Feature Values Research Articles

A Novel Continuous Sleep State Artificial Neural Network Model Based on Multi-Feature Fusion of Polysomnographic Data.

Sleep structure is crucial in sleep research, characterized by its dynamic nature and temporal progression. Traditional 30-second epochs falter in capturing the intricate subtleties of various micro-sleep states. This paper introduces an innovative artificial neural network model to generate continuous sleep depth value (SDV), utilizing a novel multi-feature fusion approach with EEG data, seamlessly integrating temporal consistency. The study involved 50 normal and 100 obstructive sleep apnea-hypopnea syndrome (OSAHS) participants. After segmenting the sleep data into 3-second intervals, a diverse array of 38 feature values were meticulously extracted, including power, spectrum entropy, frequency band duration and so on. The ensemble random forest model calculated the timing fitness value for all the features, from which the top 7 time-correlated features were selected to create detailed sleep sample values ranging from 0 to 1. Subsequently, an artificial neural network (ANN) model was trained to delineate sleep continuity details, unravel concealed patterns, and far surpassed the traditional 5-stage categorization (W, N1, N2, N3, and REM). The SDV changes from wakeful stage (mean 0.7021, standard deviation 0.2702) to stage N3 (mean 0.0396, standard deviation 0.0969). During the arousal epochs, the SDV increases from the range (0.1 to 0.3) to the range around 0.7, and decreases below 0.3. When in the deep sleep (≤0.1), the probability of arousal of normal individuals is less than 10%, while the average arousal probability of OSA patients is close to 30%. A sleep continuity model is proposed based on multi-feature fusion, which generates SDV ranging from 0 to 1 (representing deep sleep to wakefulness). It can capture the nuances of the traditional five stages and subtle differences in microstates of sleep, considered as a complement or even an alternative to traditional sleep analysis.

The Role of Typological Feature Prediction in NLP and Linguistics

Abstract Computational typology has gained traction in the field of Natural Language Processing (NLP) in recent years, as evidenced by the increasing number of papers on the topic and the establishment of a Special Interest Group on the topic (SIGTYP), including the organization of successful workshops and shared tasks. A considerable amount of work in this sub-field is concerned with prediction of typological features, for example, for databases such as the World Atlas of Language Structures (WALS) or Grambank. Prediction is argued to be useful either because (1) it allows for obtaining feature values for relatively undocumented languages, alleviating the sparseness in WALS, in turn argued to be useful for both NLP and linguistics; and (2) it allows us to probe models to see whether or not these typological features are encapsulated in, for example, language representations. In this article, we present a critical stance concerning prediction of typological features, investigating to what extent this line of research is aligned with purported needs—both from the perspective of NLP practitioners, and perhaps more importantly, from the perspective of linguists specialized in typology and language documentation. We provide evidence that this line of research in its current state suffers from a lack of interdisciplinary alignment. Based on an extensive survey of the linguistic typology community, we present concrete recommendations for future research in order to improve this alignment between linguists and NLP researchers, beyond the scope of typological feature prediction.

Skin Disease Detection Model

Abstract: The human skin is a remarkable organ susceptible to a myriad of know and unknown diseases. Many of these ailment are widespread, with some ranking among common worldwide. The complexity of diagnosing these diseases is compounded by challenges such as variations in skin texture, the presence of hair, and diverse skin colors. In some areas have limited access to medical facilities, individuals often neglect early symptoms, leading to exacerbated conditions over time. Furthermore, traditional diagnostic methods for skin diseases are time consuming. To address these challenges, there is a critical need to develop advanced diagnostic methods utilizing machine learning techniques to enhance accuracy cross various skin diseases. Machine learning algorithms have proven valuable in medical applications, leveraging image feature values to facilitate decision making. The diagnostic process involves three key stages: feature extraction, training, and testing. By employing machine learning technology, these algorithms learn from a diverse set of skin images o enhance their diagnostic capabilities. The primary goal is to significantly improved the accuracy of skin disease detection. This study focuses on utilizing color and texture features for the classification of skin diseases. The distinctive color of healthy skin differs from that affected by disease, while texture features effectively discern smoothness, coarseness, and regularity in images. Key features such as texture, color, and shape phyla pivotal role in image classification. The incorporation of convolution neural networks (CNN) further augments the capabilities of image classification in the realm of skin disease diagnosis

Partial Discharge Signal Pattern Recognition of Composite Insulation Defects in Cross-Linked Polyethylene Cables.

To investigate the pattern recognition of complex defect types in XLPE (cross-linked polyethylene) cable partial discharges and analyze the effectiveness of identifying partial discharge signal patterns, this study employs the variational mode decomposition (VMD) algorithm alongside entropy theories such as power spectrum entropy, fuzzy entropy, and permutation entropy for feature extraction from partial discharge signals of composite insulation defects. The mean power spectrum entropy (PS), mean fuzzy entropy (FU), mean permutation entropy (PE), as well as the permutation entropy values of IMF2 and IMF13 (Pe) are selected as the characteristic quantities for four categories of partial discharge signals associated with composite defects. Six hundred samples are selected from the partial discharge signals of each type of compound defect, amounting to a total of 2400 samples for the four types of compound defects combined. Each sample comprises five feature values, which are compiled into a dataset. A Snake Optimization Algorithm-optimized Support Vector Machine (SO-SVM) model is designed and trained, using the extracted features from cable partial discharge datasets as case examples for recognizing cable partial discharge signals. The identification outcomes from the SO-SVM model are then compared with those from conventional learning models. The results demonstrate that for partial discharge signals of XLPE cable composite insulation defects, the SO-SVM model yields better identification results than traditional learning models. In terms of recognition accuracy, for scratch and water ingress defects, SO-SVM improves by 14.00% over BP (Back Propagation) neural networks, by 5.66% over GA-BP (Genetic Algorithm-Back Propagation), and by 12.50% over SVM (support vector machine). For defects involving metal impurities and scratches, SO-SVM improves by 13.39% over BP, 9.34% over GA-BP, and 12.56% over SVM. For defects with metal impurities and water ingress, SO-SVM shows enhancements of 13.80% over BP, 9.47% over GA-BP, and 13.97% over SVM. Lastly, for defects combining metal impurities, water ingress, and scratches, SO-SVM registers increases of 11.90% over BP, 9.59% over GA-BP, and 12.05% over SVM.

Low-Cost Portable Road Smoothness Testing Method Based on Pseudo-Vibration Velocity Range

Road smoothness not only directly affects the safety and comfort of vehicle travel but also relates to the efficiency and cost-effectiveness of road maintenance. Traditional road smoothness detection methods usually require professional equipment and personnel, leading to high costs and cumbersome operations. Therefore, finding a low-cost, simple, and accurate method for detecting road smoothness is of great significance. This study uses vehicle-mounted acceleration sensors to detect road smoothness, establishing a correlation between driving vibration acceleration data and the international roughness index (IRI). For this research, a driving vibration acceleration data acquisition device was developed, and the driving acceleration data from the test sections were denoised and their feature values extracted. The pseudo-vibration velocity range was used as the characteristic index representing the road surface smoothness IRI value. Testing with different vehicle types showed that the method is applicable to both sedans and SUV models, yielding a relative error of 8.9% for the sedan smoothness test model and 6.7% for the SUV smoothness test model. This study contributes to conducting large-scale road smoothness detection at a low cost, improving the efficiency of road maintenance and operations.

Skin Disease Detection

The human skin is a remarkable organ susceptible to a myriad of know and unknown diseases. Many of these ailment are widespread, with some ranking among common worldwide. The complexity of diagnosing these diseases is compounded by challenges such as variations in skin texture, the presence of hair, and diverse skin colors. In some areas have limited access to medical facilities, individuals often neglect early symptoms, leading to exacerbated conditions over time. Furthermore, traditional diagnostic methods for skin diseases are time-consuming. To address these challenges, there is a critical need to develop advanced diagnostic methods utilizing machine learning techniques to enhance accuracy cross various skin diseases. Machine learning algorithms have proven valuable in medical applications, leveraging image feature values to facilitate decision-making. The diagnostic process involves three key stages: feature extraction, training, and testing. By employing machine learning technology, these algorithms learn from a diverse set of skin images o enhance their diagnostic capabilities. The primary goal is to significantly improved the accuracy of skin disease detection. This study focuses on utilizing color and texture features for the classification of skin diseases. The distinctive color of healthy skin differs from that affected by disease, while texture features effectively discern smoothness, coarseness, and regularity in images. Key features such as texture, color, and shape phyla pivotal role in image classification. The incorporation of convolution neural networks (CNN) further augments the capabilities of image classification in the realm of skin disease diagnosis

Inter-patient variability of local cycle length and duty cycle distribution of the atrial fibrillation substrate beyond pulmonary vein isolation

Abstract Appropriate targets for catheter ablation of atrial fibrillation (AFib) beyond pulmonary vein isolation (PVI) are still controversially discussed. Various approaches such as ablation of complex fractionated atrial electrograms (EGMs), areas with spatio-temporal dispersion, and areas with increased rotor density have not been reproducible or failed to improve clinical success rates. The local cycle length (CL) and duty cycle (DC) distribution during AFib offer novel measures to characterize the atrial substrate. This study aims at providing first insights into the inter-individual variability of observed CL and DC distributions, their interdependence, and their correlation to the duration of AFib episodes to improve the individual characterization of the AFib substrate and tailor ablation strategies in future. In patients with recurrence of AFib despite PVI, a high-resolution 3D map was recorded at baseline during AFib. Intracardiac EGMs were preprocessed to improve local signal quality and far field suppression. EGMs were recorded with the multipolar, basket-shaped high-density mapping catheter. Local CL and DC distribution were visualized using a scatter plot and histograms. CL quantifies the time between subsequent local activations, DC quantifies the fraction of CL with local EGM activity, and the confidence (conf) feature quantifies the spatiotemporal consistency of EGM morphology. Only EGMs with conf > 0.6 are analysed. Feature calculation does not only take into account single EGMs but a weighted conglomerate of all acquisitions within a series of beats. Adjustable CL and DC thresholds allow highlighting regions within the 3D map for individual spatial analysis. 14 patients (7x paroxysmal AFib, 7x persistent AFib) were enrolled. All subjects underwent repeat ablation after PVI due to recurrence of AFib. CL and DC histograms were of highly individual shape. The most prevalent feature value varied from 158ms to 263ms for CL and from 38% to 96% for DC; histogram widths (inter-quartile-range) from 10ms to 34ms and from 26% to 36%, respectively. Normal distributions, multimodal distributions, and skewed distributions exhibiting a dominant peak were observed for CL. Both the most prevalent CL and the width of the distribution were highly individual. DC was observed to be normally to equally distributed with mostly broad bandwidths. Linear regression revealed a positive dependency between the most prevalent CL and the width of CL distributions. The most prevalent value and the width of both CL and DC did not differ significantly for the paroxysmal and persistent subgroup. Compare figure for detailed results. CL and DC distributions can help to characterize the highly individual type of AFib. The presented results suggest that there will not be a one-size-fits-all ablation approach for the AFib substrate. Further research is needed to identify clusters of similar pattern and individually adjust the ablation strategy.

Gaussian-kernel weighted neighborhood preserving embedding algorithm and its application in fault detection

Fault detection in industrial processes is essential for enhancing production safety. Despite the application of the neighborhood preserving embedding (NPE) algorithm in fault detection as a manifold learning technique, a notable limitation exists-NPE overlooks local geometric structure, leading to suboptimal fault detection and occasional false alarms. This paper introduces the Gaussian kernel weighted NPE (KW-NPE) algorithm to address this challenge. Specifically designed for precise weight assignment in local structures, KW-NPE strategically employs the Gaussian kernel method to project the spatial neighborhood set and capture comprehensive local structural characteristics. The weight assignment, dependent on feature values, enhances the retention of intrinsic structure during dimensionality reduction. A novel objective function further augments this process.To assess performance, a comprehensive composite index is introduced in a case study, amalgamating the false alarm rate and fault detection rate. The effectiveness of the KW-NPE algorithm is demonstrated through extensive simulations and its application to the Tennessee Eastman process dataset, highlighting its superiority over conventional approaches.

Cross-Regional Crop Classification Based on Sentinel-2

Accurate crop classification is of vital importance for agricultural water management. Most researchers have achieved crop classification by model optimization in the same temporal and regional domain by adjusting the value of input features. This study aims to improve the accuracy of crop classification across temporal and spatial domains. Sentinel-2 satellite imagery is employed for crop classification training and prediction in selected farming areas of Heilongjiang Province by calculating vegetation indices and constructing sequential input feature datasets. The HUNTS filtering method was used to mitigate the influence of cloud cover, which increased the stability and completeness of the input feature data across different years. To address the issue of shifts in the input feature values during cross-scale classification, this study proposes the hypothesis testing distribution method (HTDM). This method balances the distribution of input feature values in the test set even without knowing the crop distribution, thereby enhancing the accuracy of the classification test set. The results indicate that the HTDM significantly improves prediction accuracy in cases of substantial image quality variance. In 2022, the recognition accuracy for crop types at all farms processed by the HTDM was above 87%, showcasing the strong robustness of the HTDM.

Pattern Recognition of Partial Discharge Faults in Switchgear Using a Back Propagation Neural Network Optimized by an Improved Mantis Search Algorithm.

The dependable functioning of switchgear is essential to maintain the stability of power supply systems. Partial discharge (PD) is a critical phenomenon affecting the insulation of switchgear, potentially leading to equipment failure and accidents. PDs are generally grouped into metal particle discharge, suspended discharge, and creeping discharge. Different types of PDs are closely related to the severity of a PD. Partial discharge pattern recognition (PDPR) plays a vital role in the early detection of insulation defects. In this regard, a Back Propagation Neural Network (BPNN) for PDPR in switchgear is proposed in this paper. To eliminate the sensitivity to initial values of BPNN parameters and to enhance the generalized ability of the proposed BPRN, an improved Mantis Search Algorithm (MSA) is proposed to optimize the BPNN. The improved MSA employs some boundary handling strategies and adaptive parameters to enhance the algorithm's efficiency in optimizing the network parameters of BPNN. Principal Component Analysis (PCA) is introduced to reduce the dimensionality of the feature space to achieve significant time saving in comparable recognition accuracy. The initially extracted 14 feature values are reduced to 7, reducing the BPNN parameter count from 183 with 14 features to 113 with 7 features. Finally, numerical results are presented and compared with Decision Tree (DT), k-Nearest Neighbor classifiers (KNN), and Support Vector Machine (SVM). The proposed method in this paper exhibits the highest recognition accuracy in metal particle discharge and suspended discharge.

Research on the Value and Language Features of Chinese Language and Literature Texts Based on Text Mining Technology

The linguistic characteristics of a literary work are the way of thinking embodied in the author’s use of language. From the textual value of Chinese language literature, this paper analyzes the spiritual connotation of Chinese language literature from two dimensions: reading and education. Based on the web crawler technology, we obtain the text data of Chinese language literature from three writers, Bajin, Yu Zheng and Qiong Yao, preprocess the data through data cleaning, Chinese word segmentation, de-duplication, etc., and extract the feature values of the text by using the TF-IDF algorithm. Then the text documents are mapped onto vectors using the VSM model, and the parameters of the LDA topic model are estimated by the Gibbs sampling algorithm in order to better obtain the topic changes of the Chinese language literature texts. This paper carries out linguistic feature verification from the lexical and similarity features of Chinese language literary texts. It is found that the difference in lexical density between Ba Jin’s Cold Night and Resting Garden is only 2.1 percentage points, and the frequency of the verb “to say” is 1,213 times and 735 times respectively. The average sentence lengths of Yu Zheng and Qiong Yao fluctuate within the range of [18.49,34.27], and Qiong Yao’s works have a higher thematic concentration than Zheng Zheng’s works. Analyzing the linguistic features of Chinese language literary texts based on text mining techniques helps to understand the authors’ language usage methods and helps to promote innovative expression paths in literary texts.

An inversion-based clustering approach for complex clusters

BackgroundThe choice of an appropriate similarity measure plays a pivotal role in the effectiveness of clustering algorithms. However, many conventional measures rely solely on feature values to evaluate the similarity between objects to be clustered. Furthermore, the assumption of feature independence, while valid in certain scenarios, does not hold true for all real-world problems. Hence, considering alternative similarity measures that account for inter-dependencies among features can enhance the effectiveness of clustering in various applications.MethodsIn this paper, we present the Inv measure, a novel similarity measure founded on the concept of inversion. The Inv measure considers the significance of features, the values of all object features, and the feature values of other objects, leading to a comprehensive and precise evaluation of similarity. To assess the performance of our proposed clustering approach that incorporates the Inv measure, we evaluate it on simulated data using the adjusted Rand index.ResultsThe simulation results strongly indicate that inversion-based clustering outperforms other methods in scenarios where clusters are complex, i.e., apparently highly overlapped. This showcases the practicality and effectiveness of the proposed approach, making it a valuable choice for applications that involve complex clusters across various domains.ConclusionsThe inversion-based clustering approach may hold significant value in the healthcare industry, offering possible benefits in tasks like hospital ranking, treatment improvement, and high-risk patient identification. In social media analysis, it may prove valuable for trend detection, sentiment analysis, and user profiling. E-commerce may be able to utilize the approach for product recommendation and customer segmentation. The manufacturing sector may benefit from improved quality control, process optimization, and predictive maintenance. Additionally, the approach may be applied to traffic management and fleet optimization in the transportation domain. Its versatility and effectiveness make it a promising solution for diverse fields, providing valuable insights and optimization opportunities for complex and dynamic data analysis tasks.

A texture feature extraction method considering spatial continuity and gray diversity

Texture features play an important role in the field of remote sensing classification. However, most existing methods lack a comprehensive consideration of spatial continuity, which makes them either destroy the spatial integrity of regular ground objects or fail to quantify the fragmentation degrees of irregular ground objects. These problems weak the ability of existing methods to distinguish ground objects with different fragmentation degrees. Therefore, this study proposed a new texture feature extraction method considering spatial continuity and gray diversity (SCGD). SCGD first connected all pixels in a neighborhood in series from end to end according to the row and column directions, and the diversities of the spatial continuity encoding in different directions were calculated by the Shannon index. Then, the Shannon index was used to calculate the gray diversity. Finally, SCGD calculated the weighted average of spatial continuity diversity and gray diversity to obtain the final texture feature values. Validation results indicated that SCGD can effectively distinguish ground objects with different fragmentation degrees, and its performance is better than that of traditional methods. As the spatial resolution decreases, its performance advantage becomes more obvious. Moreover, SCGD has great application potential in the field of ground object classification, and combining it with deep learning models will contribute to achieving the fine recognition of ground objects.

MVACLNet: A Multimodal Virtual Augmentation Contrastive Learning Network for Rumor Detection

In today’s digital era, rumors spreading on social media threaten societal stability and individuals’ daily lives, especially multimodal rumors. Hence, there is an urgent need for effective multimodal rumor detection methods. However, existing approaches often overlook the insufficient diversity of multimodal samples in feature space and hidden similarities and differences among multimodal samples. To address such challenges, we propose MVACLNet, a Multimodal Virtual Augmentation Contrastive Learning Network. In MVACLNet, we first design a Hierarchical Textual Feature Extraction (HTFE) module to extract comprehensive textual features from multiple perspectives. Then, we fuse the textual and visual features using a modified cross-attention mechanism, which operates from different perspectives at the feature value level, to obtain authentic multimodal feature representations. Following this, we devise a Virtual Augmentation Contrastive Learning (VACL) module as an auxiliary training module. It leverages ground-truth labels and extra-generated virtual multimodal feature representations to enhance contrastive learning, thus helping capture more crucial similarities and differences among multimodal samples. Meanwhile, it performs a Kullback–Leibler (KL) divergence constraint between predicted probability distributions of the virtual multimodal feature representations and their corresponding virtual labels to help extract more content-invariant multimodal features. Finally, the authentic multimodal feature representations are input into a rumor classifier for detection. Experiments on two real-world datasets demonstrate the effectiveness and superiority of MVACLNet on multimodal rumor detection.

Detect Gear Fault in Wind Turbine Gearbox based on Time Synchornous Averaging Without Phase Signal.

Wind energy is increasingly recognized as a worldwide sustainable and environmentally friendly power source. Nevertheless, an important barrier to further investment in wind energy is the large rate of failure that occurs to turbines. The gearbox, a crucial component, has a major effect on the performance of wind turbines. It consists of complex planetary and cylindrical gear systems, making it prone to failure and causing major defects in wind turbines. Therefore, there is an urgent need to minimize downtime and improve productivity in wind turbine gearbox operations. Recent decades have witnessed growing interest in fault diagnosis of gearboxes due to their widespread use and industry significance. The time synchronous averaging (TSA) method is widely used as a fundamental approach to detect faults in wind turbine gearboxes from mechanical vibration signals. Usually, applying this method requires a device to measure the vibration phase. Nevertheless, there are certain circumstances where installing a phase-measuring device might pose challenges. For instance, if the gearbox operates, it cannot be paused for installation. Additionally, if the gearboxes are enclosed, it becomes difficult to insert the device. The present paper presents an innovative technical approach to improve the time synchornous averaging method without requiring information about phase vibration. It also evaluates the effectiveness of the proposed method using feature values. An experimental test rig was set up to evaluate the algorithm's effectiveness, simulating different gear faults and load conditions.

A quantitative framework to support the decision between traditional, selective, and hybrid assembly

High-technological products often consist of several components, each with varying feature values that are critical to the assembly’s quality. As traditional manufacturing involves random selection of components, strict tolerance ranges are typically set to limit their feature variability, increasing scrap rates and lowering material efficiency. By smartly combining components based on their measured features, selective and hybrid assembly allow for more relaxed tolerances and more efficient use of raw materials without affecting the final product’s quality. However, the increased logistical complexity involves additional operational costs, complicating the choice for the best strategy.This study presents a decision-making model for selecting from traditional, selective, and hybrid assembly. The model is formulated by first defining the cost functions for the corresponding expected assembly cost, and then analytically determining the optimal choice boundaries between the three methods based on the involved process parameters. Estimation methods for the instance-specific process parameters are then discussed before validating the model with a fully elaborated case study.Results confirm the need for a quantitative framework to support an informed choice as this strongly depends on the involved parameter values. This study aims to pave the way towards more sustainable manufacturing by supporting practitioners to identify the conditions whereby intelligent assembly methods, in addition to allowing for scrap reduction and increased material efficiency, also represent the most cost-effective option.

Epileptic Seizure Detection from Decomposed EEG Signal through 1D and 2D Feature Representation and Convolutional Neural Network

Electroencephalogram (EEG) has emerged as the most favorable source for recognizing brain disorders like epileptic seizure (ES) using deep learning (DL) methods. This study investigated the well-performed EEG-based ES detection method by decomposing EEG signals. Specifically, empirical mode decomposition (EMD) decomposes EEG signals into six intrinsic mode functions (IMFs). Three distinct features, namely, fluctuation index, variance, and ellipse area of the second order difference plot (SODP), were extracted from each of the IMFs. The feature values from all EEG channels were arranged in two composite feature forms: a 1D (i.e., unidimensional) form and a 2D image-like form. For ES recognition, the convolutional neural network (CNN), the most prominent DL model for 2D input, was considered for the 2D feature form, and a 1D version of CNN was employed for the 1D feature form. The experiment was conducted on a benchmark CHB-MIT dataset as well as a dataset prepared from the EEG signals of ES patients from Prince Hospital Khulna (PHK), Bangladesh. The 2D feature-based CNN model outperformed the other 1D feature-based models, showing an accuracy of 99.78% for CHB-MIT and 95.26% for PHK. Furthermore, the cross-dataset evaluations also showed favorable outcomes. Therefore, the proposed method with 2D composite feature form can be a promising ES detection method.

A combination network based on graph modules for cardiac segmentation

BackgroundCardiac segmentation plays a crucial role in extracting cardiac substructures from computed tomography images, providing necessary information for subsequent cardiac disease diagnostics. Current research on cardiac segmentation algorithms faces challenges such as complex spatial distributions and limited labeled samples. The preferred method, convolutional neural network (CNN), excels at capturing features at various levels. However, the coarse convolutional operations of CNN often neglect valuable structural information. In contrast, graph neural network (GNN) excels at capturing complex spatial relationships among nodes by aggregating messages from graphs. MethodologyIn this work, a GNN-CNN combination network (GCCN) with enhanced feature extraction by GNN is proposed for cardiac segmentation. The combination network integrates the strengths of two approaches while compensating for each other's shortcomings. Firstly, 3D CT images are transformed into graphs, whose nodes with similar feature values within a certain spatial proximity are connected. Secondly, a lightweight graph attention network (LGAT) module is proposed for aggregating the messages of constructed graphs. Thirdly, the data, resampled from the feature-enhanced graphs, is fed into the 3D ResU-Net for the final segmentation process. ResultThe proposed method achieves an average Dice score of 92.98% which is 1.84% higher than 3D ResU-Net. The average Dice score of each cardiac substructure improved, particularly for the ascending aorta, achieving 96.35%, an 8.29% improvement over 3D ResU-Net. Additionally, our study demonstrates that GCCN has the improved adaptability in handling datasets with limited samples. ConclusionIn general, GCCN allows for the extraction of richer structural information from CT images, resulting in stronger feature representation of the data. Compared to pure convolutional methods, GCCN not only enhances segmentation accuracy but also exhibits greater adaptability to small-sample data. GCCN represents an exploration of applying graph neural networks to cardiac segmentation tasks from a novel perspective.

TREADS: Target Research for Anti-epileptic Drugs Using Data Science

Context: Epilepsy is a common neurological disease and is classified into different types based on features such as the kind of seizure, age of onset, part of brain effected, etc. There are nearly 30 approved anti-epileptic drugs (AEDs) for treating different epilepsies and each drug targets proteins exhibiting a specific molecular mechanism of action. There are many genes, proteins, and microRNAs known to be associated with different epileptic disorders. This rich information on epilepsy-associated data is not available at one single location and is scattered across multiple publicly available repositories. There is a need to have a single platform integrated with the data, as well as tools required for epilepsy research. Methods and Material: Text mining approaches are used to extract data from multiple biological sources. The data is curated and populated within an in-house developed epilepsy database. Machine-learning based models are built in-house to know the probability of a protein being druggable based on the significant protein features. A web interface is provided for the access of the epilepsy database as well as the ML-based tool developed in-house. Results: The epilepsy-associated data is made accessible through a web browser. For a protein of interest, the platform provides all the feature values, and the results generated using different machine learning models are displayed as visualization plots. Conclusions: To meet these objectives, we present TREADS, a platform for epilepsy research community, having both database and an ML-based tool for the study of AED targets. To access TREADS: https://treads-aer.cdacb.in

Machine learning-assisted prediction and optimization of solid oxide electrolysis cell for green hydrogen production

The solid oxide electrolysis cell (SOEC) holds great promise to efficiently convert renewable energy into hydrogen. However, traditional modeling methods are limited to a specific or reported SOEC system. Therefore, four machine learning models are developed to predict the performance of SOEC processes of various types, operating parameters, and feed conditions. The impact of these features on the SOEC's outputs is explained by the Shapley additive explanations and partial dependency plot analyses. The preferred model is integrated with a genetic algorithm to determine the optimal values of each input feature. Results show the improved extreme gradient enhanced regression (XGBoost) algorithm is the core of the machine learning model of the process since it has the highest R2 (> 0.95) in the three outputs. The electrolytic cell descriptors have a greater impact on the system performance, contributing up to 54.5%. The effective area, voltage, and temperature are the three most influential factors in the SOEC system, contributing 21.6%, 16.6%, and 13.0% to its performance. High temperature, high pressure, and low effective area are the most favorable conditions for H2 production rate. After conducting multi-objective optimization, the optimal current intensity and hydrogen production rate were determined to be 1.61 A/cm2 and 1.174 L/(h·cm2).