS from nonlinear features of SaO₂ with RBF classifiers.     

A new dynamometer measuring concentric and eccentric muscle strength in accelerated,decelerated, or isokinetic movements

Summary A new computerized dynamometer (the SPARK System) is described. The system can measure concentric and eccentric muscle strength (torque) during linear or nonlinear acceleration or deceleration, isokinetic movements up to 400° · s^–1, and isometric torque. Studies were performed to assess: I. validity and reproducibility of torque measurements; II. control of lever arm position; III. control of different velocity patterns; IV. control of velocity during subject testing; and, V. intra-individual reproducibility. No significant difference was found between torque values computed by the system and known torque values (p>0.05). No difference was present between programmed and external measurement of the lever arm position. Accelerating, decelerating and isokinetic velocity patterns were highly reproducible, with differences in elapsed time among 10 trials being never greater than 0.001 s. Velocity during concentric and eccentric isokinetic quadriceps contractions at 30° · s^–1, 120° · s^–1 and 270° · s^–1 never varied by more than 3° · s^–1 among subjects (N=21). Over three days of testing, the overall error for concentric and eccentric quadriceps contraction peak torque values for 5 angular velocities between 30° · s^–1 and 270° · s^–1 ranged from 5.8% to 9.0% and 5.8% to 9.6% respectively (N=25). The results indicate that the SPARK System provides valid and reproducible torque measurements and strict control of velocity. In addition, the intra-individual error is in accordance with those reported for other similar devices.     

The accuracy of volume estimates using ultrasound muscle thickness measurements in different muscle groups

This study aimed to investigate the accuracy of estimating the volume of limb muscles (MV) using ultrasonographic muscle thickness (MT) measurements. The MT and MV of each of elbow flexors and extensors, knee extensors and ankle plantar flexors were determined from a single ultrasonographic image and multiple magnetic resonance imaging (MRI) scans, respectively, in 27 healthy men (23–40 years of age) who were allocated to validation (n=14) and cross-validation groups (n=13). In the validation group, simple and multiple regression equations using MT and a set of MT and limb length, respectively, as independent variables were derived to estimate the MV measured by MRI. However, only the multiple regression equations were cross-validated, and so the prediction equations with r ² of 0.787–0.884 and the standard error of estimate of 22.1 cm³ (7.3%) for the elbow flexors to 198.5 cm³ (11.1%) for the knee extensors were developed using the pooled data. This approach did not induce significant systematic error in any muscle group, with no significant difference in the accuracy of estimating MV between muscle groups. In the multiple regression equations, the relative contribution of MT for predicting MV varied from 41.9% for the knee extensors to 70.4% for the elbow flexors. Thus, ultrasonographic MT measurement was a good predictor of MV when combined with limb length. For predicting MV, however, the unsuitability of a simple equation using MT only and the difference between muscle groups in the relative contribution of MT in multiple regression equations indicated a need for further research on the limb site selected and muscle analyzed for MT measurement.     

Energy source for transepithelial sodium transport in rat renal proximal tubules

Summary The effects of various metabolic inhibitors on isotonic fluid absorption (J _V ) in rat proximal tubules and on the Na⁺–K⁺-ATPase of isolated cell membranes of rat kidney cortex were investigated by the shrinking split oil droplet technique and biochemical methods respectively.Both Oligomycin (5×10^–5 M, 10^–4 M) and Antimycin A (10^–5 M, 10^–4 M) inhibited isotonic fluid absorption by 80% when applied intratubularly but only in conjunction with bovine serum albumin. At these concentrations they inhibited a Na⁺–K⁺ activated adenosine triphosphate phosphohydrolase (Na⁺–K⁺ ATPase E.C. 3.6.1.3.) of cell membranes isolated from rat kidney cortex by 77%, 82% and 55%, 95%, respectively.Sodium phosphoenolpyruvate (PEP) 5×10^–3 M could partially reverse the inhibition of the isotonic fluid absorption but only with 10^–5 M Antimycin A when the Na⁺–K⁺ ATPase inhibition was apparently small.The uncoupler, carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (10^–3 M), as well as sodium cyanide (5×10^–3 M) inhibitedJ _V 100%, but only when applied through peritubular blood capillary perfusion.From these findings it was concluded thatall proximal tubular isotonic fluid absorption is supported by energy fromoxidative processes, and that in a least 80% of this sodium reabsorption, ATP from oxidative phosphorylation is directly involved, while, for the remaining 20% non ATP energy is responsible.C. J. Martin Fellow of the National Health and Medical Research Council of Australia.     

Optimal feature selection for the assessment of vocal fold disorders

Unilateral vocal fold paralysis, vocal fold polyp, and vocal fold nodules are the most common types of neurogenic and organic vocal disorders. This article aims to distinguish these types of vocal diseases into four different classes for the purpose of automatic screening. Firstly, the reconstructed signal at each wavelet packet decomposition sub-band in five levels of decomposition with mother wavelet of (db10) is used to extract the nonlinear features of self-similarity and approximate entropy. Also, wavelet packet coefficients are used to measure energy and Shannon entropy features at different spectral sub-bands. Consequently, to find a discriminant feature vector, three different methods have been applied: Davies-Bouldin (DB) criteria, genetic algorithm (GA) with the fitness functions of support vector machine's (SVM) and k-nearest neighbor's (KNN) recognition rates. Finally, obtained feature vectors have been passed on to SVM and KNN classifiers. The results show that a feature vector of length 12 obtained by the optimization method of GA with the fitness function of SVM's recognition rate fed to SVM classifier achieves the highest classification accuracy of 91%. Furthermore, nonlinear features play an important role in pathological voice classification by participating rate of approximately 67% in the optimal feature vector.     

基于特征融合的低倍镜下大鼠骨质疏松识别

低倍镜下纹理特征不清晰,并且相较于高倍镜更易受到方向和距离的影响。为此我们提出改进的纹理特征计算方法,结合形状分析方法,进行特征级融合以提高骨质疏松识别准确率。通过对纹理参数在方向和距离上的分析,发现相关和短游程矩对方向非常敏感,并且距离在等于3时各参数变化趋于稳定。基于变异系数法利用相关和短游程矩计算各个方向权重系数,使用距离等于3时的纹理数据得到最终纹理参数结果,融合形状参数,用线性支持向量机、K-最近邻分类算法和线性判别分析方法进行分类识别。采用加权纹理参数比常规未加权的纹理参数识别准确率高,同时融合了形状参数后识别准确率比仅用纹理参数高。线性判别分析识别率最高达到了92.3%。采用加权纹理融合形状参数的方法识别准确率显著提高,具有诊断应用价值。     

Mathematical Model of Diffusion-Limited Evolution of Multiple Gas Bubbles in Tissue

Models of gas bubble dynamics employed in probabilistic analyses of decompression sickness incidence in man must be theoretically consistent and simple, if they are to yield useful results without requiring excessive computations. They are generally formulated in terms of ordinary differential equations that describe diffusion-limited gas exchange between a gas bubble and the extravascular tissue surrounding it. In our previous model (Ann. Biomed. Eng. 30: 232–246, 2002), we showed that with appropriate representation of sink pressures to account for gas loss or gain due to heterogeneous blood perfusion in the unstirred diffusion region around the bubble, diffusion-limited bubble growth in a tissue of finite volume can be simulated without postulating a boundary layer across which gas flux is discontinuous. However, interactions between two or more bubbles caused by competition for available gas cannot be considered in this model, because the diffusion region has a fixed volume with zero gas flux at its outer boundary. The present work extends the previous model to accommodate interactions among multiple bubbles by allowing the diffusion region volume of each bubble to vary during bubble evolution. For given decompression and tissue volume, bubble growth is sustained only if the bubble number density is below a certain maximum. © 2003 Biomedical Engineering Society. PAC2003: 8719Uv, 8710+e, 8715Vv, 8719Xx