基于模糊PID的睡眠呼吸机控制系统设计

持续气道正压(Continuous Positive Airway Pressure,CPAP)通气是目前治疗阻塞性睡眠呼吸暂停(Obstructive sleep apnea,OSA)最为有效的方式之一。但在实际应用中,由于受到患者自主呼吸的影响,使得气道压力很难保持稳定。为了降低患者自主呼吸对设定压力的干扰,以及消除患者呼气时的憋闷感,模糊PID控制方法被应用于睡眠呼吸机CPAP的压力控制上。本文介绍了硬件系统结构并通过查询模糊规则表的方法实现了Fuzzy PID的算法,最后,使用了气体流量分析仪(VT PLUS HF)对治疗时的压力曲线进行了测试。结果表明,压力的波动性满足了睡眠呼吸暂停治疗设备的相关标准。     

An expert system for automated recognition of patients with obstructive sleep apnea using electrocardiogram recordings

Obstructive sleep apnea (OSA) is a highly prevalent sleep disorder. The traditional diagnosis methods of the disorder are cumbersome and expensive. The ability to automatically identify OSA from electrocardiogram (ECG) recordings is important for clinical diagnosis and treatment. In this study, we proposed an expert system based on discrete wavelet transform (DWT), fast-Fourier transform (FFT) and least squares support vector machine (LS-SVM) for the automatic recognition of patients with OSA from nocturnal ECG recordings. Thirty ECG recordings collected from normal subjects and subjects with sleep apnea, each of approximately 8 h in duration, were used throughout the study. The proposed OSA recognition system comprises three stages. In the first stage, an algorithm based on DWT was used to analyze ECG recordings for the detection of heart rate variability (HRV) and ECG-derived respiration (EDR) changes. In the second stage, an FFT based power spectral density (PSD) method was used for feature extraction from HRV and EDR changes. Then, a hill-climbing feature selection algorithm was used to identify the best features that improve classification performance. In the third stage, the obtained features were used as input patterns of the LS-SVM classifier. Using the cross-validation method, the accuracy of the developed system was found to be 100% for using a subset of selected combination of HRV and EDR features. The results confirmed that the proposed expert system has potential for recognition of patients with suspected OSA by using ECG recordings.     

基于X线的成人OSA计算机辅助诊断综述

阻塞性睡眠呼吸暂停（Obstructive Sleep Apnea,OSA）是成年人较为常见的呼吸类疾病之一,该疾病的特点是睡眠过程中频繁出现上气道完全或部分塌陷,严重影响人们的睡眠质量以及身体健康。阻塞性睡眠呼吸暂停综合征的诊断主要依靠多导睡眠监测,但这种方法无法满足目前大量的诊断需求。随着人工智能的出现及发展,假设深度学习可以有效地协助医生进行诊断该综合征。主要从阻塞性睡眠呼吸暂停的临床诊断方式出发,介绍了颅面侧位片作为诊断数据集的优势,以及人工智能诊断OSA的现状,提出了人工智能辅助医师诊断OSA的技术路线,分析了目前该诊断系统仍然存在的问题和挑战。     

Detection of apneic events from single channel nasal airflow using 2nd derivative method

Detection of sleep apnea is one of the major tasks in sleep studies. Several methods, analyzing the various features of bio-signals, have been applied for automatic detection of sleep apnea, but it is still required to detect apneic events efficiently and robustly from a single nasal airflow signal under varying situations. This study introduces a new algorithm that analyzes the nasal airflow (NAF) for the detection of obstructive apneic events. It is based on mean magnitude of the second derivatives (MMSD) of NAF, which can detect respiration strength robustly under offset or baseline drift. Normal breathing epochs are extracted automatically by examining the stability of SaO(2) and NAF regularity for each subject. The standard MMSD and period of NAF, which are regarded as the values at the normal respiration level, are determined from the normal breathing epochs. In this study, 24 Polysomnography (PSG) recordings diagnosed as obstructive sleep apnea (OSA) syndrome were analyzed. By analyzing the mean performance of the algorithm in a training set consisting of three PSG recordings, apnea threshold is determined to be 13% of the normal MMSD of NAF. NAF signal was divided into 1-s segments for analysis. Each segment is compared with the apnea threshold and classified into apnea events if the segment is included in a group of apnea segments and the group satisfies the time limitation. The suggested algorithm was applied to a test set consisting of the other 21 PSG recordings. Performance of the algorithm was evaluated by comparing the results with the sleep specialist's manual scoring on the same record. The overall agreement rate between the two was 92.0% (kappa=0.78). Considering its simplicity and lower computational load, the suggested algorithm is found to be robust and useful. It is expected to assist sleep specialists to read PSG more quickly and will be useful for ambulatory monitoring of apneas using airflow signals.     

Automatic detection of respiratory arrests in OSA patients using PPG and machine learning techniques

Obstructive sleep apnea is a syndrome which is characterized by the decrease in air flow or respiratory arrest depending on upper respiratory tract obstructions recurring during sleep and often observed with the decrease in the oxygen saturation. The aim of this study was to determine the connection between the respiratory arrests and the photoplethysmography (PPG) signal in obstructive sleep apnea patients. Determination of this connection is important for the suggestion of using a new signal in diagnosis of the disease. Thirty-four time-domain features were extracted from the PPG signal in the study. The relation between these features and respiratory arrests was statistically investigated. The Mann–Whitney U test was applied to reveal whether this relation was incidental or statistically significant, and 32 out of 34 features were found statistically significant. After this stage, the features of the PPG signal were classified with k-nearest neighbors classification algorithm, radial basis function neural network, probabilistic neural network, multilayer feedforward neural network (MLFFNN) and ensemble classification method. The output of the classifiers was considered as apnea and control (normal). When the classifier results were compared, the best performance was obtained with MLFFNN. Test accuracy rate is 97.07 % and kappa value is 0.93 for MLFFNN. It has been concluded with the results obtained that respiratory arrests can be recognized through the PPG signal and the PPG signal can be used for the diagnosis of OSA.

     

A new approach for estimation of obstructive sleep apnea syndrome

Obstructive sleep apnea syndrome (OSAS) is a situation where repeatedly upper airway stops off while the respiratory effort continues during sleep at least for 10 s. Apart from polysomnography, many researchers have concentrated on exploring alternative methods for OSAS detection. However, not much work has been done on using non-Gaussian and nonlinear behavior of the electroencephalogram (EEG) signals. Bispectral analysis is an advanced signal processing technique particularly used for exhibiting quadratic phase-coupling that may arise between signal components with different frequencies. From this perspective, in this study, a new technique for recognizing patients with OSAS was introduced using bispectral characteristics of EEG signal and an artificial neural network (ANN). The amount of Quadratic phase coupling (QPC) in each subband of EEG (namely; delta, theta, alpha, beta and gamma) was calculated over bispectral density of EEG. Then, these QPCs were fed to the input of the designed ANN. The neural network was configured with two outputs: one for OSAS and one for estimation of normal situation. With this technique a global accuracy of 96.15% was achieved. The proposed technique could be used in designing automatic OSAS identification systems which will improve medical service.     

Artificial neural network and dataset optimization for implementation of linear system models in resource-constrained embedded systems

On-board training of artificial neural network (ANN) is important in instances where real time data are required for model training. Provision of on-board intelligence enables the developed systems to self-recalibrate and enhances their efficiencies. In this work, investigations have been performed to determine optimized parameters of ANN model for linear systems. The performance parameters that is, model parameters, memory requirements, accuracy and processing time are chosen by considering the model to be installed on commercially available microcontrollers that have very limited on-board memory. Minimum data requirements for training ANN models of linear systems are also explored for better performance. All dataset ranges are normalized in order to exclude the effects of range differences. It is shown that for linear systems, 1–3–1 architecture produces best results against ≤100 data points when Bayesian Regularization (BR) training function is used along with Log Sigmoid Activation function. Simulations for 1–3–1 architecture are then performed for datasets having 10, 25, 50 and 100 data points. The results show that training with 25 data points produces over-all better performance than other datasets. A large dataset utilizes more training time and memory whereas a smaller dataset produces relatively lesser accuracy. The effects of clustered data and uniformly distributed data are also explored. It is found that total epochs in case of clustered data are significantly higher than uniformly distributed data. The combination of these optimized parameters that is, 1–3–1 architecture, with BR and Log function, for ≤100 data points can be used for the development and implementation of linear components or systems in resource-constrained embedded systems.     

Classification of sleep stages using class-dependent sequential feature selection and artificial neural network

Several studies have been conducted for automatic classification of sleep stages to ease time-consuming manual scoring process that can involve a high degree of experience and subjectivity. But none of them has found a practical usage in medical area so far because of their under acceptable success rates. In this study, a different classification scheme is proposed to increase the success rate in automatic sleep stage scoring in which sleep stages were classified as Awake, Non-REM1, Non-REM2, Non-REM3 and REM stages. Using EEG, EMG and EOG recordings of five healthy subjects, a modified version of sequential feature selection method was applied to the sleep epochs in class by class basis and different artificial neural network (ANN) architectures were trained for each class. That is to say, sleep stages were classified with five ANN architectures each of which uses different features and different network parameters for classification. The highest classification accuracy was obtained for REM sleep as 95.13 % in addition to the lowest classification accuracy of 86.42 % for Non-REM3 sleep. The overall accuracy, on the other hand, was recorded as 90.93 %, which is a comparatively good result when the other studies using all stages are taken into account.     

Automated detection of cyclic alternating pattern and classification of sleep stages using deep neural network

The visual sleep stages scoring by human experts is the current gold standard for sleep analysis. However, this method is tedious, time-consuming, prone to human errors, and unable to detect microstructure of sleep such as cyclic alternating pattern (CAP) which is an important diagnostic factor for the detection of sleep disorders such as insomnia and obstructive sleep apnea (OSA). The CAP is only observed as subtle changes in the electroencephalogram (EEG) signals during non-rapid eye movement (NREM) sleep, making it very difficult for human experts to discern. Hence, it is important to have an automated system developed using artificial intelligence for accurate and robust detection of CAP and sleep stages classification. In this study, a deep learning model based on 1-dimensional convolutional neural network (1D-CNN) is proposed for CAP detection and homogenous 3-class sleep stages classification, namely wakefulness (W), rapid eye movement (REM) and NREM sleep. The proposed model is developed using standardized EEG recordings. Our developed CNN network achieved good model performance for 3-class sleep stages classification with a classification accuracy of 90.46%. Our proposed model also yielded a classification accuracy of 73.64% using balanced CAP dataset, and sensitivity of 92.06% with unbalanced CAP dataset. Our proposed model correctly identified majority of A-phases which comprised of only 12.6% in the unbalanced dataset. The performance of the developed prototype is ready to be tested with more data before clinical application.

     

Detection and classification of sleep apnea using genetic algorithms and SVM‐based classification of thoracic respiratory effort and oximetric signal features

Sleep apnea is a relatively prevalent breathing disorder characterized by temporary interruptions in airflow during sleep. There are 2 major types of sleep apnea. Obstructive sleep apnea occurs when air cannot flow through the upper airway despite efforts to breathe. Central sleep apnea occurs when the brain fails to signal to the muscles to maintain breathing. The standard diagnostic test is polysomnography, which is expensive and time consuming. The aim of this study was to design an automatic diagnostic and classifying algorithm for sleep apneas employing thoracic respiratory effort and oximetric signals. This algorithm was trained and tested applying a database of 54 subjects who had undergone polysomnography. A feature extraction stage was conducted to compute features. An optimal genetic algorithm was applied to select optimal features of these 2 kinds of signals. The classification technique was based on the support vector machine classifier to classify the selected features in 3 classes as healthy, obstructive, and central sleep apnea events. The results show that our automated classification algorithm can diagnose sleep apnea and its types with an average accuracy level of 90.2% (87.5‐95.8) in the test set and 90.9% in the validation set with high acceptable accuracy.     

Statistical shape model for manifold regularization: Gleason grading of prostate histology

Gleason patterns of prostate cancer histopathology, characterized primarily by morphological and architectural attributes of histological structures (glands and nuclei), have been found to be highly correlated with disease aggressiveness and patient outcome. Gleason patterns 4 and 5 are highly correlated with more aggressive disease and poorer patient outcome, while Gleason patterns 1–3 tend to reflect more favorable patient outcome. Because Gleason grading is done manually by a pathologist visually examining glass (or digital) slides subtle morphologic and architectural differences of histological attributes, in addition to other factors, may result in grading errors and hence cause high inter-observer variability. Recently some researchers have proposed computerized decision support systems to automatically grade Gleason patterns by using features pertaining to nuclear architecture, gland morphology, as well as tissue texture. Automated characterization of gland morphology has been shown to distinguish between intermediate Gleason patterns 3 and 4 with high accuracy. Manifold learning (ML) schemes attempt to generate a low dimensional manifold representation of a higher dimensional feature space while simultaneously preserving nonlinear relationships between object instances. Classification can then be performed in the low dimensional space with high accuracy. However ML is sensitive to the samples contained in the dataset; changes in the dataset may alter the manifold structure. In this paper we present a manifold regularization technique to constrain the low dimensional manifold to a specific range of possible manifold shapes, the range being determined via a statistical shape model of manifolds (SSMM). In this work we demonstrate applications of the SSMM in (1) identifying samples on the manifold which contain noise, defined as those samples which deviate from the SSMM, and (2) accurate out-of-sample extrapolation (OSE) of newly acquired samples onto a manifold constrained by the SSMM. We demonstrate these applications of the SSMM in the context of distinguish between Gleason patterns 3 and 4 using glandular morphologic features in a prostate histopathology dataset of 58 patient studies. Identifying and eliminating noisy samples from the manifold via the SSMM results in a statistically significant improvement in area under the receiver operator characteristic curve (AUC), 0.832 ± 0.048 with removal of noisy samples compared to a AUC of 0.779 ± 0.075 without removal of samples. The use of the SSMM for OSE of newly acquired glands also shows statistically significant improvement in AUC, 0.834 ± 0.051 with the SSMM compared to 0.779 ± 0.054 without the SSMM. Similar results were observed for the synthetic Swiss Roll and Helix datasets.     

Effect of prescribed sleep surfaces on back pain and sleep quality in patients diagnosed with low back and shoulder pain

The purpose of this study was to assess sleep quality and comfort of participants diagnosed with low back pain and stiffness following sleep on individually prescribed mattresses based on dominant sleeping positions. Subjects consisted of 27 patients (females, n = 14; males, n = 13; age 44.8 yrs ± SD 14.6, weight 174 lb. ±SD 39.6, height 68.3 in. ± SD 3.7) referred by chiropractic physicians for the study. For the baseline (pretest) data subjects recorded back and shoulder discomfort, sleep quality and comfort by visual analog scales (VAS) for 21 days while sleeping in their own beds. Subsequently, participants’ beds were replaced by medium-firm mattresses specifically layered with foam and latex based on the participants’ reported prominent sleeping position and they again rated their sleep comfort and quality daily for the following 12 weeks. Analysis yielded significant differences between pre- and post means for all variables and for back pain, we found significant (p < 0.01) differences between the first posttest mean and weeks 4 and weeks 8–12, thus indicating progressive improvement in both back pain and stiffness while sleeping on the new mattresses. Additionally, the number of days per week of experiencing poor sleep and physical discomfort decreased significantly. It was concluded that sleep surfaces are related to sleep discomfort and that is indeed possible to reduce pain and discomfort and to increase sleep quality in those with chronic back pain by replacing mattresses based on sleeping position.     

Performance comparison of artificial neural network and logistic regression model for differentiating lung nodules on CT scans

PurposeTo compare the diagnostic performances of artificial neural networks (ANNs) and multivariable logistic regression (LR) analyses for differentiating between malignant and benign lung nodules on computed tomography (CT) scans.MethodsThis study evaluated 135 malignant nodules and 65 benign nodules. For each nodule, morphologic features (size, margins, contour, internal characteristics) on CT images and the patient’s age, sex and history of bloody sputum were recorded. Based on 200 bootstrap samples generated from the initial dataset, 200 pairs of ANN and LR models were built and tested. The area under the receiver operating characteristic (ROC) curve, Hosmer–Lemeshow statistic and overall accuracy rate were used for the performance comparison.ResultsANNs had a higher discriminative performance than LR models (area under the ROC curve: 0.955 ± 0.015 (mean ± standard error) and 0.929 ± 0.017, respectively, p < 0.05). The overall accuracy rate for ANNs (90.0 ± 2.0%) was greater than that for LR models (86.9 ± 1.6%, p < 0.05). The Hosmer–Lemeshow statistic for the ANNs was 8.76 ± 6.59 vs. 6.62 ± 4.03 (p > 0.05) for the LR models.ConclusionsWhen used to differentiate between malignant and benign lung nodules on CT scans based on both objective and subjective features, ANNs outperformed LR models in both discrimination and clinical usefulness, but did not outperform for the calibration.     

A study of the non-linear adjustment for analogy based software cost estimation

Cost estimation is one of the most important but most difficult tasks in software project management. Many methods have been proposed for software cost estimation. Analogy Based Estimation (ABE), which is essentially a case-based reasoning (CBR) approach, is one popular technique. To improve the accuracy of ABE method, several studies have been focusing on the adjustments to the original solutions. However, most published adjustment mechanisms are based on linear forms and are restricted to numerical type of project features. On the other hand, software project datasets often exhibit non-normal characteristics with large proportions of categorical features. To explore the possibilities for a better adjustment mechanism, this paper proposes Artificial Neural Network (ANN) for Non-linear adjustment to ABE (NABE) with the learning ability to approximate complex relationships and incorporating the categorical features. The proposed NABE is validated on four real world datasets and compared against the linear adjusted ABEs, CART, ANN and SWR. Subsequently, eight artificial datasets are generated for a systematic investigation on the relationship between model accuracies and dataset properties. The comparisons and analysis show that non-linear adjustment could generally extend ABE’s flexibility on complex datasets with large number of categorical features and improve the accuracies of adjustment techniques.     

Sleep in a live-in mining operation: The influence of start times and restricted non-work activities

The amount of sleep obtained between shifts is influenced by numerous factors including the length of work and rest periods, the timing of the rest period relative to the endogenous circadian cycle and personal choices about the use of non-work time. The current study utilised a real-world live-in mining environment to examine the amount of sleep obtained when access to normal domestic, family and social activities was restricted. Participants were 29 mining operators (26 male, average age 37.4 ± 6.8 years) who recorded sleep, work and fatigue information and wore an activity monitor for a cycle of seven day shifts and seven night shifts (both 12 h) followed by either seven or fourteen days off. During the two weeks of work participants lived on-site. Total sleep time was significantly less (p < 0.01) while on-site on both day (6.1 ± 1.0 h) and night shifts (5.7 ± 1.5 h) than days off (7.4 ± 1.4 h). Further, night shift sleep was significantly shorter than day-shift sleep (p < 0.01). Assessment of subjective fatigue ratings showed that the sleep associated with both days off and night shifts had a greater recovery value than sleep associated with day shifts (p < 0.01). While on-site, participants obtained only 6 h of sleep indicating that the absence of competing domestic, family and social activities did not convert to more sleep. Factors including shift start times and circadian influences appear to have been more important.     

Methodology for automatic bioacoustic classification of anurans based on feature fusion

The automatic recognition of anurans by their calls provides indicators of ecosystem health and habitat quality. This paper presents a new methodology for the acoustic classification of anurans using a fusion of frequency domain features, Mel and Linear Frequency Cepstral Coefficients (MFCCs and LFCCs), with time domain features like entropy and syllable duration through intelligent systems. This methodology has been validated in three databases with a significant number of different species proving the strength of this approach. First, the audio recordings are automatically segmented into syllables which represent different anuran calls. For each syllable, both types of features are computed and evaluated separately as in previous works. In the experiments, a novel data fusion method has been used showing an increase of the classification accuracy which achieves an average of 98.80% ± 2.43 in 41 anuran species from AmphibiaWeb database, 96.90% ± 3.57 in 58 frogs from Cuba and 95.48% ± 4.97 in 100 anurans from southern Brazil and Uruguay; reaching a classification rate of 95.38% ± 5.05 for the aggregate dataset of 199 species.     

Towards an accurate sleep apnea detection based on ECG signal: The quintessential of a wise feature selection

A wise feature selection from minute-to-minute Electrocardiogram (ECG) signal is a challenging task for many reasons, but mostly because of the promise of the accurate detection of clinical disorders, such as the sleep apnea. In this study, the ECG signal was modeled in order to obtain the Heart Rate Variability (HRV) and the ECG-Derived Respiration (EDR). Selected features techniques were used for benchmark with different classifiers such as Artificial Neural Networks (ANN) and Support Vector Machine(SVM), among others. The results evidence that the best accuracy was 82.12%, with a sensitivity and specificity of 88.41% and 72.29%, respectively. In addition, experiments revealed that a wise feature selection may improve the system accuracy. Therefore, the proposed model revealed to be reliable and simpler alternative to classical solutions for the sleep apnea detection, for example the ones based on the Polysomnography.     

Predicting the compressive strength of ground granulated blast furnace slag concrete using artificial neural network

In this study, an artificial neural networks study was carried out to predict the compressive strength of ground granulated blast furnace slag concrete. A data set of a laboratory work, in which a total of 45 concretes were produced, was utilized in the ANNs study. The concrete mixture parameters were three different water–cement ratios (0.3, 0.4, and 0.5), three different cement dosages (350, 400, and 450 kg/m³) and four partial slag replacement ratios (20%, 40%, 60%, and 80%). Compressive strengths of moist cured specimens (22 ± 2 °C) were measured at 3, 7, 28, 90, and 360 days. ANN model is constructed, trained and tested using these data. The data used in the ANN model are arranged in a format of six input parameters that cover the cement, ground granulated blast furnace slag, water, hyperplasticizer, aggregate and age of samples and, an output parameter which is compressive strength of concrete. The results showed that ANN can be an alternative approach for the predicting the compressive strength of ground granulated blast furnace slag concrete using concrete ingredients as input parameters.     

Proposing two new metaheuristic algorithms of ALO-MLP and SHO-MLP in predicting bearing capacity of circular footing located on horizontal multilayer soil

In this study, for the issue of shallow circular footing’s bearing capacity (also shown as F_ult), we used the merits of artificial neural network (ANN), while optimized it by two metaheuristic algorithms (i.e., ant lion optimization (ALO) and the spotted hyena optimizer (SHO)). Several studies demonstrated that ANNs have significant results in terms of predicting the soil’s bearing capacity. Nevertheless, most models of ANN learning consist of different disadvantages. Accordantly, we focused on the application of two hybrid models of ALO–MLP and SHO–MLP for predicting the F_ult placed in layered soils. Moreover, we performed an Extensive Finite Element (FE) modeling on 16 sets of soil layer (soft soil placed onto stronger soil and vice versa) considering a database that consists of 703 testing and 2810 training datasets for preparing the training and testing datasets. The independent variables in terms of ALO and SHO algorithms have been optimized by taking into account a trial and error process. The input data layers consisted of (i) upper layer foundation/thickness width (h/B) ratio, (ii) bottom and topsoil layer properties (for example, six of the most important properties of soil), (iii) vertical settlement (s), (iv) footing width (B), where the main target was taken F_ult. According to RMSE and R², values of (0.996 and 0.034) and (0.994 and 0.044) are obtained for training dataset and values of (0.994 and 0.040) and (0.991 and 0.050) are found for the testing dataset of proposed SHO–MLP and ALO–MLP best-fit prediction network structures, respectively. This proves higher reliability of the proposed hybrid model of SHO–MLP in approximating shallow circular footing bearing capacity.

     

Extracting rules for classification problems: AIS based approach

Although Artificial Neural Network (ANN) usually reaches high classification accuracy, the obtained results in most cases may be incomprehensible. This fact is causing a serious problem in data mining applications. The rules that are derived from ANN are needed to be formed to solve this problem and various methods have been improved to extract these rules. In our previous work, a hybrid neural network was presented for classification (Kahramanli & Allahverdi, 2008). In this study a method that uses Artificial Immune Systems (AIS) algorithm has been presented to extract rules from trained hybrid neural network. The data were obtained from the University of California at Irvine (UCI) machine learning repository. The datasets are Cleveland heart disease and Hepatitis data. The proposed method achieved accuracy values 96.4% and 96.8% for Cleveland heart disease dataset and Hepatitis dataset respectively. It is been observed that these results are one of the best results comparing with results obtained from related previous studies and reported in UCI web sites.