A novel approach for estimating initial sound level for speech reception threshold test.

BACKGROUND: The speech reception threshold (SRT), synonymous with the speech recognition threshold, denotes the minimum hearing level required for an individual to discern 50% of presented speech material. This threshold is measured independently in each ear with a repetitive up-down adjustment of stimulus level starting from the initial SRT value derived from pure tone thresholds (PTTs), measured via pure-tone audiometry (PTA). However, repetitive adjustments in the test contributes to increased fatigue for both patients and audiologists, compromising the reliability of the hearing tests. OBJECTIVE: Determining the first (initial) sound level closer to the finally determined SRT value, is important to reduce the number of repetitions. The existing method to determine the initial sound level is to average the PTTs called pure tone average (PTAv). METHODS: We propose a novel method using a machine learning approach to estimate a more optimal initial sound level for the SRT test. Specifically, a convolutional neural network with 1-dimensional filters (1D CNN) was implemented to predict a superior initial level than the conventional methods. RESULTS: Our approach produced a reduction of 37.92% in the difference between the initial stimulus level and the final SRT value. CONCLUSIONS: This outcome substantiates that our approach can reduce the repetitions for finding the final SRT, and, as the result, the hearing test time can be reduced.

Sleep snoring detection using multi-layer neural networks.

Snoring detection is important for diagnosing obstructive sleep apnea syndrome (OSAS) and other respiratory sleep disorders. In general, audio signal processing such as snoring sound analysis uses the frequency characteristics of the signal. Recently, a correlational filter Multilayer Perceptron neural network (f-MLP) has been proposed, which has the first hidden layer of correlational filter operations in frequency domain. It demonstrated a superior classification performance for the pattern sets; of these, frequency information is the dominant feature for classification. The first hidden layer is implemented with the correlational filter operation; its output is the power spectrum of the filter output, while the other layers are the same as the ordinary multilayer Perceptron (o-MLP). By using the back-propagation learning algorithm for the correlational filter layer, f-MLP was able to self-adapt the filter coefficients to produce its output with more discrimination power for classification in the higher layer. In this research, this f-MLP was applied for sleep snoring signal detection. As a result, the f-MLP achieved an average detection rate of 96% for the test patterns, compared to the conventional multilayer neural network that demonstrates an 82% average detection rate.

Hematocrit estimation using online sequential extreme learning machine.

Hematocrit is a blood test that is defined as the volume percentage of red blood cells in the whole blood. It is one of the important indicators for clinical decision making and the most effective factor in glucose measurement using handheld devices. In this paper, a method for hematocrit estimation that is based upon the transduced current curve and the neural network is presented. The salient points of this method are that (1) the neural network is trained by the online sequential extreme learning machine (OS-ELM) in which the devices can be still trained with new samples during the using process and (2) the extended features are used to reduce the number of current points which can save the battery power of devices and speed up the measurement process.

Influence of pelvic asymmetry and idiopathic scoliosis in adolescents on postural balance during sitting.

The effects of pelvic asymmetry and idiopathic scoliosis on postural balance during sitting were studied by measuring inclination angles, pressure distribution, and electromyography. Participants were classified into a control group, pelvic asymmetry group, scoliosis group, and scoliosis with pelvic asymmetry and then performed anterior, posterior, left, and right pelvic tilting while sitting on the unstable board for 5 seconds to assess their postural balance. Inclination and obliquity angles between the groups were measured by an accelerometer located on the unstable board. Pressure distribution (maximum force and peak pressure) was analyzed using a capacitive seat sensor. In addition, surface electrodes were attached to the abdominal and erector spinae muscles of each participant. Inclination and obliquity angles increased more asymmetrically in participants with both pelvic asymmetry and scoliosis than with pelvic asymmetry or scoliosis alone. Maximum forces and peak pressures of each group showed an asymmetrical pressure distribution caused by the difference in height between the left and right pelvis and curve type of the patients' spines when performing anterior, posterior, left, and right pelvic tilting while sitting. Muscle contraction patterns of external oblique, thoracic erector spinae, lumbar erector spinae, and lumbar multifidus muscles may be influenced by spine curve type and region of idiopathic scoliosis. Asymmetrical muscle activities were observed on the convex side of scoliotic patients and these muscle activity patterns were changed by the pelvic asymmetry. From these results, it was confirmed that pelvic asymmetry and idiopathic scoliosis cause postural asymmetry, unequal weight distribution, and muscular imbalance during sitting.

Effects of custom-made insoles on idiopathic pes cavus foot during walking.

From a subject group of pes cavus, the purpose of this study was to evaluate the biomechanical characteristics of lower limbs, based on plantar foot pressure and electromyography (EMG) activities, by the effects on two kind of custom-made insoles. Ten individuals among thirty females with a clinical diagnosis of idiopathic pes cavus (mean age (SD): 22.3 (0.08) years) were selected for the study. The plantar foot pressure data and EMG activities of four lower limb muscles were collected, when subjects walked on a treadmill, under three different experimental conditions. The plantar foot pressure data was analyzed, after the bilateral foot was divided into three areas of masks and into four sections of stance phase, to compare plantar foot pressure. The EMG activities were analyzed for integrated EMG (IEMG) value. The results show that plantar foot pressure concentrated in particular parts is decreased by custom-made insoles. In the case of EMG, all the muscle activities decreased significantly. The custom-made insoles dispersed pressure concentrated by the higher medial longitudinal arch and improved the efficient use of muscles. In particular, the extension structure in the forefoot of custom-made insoles was more efficient for pes cavus. Therefore, it could help patients to walk, by offering support to prevent the disease of pes cavus deformity, and to relieve the burden and fatigue in the lower limbs on gait.

Efficient snoring and breathing detection based on sub-band spectral statistics.

Snoring, a common symptom in the general population may indicate the presence of obstructive sleep apnea (OSA). In order to detect snoring events in sleep sound recordings, a novel method was proposed in this paper. The proposed method operates by analyzing the acoustic characteristics of the snoring sounds. Based on these acoustic properties, the feature vectors are obtained using the mean and standard deviation of the sub-band spectral energy. A support vector machine is then applied to perform the frame-based classification procedure. This method was demonstrated experimentally to be effective for snoring detection. The database for detection included full-night audio recordings from four individuals who acknowledged having snoring habits. The performance of the proposed method was evaluated by classifying different events (snoring, breathing and silence) from the sleep sound recordings and comparing the classification against ground truth. The proposed algorithm was able to achieve an accuracy of 99.61% for detecting snoring events, 99.16% for breathing, and 99.55% for silence.

Reduction of PDC1 expression in S. cerevisiae with xylose isomerase on xylose medium.

Ethanol production using hemicelluloses has recently become a focus of many researchers. In order to promote D: -xylose fermentation, we cloned the bacterial xylA gene encoding for xylose isomerase with 434 amino acid residues from Agrobacterium tumefaciens, and successfully expressed it in Saccharomyces cerevisiae, a non-xylose assimilating yeast. The recombinant strain S. cerevisiae W303-1A/pAGROXI successfully colonized a minimal medium containing D: -xylose as a sole carbon source and was capable of growth in minimal medium containing 2% xylose via aerobic shake cultivation. Although the recombinant strain assimilates D: -xylose, its ethanol productivity is quite low during fermentation with D: -xylose alone. In order to ascertain the key enzyme in ethanol production from D: -xylose, we checked the expression levels of the gene clusters involved in the xylose assimilating pathway. Among the genes classified into four groups by their expression patterns, the mRNA level of pyruvate decarboxylase (PDC1) was reduced dramatically in xylose media. This reduced expression of PDC1, an enzyme which converts pyruvate to acetaldehyde, may cause low ethanol productivity in xylose medium. Thus, the enhancement of PDC1 gene expression may provide us with a useful tool for the fermentation of ethanol from hemicellulose.

Performance comparison of SLFN training algorithms for DNA microarray classification.

The classification of biological samples measured by DNA microarrays has been a major topic of interest in the last decade, and several approaches to this topic have been investigated. However, till now, classifying the high-dimensional data of microarrays still presents a challenge to researchers. In this chapter, we focus on evaluating the performance of the training algorithms of the single hidden layer feedforward neural networks (SLFNs) to classify DNA microarrays. The training algorithms consist of backpropagation (BP), extreme learning machine (ELM) and regularized least squares ELM (RLS-ELM), and an effective algorithm called neural-SVD has recently been proposed. We also compare the performance of the neural network approaches with popular classifiers such as support vector machine (SVM), principle component analysis (PCA) and fisher discriminant analysis (FDA).

Gene expression profiling in mouse liver infected with Clonorchis sinensis metacercariae.

Clonorchis sinensis, the parasite that causes clonorchiasis, is endemic in many Asian countries, and infection with the organism drives changes in the liver tissues of the host. However, information regarding the molecular events in clonorchiasis remains limited, and little is currently known about host-pathogen interactions in clonorchiasis. In this study, we assessed the gene expression profiles in mice livers via DNA microarray analysis 1, 2, 4, and 6 weeks after induced metacercariae infection. Functional clustering of the gene expression profile showed that the immunity-involved genes were induced in the livers of the mice at the early stage of metacercariae infection, whereas immune responses were reduced in the 6-week liver tissues after infection in which the metacercariae became adult flukes. Many genes involved in fatty acid metabolism, including Peci, Cyp4a10, Acat1, Ehhadh, Gcdh, and Cyp2 family were downregulated in the infected livers. On the other hand, the liver tissues infected with the parasite expressed Wnt signaling molecules such as Wnt7b, Fzd6, and Pdgfrb and cell cycle-regulating genes including cyclin-D1, Cdca3, and Bcl3. These investigations constitute an excellent starting point for increased understanding of the molecular mechanisms underlying host-pathogen interaction during the development of C. sinensis in the host liver.

An improvement of extreme learning machine for compact single-hidden-layer feedforward neural networks.

Recently, a novel learning algorithm called extreme learning machine (ELM) was proposed for efficiently training single-hidden-layer feedforward neural networks (SLFNs). It was much faster than the traditional gradient-descent-based learning algorithms due to the analytical determination of output weights with the random choice of input weights and hidden layer biases. However, this algorithm often requires a large number of hidden units and thus slowly responds to new observations. Evolutionary extreme learning machine (E-ELM) was proposed to overcome this problem; it used the differential evolution algorithm to select the input weights and hidden layer biases. However, this algorithm required much time for searching optimal parameters with iterative processes and was not suitable for data sets with a large number of input features. In this paper, a new approach for training SLFNs is proposed, in which the input weights and biases of hidden units are determined based on a fast regularized least-squares scheme. Experimental results for many real applications with both small and large number of input features show that our proposed approach can achieve good generalization performance with much more compact networks and extremely high speed for both learning and testing.

Pattern analysis of serum proteome distinguishes renal cell carcinoma from other urologic diseases and healthy persons.

Despite having a relatively low incidence, renal cell carcinoma (RCC) is one of the most lethal urologic cancers. For successful treatment including surgery, early detection is essential. Currently there is no screening method such as biomarker assays for early diagnosis of RCC. Surface-enhanced laser desorption/ionization-time of flight mass spectrometry (SELDI-TOF) is a recent technical advance that can be used to identify biomarkers for cancers. In this study, we investigated whether SELDI protein profiling and artificial intelligence analysis of serum could distinguish RCC from healthy persons and other urologic diseases (nonRCC). The SELDI-TOF data was acquired from a total of 36 serum samples with weak cation exchange-2 protein chip arrays and filtered using ProteinChip software. We used a decision tree algorithm c4.5 to classify the three groups of sera. Five proteins were identified with masses of 3900, 4107, 4153, 5352, and 5987 Da. These biomarkers can correctly separate RCC from healthy and nonRCC samples.