基于双谱分析—BP神经网络的ICU急性低血压发生预测模型研究

目的:ICU中,急性低血压的发生严重威胁着患者的生命安全,能够及时准确地对其预测具有重要临床意义。为了提高医生对患者发病提前判断的准确性,本文研究了一种基于双谱分析和BP神经网络的急性低血压发生的预测模型。方法:应用双谱分析提取动脉血压数据特征,构建特征向量,利用BP神经网络训练出分类预测模型,实现对急性低血压发生的提前预测。结果:经过不断的优化调整,本文最终构造了一个三层的BP神经网络预测模型,具有良好的自动预测能力。结论:实验表明,本方法能达到比较好的分类预测效果,可为ICU中急性低血压发生的提前预测和干预提供辅助参考。     

基于AR模型和支持向量机的急性低血压预测

ICU中,急性低血压的发生严重威胁着患者的生命安全,临床上对其预测性判断主要依靠医生经验。为实现急性低血压预测,利用PhsioNet的MIMIC II数据库ICU监护中的患者临床记录,对发生与未发生急性低血压两者间的平均动脉压信号进行AR模型的功率谱估计,运用医学信息学理论,选取功率谱幅度的中位数、平均值、最大值、标准偏差和极差用于支持向量机分类预测器的学习和训练,建立分类预测模型。预测模型对测试集进行分类预测,得到预测正确率为87.5%,表明相对于直接提取患者平均动脉压信号的统计特征参数作为预测特征,本方法具有更好的预测效果,有利于实现急性低血压提前预测。     

三维步态分析检测基于BP神经网络的膝踝运动

背景：虽然在人体步态方面已有大量研究,但针对膝踝协调运动的研究很少。 目的：用BP神经网络分析膝踝协调运动关系。 方法：利用三维步态分析系统检测了30名健康志愿者以快、中、慢3种步速行走时的步态数据,进行统计分析;并通过建立BP神经网络预测数据,同时对膝踝协调的控制方法进行探讨。 结果与结论：不同的人步态不同,但BP神经网络预测所得曲线与实验基本一致,证实了用BP神经网络做膝踝运动关系预测的合理性和可行性,很好的研究了膝踝协调性,给全智能膝踝协调控制假肢的研发提供了理论依据。     

基于支持向量机的ICU中急性低血压预测模型研究

在重症监护室(ICU)的监护中,急性低血压(AHE)的发生严重威胁着患者的生命安全,临床上主要依靠医生的经验处置。本文运用医学信息学的理论,研究一种ICU中AHE发生的预测模型。利用ICU监护中血压变化的连续记录数据,分析发生与未发生AHE两者间平均动脉压(MAP)信号的变化趋势与特点,基于统计学习理论的支持向量机(SVM)方法,选取中位数、平均值等统计特征参数用于学习和训练,建立分类预测模型。在此基础上,对不同核函数构成的分类器和预测算法进行了比较分析。实验验证,本方法能够达到比较好的分类预测效果,有利于AHE发生的提前预测。     

ICU中发生急性低血压的预测方法研究进展

在ICU监护中,急性低血压的发生可能引起严重的后果,甚至威胁患者的生命安全,如何检测和提前预报急性低血压发生已成为医学界必须重视的临床问题。近年来,医疗监护技术和信号分析处理方法的迅速发展,促进了对急性低血压发生的预测方法的相关研究,研究内容主要侧重于两个方面:一是通过研究与可能发生急性低血压相关的生命体征参数,发现动脉压、心率及血氧饱和度等可以作为有效的指标参数;二是研究这些参数的变化趋势,通过一定的时间窗和阈值判断,可达到提前1小时预测急性低血压的发生。将数字信号处理技术和临床监护参数相结合,是实现智能化监护技术发展的方向,在继续丰富监护数据库的基础上,研究者们正致力于应用医学信息学方法认识急性低血压发生的规律,寻求提前预测急性低血压发生的方法,进而设计智能化预测软件。以上研究有利于实现急性低血压提前预测,提前干预,大大降低抢救风险,具有重大的临床研究应用价值。     

基于小波变换和BP神经网络的室性早搏(PVC)识别

目的:室性早搏(PVC)是最常见的心律失常,能够实时、准确检测出它具有重要的临床意义。为了能够及早地发现室性早搏(PVC)、提高医生对室性早搏(PVC)诊断的准确性,本文研究了基于小波变换和BP神经网络的室性早搏(PVC)识别方法。方法:首先选取ECG信号的特征参数,用小波变换检测其特征点,提取特征参数,构造特征向量,然后再用训练好的BP神经网络对室性早搏(PVC)进行识别,从而实现ECG信号的正常心律和室性早搏的自动检测识别。结果:本文构造了一个7-10-1结构的三层BP神经网络,用MIT-BIH标准心电数据库中的心电信号数据对构造的神经网络进行训练、检测识别,识别率达到预期的效果。结论:分析MIT-BIH心电数据库六组试验数据的实验结果,证明了基于小波变换和BP神经网络的室性早搏(PVC)识别的方法具有较高的PVC自动检测识别率,对医生的诊断有良好的辅助作用,具有较高的应用价值。     

基于非线性混沌分析的急性低血压预测

重症监护室中急性低血压发生的预测研究是临床医学的重点与难点。本文应用非线性混沌分析方法,对MIMICⅡ临床记录中患者的平均动脉压时间序列信号进行分析,构建患者的李雅普诺夫指数变化曲线。研究发现患者在急性低血压症状发生前一般会出现明显的指数曲线突变情况,这为急性低血压的有效预测提供了直观的依据,为急性低血压的理论研究与应用提供了一条可参考的思路。     

基于GA-BP神经网络进行呼吸运动预测的研究

放射治疗是胸腹部肿瘤的常用治疗手段,但由于病人在治疗过程中的呼吸使胸腹部肿瘤产生运动位移降低了放射治疗的质量。人们希望在治疗的过程中对目标进行运动估计,以减少呼吸运动在治疗过程中的影响。BP人工神经网络由于其良好的非线性逼近特性被用于呼吸运动估计,可以较好地反映呼吸运动的发展趋势;但BP神经网络本质是梯度下降法,容易陷入局部最优。利用遗传算法(GA)和BP网络相结合的方法,遗传算法具有良好的全局搜索能力,弥补BP神经网络易于陷入局部最优的缺点。通过对9例呼吸运动信号进行对比实验,结果表明用GA-BP神经网络的预测精度高于单纯使用BP神经网络的预测精度。     

基于B样条小波变换的视觉诱发电位的识别方法

针对已知的视觉诱发电位信号处理及特征提取存在的抗干扰能力差及识别率低的问题,提出一种基于B样条小波变换与BP神经网络的视觉诱发电位(VEP)的特征提取及识别方法。首先,对VEP进行少次平均预处理,以增强信号的信噪比;其次,基于B样条小波变换对VEP进行特征选取,并通过BP神经网络分类器对特征进行分类。与现有方法相比,所提出的方法能够较好地提取视觉诱发电位特征,对于含噪声的信号具有较好的抗干扰能力和识别能力。实验结果表明,采用B样条小波变换结合BP神经网络的方法,对视觉诱发电位的平均识别率为90.4%,优于其他方法,验证了所提出方法的正确性和有效性。     

基于肌电信号的膝关节跨越障碍角度预测方法

为解决人体跨越障碍物时膝关节角度输出的问题,针对性设计一种穿戴式信号获取实验台,对下肢运动姿态进行运动分析,将肌肉电信号及关节角度信号作为运动数据,对信号进行处理后利用BP神经网络预测跨越障碍时输出角度,提出一种利用BP神经网络算法,根据不同大腿抬起高度,分析膝关节运动主动肌与被动肌发力程度,预测输出人体跨越障碍时膝关节角度的方法,能够有效帮助假肢膝关节或康复机器人实现跨越障碍的复杂动作。     

基于波变换的心室晚电位神经网络检测

本文采用小波变换和频谱标测技术从信号平均心电图中提取一组含有晚电位信息的比值特征参数，利用人工神经网络ＢＰ算法，完成对心室晚电位的检测。研究结果表明，采用神经网络检测晚电位的方法不仅具有较高的检测正确率，而且可以放松对信号平均心电图分析时段选取及ＱＲＳ终点定位准确度的要求。     

Prediction of oculocardiac reflex in strabismus surgery using neural networks.

Successfully predicting an oculocardiac reflex (OCR) is difficult to achieve despite various proposed maneuvers. The aim of this study was to test the models built up by neural networks to predict the occurrence of OCR during strabismus surgery in children. Premedication was not given. Atropine 0.01 mg/kg was medicated just before induction. Induction was performed with fentanyl or ketorolac, followed by propofol. Atracurium or vecuronium was given for intubation. Anesthesia was maintained with O2-N2O with continuous propofol infusion. Chi-square test was performed for induction agents, gender, weight, muscle blockade, repaired muscle, number of repaired muscles, duration of operation to detect any association between the occurrence of OCR and to develop the model of neural networks. The multi-layer perceptron, radial basis function and Bayesian backpropagation network were tested. The occurrence of OCR was significantly associated with gender and repaired muscle (p < 0.05). Gender, repaired muscle and age were considered as input for the multi-layer perceptron, radial basis function and Bayesian backpropagation network. Three neural networks had predicted the same correction rate in the occurrence of OCR as being 87.5% overall among 16 patients' records tested. These models are conceptually different in predicting compared to conventional maneuvers, and have the advantage of testing individually and foretelling the propensity. By comparison neural networks use grouped experiential data and predict OCR by the learning rule. Neural networks require a relatively abundant number of experienced and homogenous patients' records to establish an accurate model. The multi-layer perceptron, radial basis function and Bayesian backpropagation modeling network may be an alternative way, and preferable to vagal tone maneuvers if the associated relationships to the occurrence of OCR are more clearly defined.     

Spontaneous arousals during quiet sleep in piglets: a visual and wavelet-based analysis

STUDY OBJECTIVES: The objectives of the study were to characterize spontaneous arousals during NREM sleep in piglets and to compare two methods of identifying these events: a "visual" technique using spectral analysis and an automated technique using wavelets. Our goal was to understand the benefits and limits of these methods when applied to sleep in human infants. DESIGN: Arousals were identified by evaluating rapid changes in EEG low frequency activity, blood pressure (BP), and heart rate (HR). A cortical arousal was defined as a rapid decrease in EEG low frequency activity. An autonomic arousal was defined by a transient increase in heart rate or a transient change in mean arterial BP (MAP). SETTING: Laboratory study in sleeping and awake piglets. PARTICIPANTS: Five 1-2 week old piglets. INTERVENTIONS: Chronically instrumented with a femoral arterial line, EEG, EOG, EMG electrodes, and a micro-dialysis probe with its tip located in the rostral ventral medulla. Artificial CSF (aCSF) was dialyzed into the RVM throughout the experiments Measurements: For the visual analysis, the average delta power (0.5-4 Hz) for each 5-second epoch was determined using spectral analysis. MAP and HR were analyzed in 1-second bins. Video images were analyzed for body movements and eye openings. Transient changes in blood pressure, HR, and delta power were then visually identified. For the wavelet analysis, a quantitative, automated technique with a defined "wakefulness threshold" was used to identify rapid decreases in EEG low frequency activity and the rate of change of MAP. RESULTS: Using the visual method, 117 episodes associated with stereotypical hemodynamic, EEG, and behavioral changes (startle) were identified. Seventy five events occurred in isolation or were first in a series of "multiple" events, 41 "multiple" events were defined as events occurring <20 seconds following a previous event. Eighteen events were associated with the termination of apnea. In isolated events or those occurring first in a series, the onset of changes in HR and BP clearly preceded the decrease in EEG amplitude and delta power. Using wavelet analysis, 73 EEG arousals and 115 MAP transients were identified independently; 62% of the EEG events were associated with a transient change in MAP and HR, and in these cases the onset of the hemodynamic events preceded EEG arousals. EEG arousals and MAP transients, however, also occurred alone and not associated with a stereotypical pattern of a startle, changes in MAP and HR and the EEG. CONCLUSIONS: Many of these spontaneous arousals represent integrated EEG, hemodynamic, and behavioral processes similar to arousal phenomena described in adult rats and human infants, but the pattern of spontaneous arousals appears to be more heterogeneous than has been described for arousals induced by exogenous stimuli. Both the visual and wavelet analysis identified these events, but the wavelet technique has the potential advantage of better time resolution and automation of the analysis.     

COMPUTER-AIDED DETECTION OF THE EPILEPTIC WAVES IN EEG:A REALIZED STRATEGY BY ADOPTING MULTI-METHOD

A.THESOURCESOFDATAThedatausedinthisresearchcamefromtheEEGlaboratoryoftheTianjinGeneralhospital.TheywereobtainedfromtWelvepatientsattheageofo.9~27,withthesamPlingrateof5l2persecondandsamplingPeriodof8minutes.Thesesignalscontainvariousbackgroundwavesandmostofthedatacontainvariouskindsofartifacts.Twopatients'datawereselectedtoobtaintheparametersandtrainoursystem,Whiletheothertenwaslefttotestitsefficiency.B.THECOMPOSITIONoFTHESYSTEMAtthebeginning,theEEGdatawassenttothecomPute…     

Classification of tetraplegics through automatic movement evaluation

The general problem of classification of functional movements in humans with spinal cord injuries requires the following questions to be answered: what are the essential kinematic parameters that we have to observe during the movement? Is it possible to estimate preserved motor skills based on kinematics? Which computational method for identification is suited to geometric feature analysis? To answer these questions we have developed the methodology which has two phases: (1) recordings of a series of specified arm movements; and (2) custom made software for graphical presentation of arm movements and the design of wavelet and neural networks for movement classification. The proposed protocol is automated and both graphical presentation and neural networks allow easy interpretation of the instrumented assessment to accomplish automatic classification of arm movements in tetraplegics. The protocol was evaluated on 16 spinal cord injury (SCI) patients and seven healthy control subjects for three different arm movements. The classification rate yielded results in the range 46-100% for movement trials that were tested. The application of neural networks for classification of arm movements is completed with results using different neural networks: backpropagation, radial basis, recurrent (Elman), self-organizing and Learning Vector Quantization (LVQ).     

Serum extracellular vesicle-derived ASS1 is a promising predictor for the occurrence of HEV-ALF

Development of biomarkers for predicting the occurrence of hepatitis E virus related-acute liver failure (HEV-ALF) is conducive to prevention and early intervention. Serum samples from 250 HEV-ALF patients, 250 patients with acute hepatitis E (AHE) and 250 health controls (HCs) were collected. We assessed the predictive ability of extracellular vesicle (EV)-derived argininosuccinate synthase 1 (ASS1) levels for HEV-ALF occurrence. Serum EVs were successfully isolated. EV-derived ASS1 levels in the HEV-ALF patients were significantly higher than those in the AHE patients and HCs. In HEV-ALF patients, EV-derived ASS1 levels were positively correlated with the number of failed organs and disease progression. The logistical regression showed that EV-derived ASS1 level is an independent risk factor for HEV-ALF, and orthogonal partial least squares discriminant analysis (OPLS-DA) also suggested that EV-derived ASS1 level has high predictive capability. Besides, the area under the curve (AUC) of EV-derived ASS1 level to predict HEV-ALF occurrence was 0.728 (0.684–0.772) with the sensitivity and specificity being 72.80% and 64.80%, which had a high decision-making ability. Furthermore, there existed no significant difference between the age ≥60 and age <60 groups in EV-derived ASS1 levels. Serum EV-derived ASS1 level is a promising predictor for the occurrence of HEV-ALF.     

基于模糊小波神经网络的磁共振图像分割方法

针对磁共振图像分割的特点,将小波分析优异的局部特性、模糊逻辑的定性知识表达能力和神经网络的自学习能力结合起来,提出了一种基于模糊小波神经网络的磁共振图像分割方法.该方法采用小波函数作为模糊隶属函数,利用神经网络实现模糊推理,并采用反向误差传播算法对网络进行训练.实验结果表明,这种基于模糊小波神经网络的磁共振图像分割方法与普通神经网络分割方法相比,具有更高的分割精度和更快的训练收敛速度.