In order to improve the accuracy and efficiency of automatic seizure detection, the paper proposes a method based on improved genetic algorithm optimization back propagation (IGA-BP) neural network for epilepsy diagnosis, and uses the method to achieve detection of clinical epilepsy rapidly and effectively. Firstly, the method extracted the linear and nonlinear features of the epileptic electroencephalogram (EEG) signals and used a Gaussian mixture model (GMM) to perform cluster analysis on EEG features. Next, expectation maximization (EM) algorithm was used to estimate GMM parameters to calculate the optimal parameters for the selection operator of genetic algorithm (GA). The initial weights and thresholds of the BP neural network were obtained through using the improved genetic algorithm. Finally, the optimized BP neural network is used for the classification of the epileptic EEG signals to detect the epileptic seizure automatically. Compared with the traditional genetic algorithm optimization back propagation (GA-BP), the IGA-BP neural network can improve the population convergence rate and reduce the classification error. In the process of automatic detection of epilepsy, the method improves the detection accuracy in the automatic detection of epilepsy disorders and reduced inspection time. It has important application value in the clinical diagnosis and treatment of epilepsy.
Real-time continuous glucose monitoring can help diabetics to control blood sugar levels within the normal range. However, in the process of practical monitoring, the output of real-time continuous glucose monitoring system is susceptible to glucose sensor and environment noise, which will influence the measurement accuracy of the system. Aiming at this problem, a dual-calibration algorithm for the moving-window double-layer filtering algorithm combined with real-time self-compensation calibration algorithm is proposed in this paper, which can realize the signal drift compensation for current data. And a real-time continuous glucose monitoring instrument based on this study was designed. This real-time continuous glucose monitoring instrument consisted of an adjustable excitation voltage module, a current-voltage converter module, a microprocessor and a wireless transceiver module. For portability, the size of the device was only 40 mm × 30 mm × 5 mm and its weight was only 30 g. In addition, a communication command code algorithm was designed to ensure the security and integrity of data transmission in this study. Results of experiments in vitro showed that current detection of the device worked effectively. A 5-hour monitoring of blood glucose level in vivo showed that the device could continuously monitor blood glucose in real time. The relative error of monitoring results of the designed device ranged from 2.22% to 7.17% when comparing to a portable blood meter.
Under the current situation of the rapid development of brain-like artificial intelligence and the increasingly complex electromagnetic environment, the most bionic and anti-interference spiking neural network has shown great potential in computing speed, real-time information processing, and spatiotemporal data processing. Spiking neural network is the core part of brain-like artificial intelligence, which realizes brain-like computing by simulating the structure of biological neural network and the way of information transmission. This article first summarizes the advantages and disadvantages of the five models, and analyzes the characteristics of several network topologies. Then, it summarizes the spiking neural network algorithms. The unsupervised learning based on spike timing dependent plasticity (STDP) rules and four types of supervised learning algorithms are analyzed. Finally, the research on brain-like neuromorphic chips at home and abroad are reviewed. This paper aims to provide learning ideas and research directions for new colleagues in the field of spiking neural network.
As one of the standard electrophysiological signals in the human body, the photoplethysmography contains detailed information about the blood microcirculation and has been commonly used in various medical scenarios, where the accurate detection of the pulse waveform and quantification of its morphological characteristics are essential steps. In this paper, a modular pulse wave preprocessing and analysis system is developed based on the principles of design patterns. The system designs each part of the preprocessing and analysis process as independent functional modules to be compatible and reusable. In addition, the detection process of the pulse waveform is improved, and a new waveform detection algorithm composed of screening-checking-deciding is proposed. It is verified that the algorithm has a practical design for each module, high accuracy of waveform recognition and high anti-interference capability. The modular pulse wave preprocessing and analysis software system developed in this paper can meet the individual preprocessing requirements for various pulse wave application studies under different platforms. The proposed novel algorithm with high accuracy also provides a new idea for the pulse wave analysis process.
In this study, surface electromyography (sEMG) of the lower limbs of cerebral-palsy (CP) subjects in gait cycle was recorded and its parameters of gait cycle characters were analyzed to assess their clinical severity. Three algorithms, including integrated profile (IP), sample-entropy (SampEN) and smooth nonlinear energy operator (SNEO) algorithm, were applied to calculate the duration of walking sEMG segments in simulated SEMG signals. After that, the efficiency and accuracy were compared among these three algorithms. SNEO was then selected as the optimal algorithm among the three algorithms and employed for real sEMG signal processing of CP subjects. The results indicated that there was no significant difference in the accuracy of sEMG segement detection for the three algorithms. However, the computation speed of SNEO algorithm was much faster than those of the others and thus it was a suitable algorithm for detecting walking sEMG segments of CP subjects. In addition, the positive correlation was found between the clinical severity and the mean duration of walking sEMG segments in CP subjects. The results indicated that there was a significant difference in the three groups of CP subjects with different levels of severity. Our findings showed that the mean duration of walking sEMG segments could be considered as an assistant index to evaluate the clinical severity of CP subjects.
Objective To construct and compare logistic regression and decision tree models for predicting systemic inflammatory response syndrome (SIRS) in patients with type B aortic dissection (TBAD) after interventional surgery. Methods A retrospective analysis was conducted on clinical data of TBAD patients at Peking University Shenzhen Hospital from 2020 to 2024. The patients were divided into a SIRS group and a non SIRS group based on whether SIRS occurred within 24 hours after surgery. Multivariate logistic regression was used to analyze the influencing factors of SIRS occurrence in TBAD intervention patients, and a decision tree model was constructed using SPSS Modeler to compare the predictive performance of the two models. Results A total of 742 patients with TBAD were included, including 579 males and 163 females, aged between 27 and 97 (58.85±10.79) years. Within 24 hours after intervention, a total of 506 patients developed SIRS, with an incidence rate of 68.19%. Logistic regression analysis showed that the extensive involvement of the dissection, the surgical time≥ 2 hours, PET coated stents implanted, serum creatinine, white blood cell count, C-reactive protein, monocyte count (MONO), neutrophil count levels elevated, estimated glomerular filtration rate and decreased albumin levels were independent risk factors for SIRS (P<0.05). The decision tree model selected a total of 10 explanatory variables and 6 layers with 37 nodes, among which MONO was the most important predictor. The area under the decision tree model curve was 0.829 [95% CI (0.800, 0.856)], which was better than the logistic regression model's 0.690 [95% CI (0.655, 0.723)], and the difference was statistically significant (P<0.001). Conclusion The incidence of SIRS after TBAD intervention is high, and the decision tree model has better predictive performance than logistic regression. It can identify high-risk patients with higher accuracy and provide a practical tool for early clinical intervention.
In this paper, we present a 3D virtual phantom design software, which was developed based on object-oriented programming methodology and dedicated to medical physics research. This software was named Magical Phantom (MPhantom), which is composed of 3D visual builder module and virtual CT scanner. The users can conveniently construct any complex 3D phantom, and then export the phantom as DICOM 3.0 CT images. MPhantom is a user-friendly and powerful software for 3D phantom configuration, and has passed the real scene's application test. MPhantom will accelerate the Monte Carlo simulation for dose calculation in radiation therapy and X ray imaging reconstruction algorithm research.
The microarray technology used in biological and medical research provides a new idea for the diagnosis and treatment of cancer. To find different types of cancer and to classify the cancer samples accurately, we propose a new cluster ensemble framework Dual Neural Gas Cluster Ensemble (DNGCE), which is based on neural gas algorithm, to discover the underlying structure of noisy cancer gene expression profiles. This framework DNGCE applies the neural gas algorithm to perform clustering not only on the sample dimension, but also on the attribute dimension. It also adopts the normalized cut algorithm to partition off the consensus matrix constructed from multiple clustering solutions. We obtained the final accurate results. Experiments on cancer gene expression profiles illustrated that the proposed approach could achieve good performance, as it outperforms the single clustering algorithms and most of the existing approaches in the process of clustering gene expression profiles.
Early accurate detection of inferior myocardial infarction is an important way to reduce the mortality from inferior myocardial infarction. Regrading the existing problems in the detection of inferior myocardial infarction, complex model structures and redundant features, this paper proposed a novel inferior myocardial infarction detection algorithm. Firstly, based on the clinic pathological information, the peak and area features of QRS and ST-T wavebands as well as the slope feature of ST waveband were extracted from electrocardiogram (ECG) signals leads Ⅱ, Ⅲ and aVF. In addition, according to individual features and the dispersion between them, we applied genetic algorithm to make judgement and then input the feature with larger degree into support vector machine (SVM) to realize the accurate detection of inferior myocardial infarction. The proposed method in this paper was verified by Physikalisch-Technische Bundesanstalt (PTB) diagnostic electrocardio signal database and the accuracy rate was up to 98.33%. Conforming to the clinical diagnosis and the characteristics of specific changes in inferior myocardial infarction ECG signal, the proposed method can effectively make precise detection of inferior myocardial infarction by morphological features, and therefore is suitable to be applied in portable devices development for clinical promotion.
Detection and classification of malignant arrhythmia are key tasks of automated external defibrillators. In this paper, 21 metrics extracted from existing algorithms were studied by retrospective analysis. Based on these metrics, a back propagation neural network optimized by genetic algorithm was constructed. A total of 1,343 electrocardiogram samples were included in the analysis. The results of the experiments indicated that this network had a good performance in classification of sinus rhythm, ventricular fibrillation, ventricular tachycardia and asystole. The balanced accuracy on test dataset reached up to 99.06%. It illustrates that our proposed detection algorithm is obviously superior to existing algorithms. The application of the algorithm in the automated external defibrillators will further improve the reliability of rhythm analysis before defibrillation and ultimately improve the survival rate of cardiac arrest.