Search

Background: The ActiPASS software was developed from the open-source Acti4 activity classification algorithm for thigh-worn accelerometry. However, the original algorithm has not been validated in children or compared with a child-specific set of algorithm thresholds. This study aims to evaluate the accuracy of ActiPASS in classifying activity types in children using 2 published sets of Acti4 thresholds. Methods: Laboratory and free-living data from 2 previous studies were used. The laboratory condition included 41 school-aged children (11.0 [4.8] y; 46.5% male), and the free-living condition included 15 children (10.0 [2.6] y; 66.6% male). Participants wore a single accelerometer on the dominant thigh, and annotated video recordings were used as a reference. Postures and activity types were classified with ActiPASS using the original adult thresholds and a child-specific set of thresholds. Results: Using the original adult thresholds, the mean balanced accuracy (95% CI) for the laboratory condition ranged from 0.62 (0.56–0.67) for lying to 0.97 (0.94–0.99) for running. For the free-living condition, accuracy ranged from 0.61 (0.48–0.75) for lying to 0.96 (0.92–0.99) for cycling. Mean balanced accuracy for overall sedentary behavior (sitting and lying) was ≥0.97 (0.95–0.99) across all thresholds and conditions. No meaningful differences were found between the 2 sets of thresholds, except for superior balanced accuracy of the adult thresholds for walking under laboratory conditions. Conclusions: The results indicate that ActiPASS can accurately classify different basic types of physical activity and sedentary behavior in children using thigh-worn accelerometer data. [ABSTRACT FROM AUTHOR]

Modern manufacturing is the world's largest and most automated industrial sector. The rise of Industry 4.0 technologies such as Big Data, Internet of Things (IoT) devices, and Machine Learning has enabled a better connection with machines and factory systems. Data harvesting allowed for a more seamless and comprehensive implementation of the knowledge-based decision-making process. New models that provide a competitive edge must be created by combining the Lean paradigm with the new technologies of Industry 4.0. This paper presents novel computer-based vision models for automated detection and classification of damaged packages from intact packages. In high-volume production environments, the package manual inspection process through the human eye consumes inordinate amounts of time poring over physical packages. Our proposed three different computer-based vision approaches detect damaged packages to prevent them from moving to shipping operations that would otherwise incur waste in the form of wasted operating hours, wasted resources and lost customer satisfaction. The proposed approaches were carried out on a data set consisting of package images and achieved high precision, accuracy, and recall values during the training and validation stage, with the resultant trained YOLO v7 model. [ABSTRACT FROM AUTHOR]

Purpose of the article: Patients with penicillin allergy labels are more likely to have postoperative wound infections. When penicillin allergy labels are interrogated, a significant number of individuals do not have penicillin allergies and may be delabeled. This study was conducted to gain preliminary evidence into the potential role of artificial intelligence in assisting with perioperative penicillin adverse reaction (AR) evaluation. Material and methods: A single-centre retrospective cohort study of consecutive emergency and elective neurosurgery admissions was conducted over a two-year period. Previously derived artificial intelligence algorithms for the classification of penicillin AR were applied to the data. Results: There were 2063 individual admissions included in the study. The number of individuals with penicillin allergy labels was 124; one patient had a penicillin intolerance label. Of these labels, 22.4% were not consistent with classifications using expert criteria. When the artificial intelligence algorithm was applied to the cohort, the algorithm maintained a high level of classification performance (classification accuracy 98.1% for allergy versus intolerance classification). Conclusions: Penicillin allergy labels are common among neurosurgery inpatients. Artificial intelligence can accurately classify penicillin AR in this cohort, and may assist in identifying patients suitable for delabeling. [ABSTRACT FROM AUTHOR]

This paper presents a novel weighted likelihood ensemble approach for predicting pipe failures in water distribution networks (WDNs). The proposed method leverages ensemble modeling, specifically stacking, to enhance prediction capability. The study utilizes a data set of water pipe failures from 2006 to 2017, segmented into different time intervals. Various classification algorithms, including logistic regression (LR), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGB), are employed to predict failures within these segments. These individual models are then combined to create ensemble models. The results show that the stacked models consistently outperform the models that use the training data set as a whole. Along with traditional evaluation metrics, practical assessments are conducted, considering different percentages of pipes for replacement. These evaluations align with tactical and strategic maintenance plans. Remarkably, the most significant improvements are observed in models with lower replacement percentages. The novel aspect of this approach lies in assigning weights to prediction results from different models, each utilizing distinct time segments of data. By developing a meta-model with linear regression based on weighted likelihoods of pipe failures, this method provides valuable insights for asset managers and decision makers. It aids in prioritizing pipe rehabilitation programs, with the potential for further refinement as new failure data becomes available. [ABSTRACT FROM AUTHOR]

Efficiency and optimization in virtual reality (VR) technology is an urgent need, especially in the context of optimizing algorithms to recognize user emotions while using VR. Efficient VR technology can improve user experience and enable more immersive and responsive interactions. This study adopts the preferred reporting items for systematic reviews and metaanalyses (PRISMA) (2020) method to identify and analyze gaps in the existing literature, focusing on the optimization of electroencephalogram (EEG) signal classification algorithms to recognize VR users' emotions. The literature search was conducted through the Scopus database, with article selection based on the type of emotion classified, the classification method used, the limitations of the research, and the results obtained. Of the 1478 articles found, 74 articles passed the initial selection stage, and the final stage 13 articles were selected for further analysis. The selected articles provide important insights into the development of EEG classification algorithms for VR users, especially in multi-user settings. The findings identify potential and opportunities in the development of more efficient and accurate EEG signal classification algorithms for VR users. By focusing on emotion classification in a multi-user VR environment, this research contributes to improving the efficiency of VR technology and supporting a better and more responsive user experience. [ABSTRACT FROM AUTHOR]

Fault diagnosis plays an integral role in machine health monitoring. However, in practical applications, there are obvious differences in class distribution within the data, leading to poor performance of the algorithm in identifying a few classes. Meanwhile, overfitting and computational resource requirements have become a challenge. Recently, the stacking model has been promoted in the field of fault diagnosis, but its performance evaluation of stacking models in many literature is not comprehensive enough. In this paper, an Advanced Ensemble Trees model (AET) is proposed. The SMOTE (Synthetic Minority Oversampling Technique) resampling technique is used to optimise the dataset balance. Then, the advantages of Support Vector Machines (SVM) and multi-tree models are combined to form a robust base model using hyper-parameter tuning. Simple Logistic Regression (LR) is used as a meta-model to construct the new stacking model. Through extensive experimental validation, it is found that the AET model is close to 99% in several key performance metrics and outperforms existing machine learning methods and relatively short model training time. [ABSTRACT FROM AUTHOR]

Recognition technology based on millimeter wave radar (MMW) can operate in all-weather conditions and has received much attention in the field of intelligent connected vehicles (ICVs). However, the label information of the targets cannot be directly obtained from the original radar point clouds, making it necessary to develop advanced recognition algorithms. This paper proposes a target recognition algorithm based on machine learning that utilizes radar point clouds and leverages the radar reflection intensity to improve target recognition accuracy. Firstly, regional division and density clustering techniques are employed to preprocess the original point clouds from the MMW and segment them into meaningful regions, thereby reducing the computational burden; Secondly, relevant features are extracted from the processed radar point cloud, including radar scattering cross section and its related features. Finally, to improve target recognition accuracy, this paper proposes a grid search optimization principal component analysis support vector machine (GS-PCA-SVM) classification algorithm. The algorithm uses PCA to reduce the dimensionality of the data while preserving key information; then, it optimizes the parameters and kernel function of SVM by using the GS method to improve the performance of the classifier. The experimental results indicate that the recognition algorithm proposed in this paper achieves accuracies of 80%, 93%, and 95% on static, dynamic, and mixed datasets, respectively. Real vehicle experiments also prove that this algorithm has high accuracy and reliability when applied to ICV. [ABSTRACT FROM AUTHOR]

Copyright of Telecommunication Engineering is the property of Telecommunication Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Introduction: Freezing of gait (FOG) is a disabling symptom for people with Parkinson's disease (PwPD). Turning on the spot for one minute in alternating directions (360 turn) while performing a cognitive dual-task (DT) is a fast and sensitive way to provoke FOG. The FOG-index is a widely used wearable sensor-based algorithm to quantify FOG severity during turning. Despite that, the FOG-index's classification performance and criterion validity is not tested against the gold standard (i.e., video-rated time spent freezing). Therefore, this study aimed to evaluate the FOG-index's classification performance and criterion validity to assess FOG severity during 360 turn. Additionally, we investigated the FOG-index's optimal cutoff values to differentiate between PwPD with and without FOG. Methods: 164 PwPD self-reported the presence of FOG on the New Freezing of Gait Questionnaire (NFOGQ) and performed the DT 360 turn in the ON medication state while being videoed and wearing five wearable sensors. Two independent clinical experts rated FOG on video. ROC-AUC values assessed the FOG-index's classification accuracy against self-reported FOG and expert ratings. Spearman-rho was used to evaluate the correlation between expert and FOG-index ratings of FOG severity. Results: Twenty-eight patients self-reported FOG, while 104 were classified as a freezer by the experts. The FOG-index had limited classification agreement with the NFOGQ (AUC = 0.60, p = 0.115, sensitivity 46.4%, specificity 72.8%) and the experts (AUC = 0.65, p < 0.001, sensitivity 68.3%, specificity 61.7%). Only weak correlations were found between the algorithm outputs and expert ratings for FOG severity (rho = 0.13–0.38). Conclusion: A surprisingly large discrepancy was found between self-reported and expert-rated FOG during the 360 turning task, indicating PwPD do not always notice FOG in daily life. The FOG-index achieved suboptimal classification performance and poor criterion validity to assess FOG severity. Regardless, 360 turning proved a sensitive task to elicit FOG. Further development of the FOG-index is warranted, and long-term follow-up studies are needed to assess the predictive value of the 360 turning task for classifying FOG conversion. [ABSTRACT FROM AUTHOR]

Pharmacogenetics is a branch of genomic medicine aiming to personalize drug prescription guidelines based on individual genetic information. This concept might lead to a reduction in adverse drug reactions, which place a heavy burden on individual patients' health and the economy of the healthcare system. The aim of this study was to present insights gained from the pharmacogenetics-based clustering of over 500 patients from the Croatian population. The data used in this article were obtained by the pharmacogenetic testing of 522 patients from the Croatian population. The patients were clustered based on the genotypes of 28 pharmacologically relevant genes. Dimensionality reduction was employed using the UMAP algorithm, after which clusters were defined using HDBSCAN. Validation of clustering was performed by decision tree analysis and predictive modeling using the RandomForest, XGBoost, and ExtraTrees classification algorithms. The clustering algorithm defined six clusters of patients based on two UMAP components (silhouette score = 0.782). Decision tree analysis demonstrated CYP2D6 and SLCO1B1 genotypes as the main points of cluster determination. Predictive modeling demonstrated an excellent ability to discern the cluster of each patient based on all genes (avg. ROC-AUC = 0.998), CYP2D6 and SLCO1B1 (avg. ROC-AUC = 1.000), and CYP2D6 alone (avg. ROC-AUC = 0.910). Membership in each cluster provided clinically relevant information, in the context of ruling out certain favorable or unfavorable phenotypes. However, this study's main limitation is its cohort size. Through further research and investigation of a larger number of patients, more accurate and clinically applicable associations between pharmacogenetic genotypes and phenotypes might be discovered. [ABSTRACT FROM AUTHOR]

The identification of significant predictors with higher model performance is the key objective in classification domain. A machine learning-based variable selection technique termed as CARS-Logistic model is proposed by coupling competitive adaptive re-weighted sampling(CARS) and logistic regression for binary classification. Based on five assessment criteria, the proposed method is found to be more efficient than Forward selection logistic regression model. The CARS-Logistic model is executed to determine the significant factors of perinatal mortality in Pakistan. The identified hazards communicated social, cultural, financial, and health-related characteristics which contain key information about perinatal mortality in Pakistan for policymakers. [ABSTRACT FROM AUTHOR]

In this work, the potential of cryogen-free low-field 1H-NMR spectroscopy combined with chemometrics was examined to authenticate honeys based on their botanical origin and to determine the presence and content of low-priced sugar syrups in genuine honeys. The multivariate classification data algorithms of PCA, HCA and SIMCA were constructed over the integrated data of benchtop NMR spectroscopy, by using 95 genuine honey samples from 15 different types of monofloral, polyfloral and honeydew honeys. The supervised PLS-R models were also constructed to identify quantitative detection limits for the cheap sugar syrups in adulterated honey–sugar syrup sample sets. The monofloral honeys were especially forecasted with an accuracy of 100% by the SIMCA model, and they offered a distinct demarcation of the honeys and were employed to classify honeys. The minimum detection limits for syrup adulterants were 0.96% in monofloral honeys, 1.55% in polyfloral honeys and 1.93% in honeydew honeys, respectively, with high R2 values (> 0.9970). It is concluded that the profiles of major carbohydrate molecules and minor components by low-field 1H-NMR spectroscopy with chemometrics could be a powerful and potential device for the demarcation of honeys from diverse botanical origins and the quantification of low-priced sugar syrups in pure honeys. [ABSTRACT FROM AUTHOR]

Studying sleep stages is crucial for understanding sleep architecture, which can help identify various health conditions, including insomnia, sleep apnea, and neurodegenerative diseases, allowing for better diagnosis and treatment interventions. In this paper, we explore the effectiveness of generalized weighted permutation entropy (GWPE) in distinguishing between different sleep stages from EEG signals. Using classification algorithms, we evaluate feature sets derived from both standard permutation entropy (PE) and GWPE to determine which set performs better in classifying sleep stages, demonstrating that GWPE significantly enhances sleep stage differentiation, particularly in identifying the transition between N1 and REM sleep. The results highlight the potential of GWPE as a valuable tool for understanding sleep neurophysiology and improving the diagnosis of sleep disorders. [ABSTRACT FROM AUTHOR]

Tornadoes are severe convective weather exhibiting localized and sudden occurrences. Weather radar is widely regarded as the most effective tool for monitoring tornadoes and issuing early warnings. Dual-polarization updating has significantly improved tornado prediction and forecasting abilities. This article proposes an innovative tornado detection algorithm based on XGBoost which is suitable for dual-polarization radar data, was upgraded and has been used in China since 2019, and has been applied in the Tornado Key Open Laboratory of the China Meteorological Administration. The characteristics associated with the velocity attributes, reflectivity, velocity spectrum width, differential reflectivity, and correlation coefficient are used in the algorithm to achieve real-time tornado detection. Experimental evaluations have demonstrated that the proposed algorithm significantly improves tornado detection rates and leading times. Compared with the traditional TDA-RF algorithm based on Doppler weather radar data, the TDA-XGB algorithm introduces dual polarization parameters (such as differential reflectivity and the correlation coefficient), which effectively improves tornado identification performance. In addition, the TDA-XGB algorithm combines artificial intelligence-assisted learning to optimize the traditional algorithm based on the tornado vortex signature (TVS) and tornado debris signature (TDS), further improving the detection effect. Furthermore, the algorithm provides classification probabilities in the genesis and evolution of tornadoes, thereby supporting forecasters in their efforts to anticipate and issue timely tornado warnings. [ABSTRACT FROM AUTHOR]

People tend to share their opinions on social media daily. This text needs to be accurately mined for different purposes like enhancements in services and/or products. Mining and analyzing Arabic text have been a big challenge due to many complications inherited in Arabic language. Although, many research studies have already investigated the Arabic text sentiment analysis problem, this paper investigates the specific research topic that addresses Arabic comparative opinion mining. This research topic is not widely investigated in many research studies. This paper proposes a lexicon-based framework which includes a set of proposed algorithms for the mining and analysis of Arabic comparative sentences. The proposed framework comprises a set of contributions including an Arabic comparative sentence keywords lexicon and a proposed algorithm for the identification of Arabic comparative sentences, followed by a second proposed algorithm for the classification of identified comparative sentences into different types. The framework also comprises a third proposed algorithm that was developed to extract relations between entities in each of the identified comparative sentence types. Finally, two proposed algorithms were developed for the extraction of the preferred entity in each sentence type. The framework was evaluated using three different Arabic language datasets. The evaluation metrics used to obtain the evaluation results include precision, recall, F-score, and accuracy. The average values of the evaluation metrics for the proposed sentences identification algorithm reached 97%. The average evaluation values of the evaluation metrics for the proposed sentence type identification algorithm reached 96%. Finally, the average results showed 97% relation word extraction precision for the proposed relation extraction algorithm. [ABSTRACT FROM AUTHOR]

Accurate indoor positioning is essential for many applications. However, current methods often fall short in complex environments due to signal fluctuations. We propose a new indoor positioning approach, that is, improved sequential deep learning (ISDL), to address this issue. First, we apply sequential classification algorithms to progressively narrow the search space, reducing potential location regions into smaller neighborhoods. Next, we combine a deep neural network (DNN) with Weighted K-Nearest Neighbors (WKNN) to refine the final location prediction. Then, we validate our method using the publicly available U J I n d o o r L o c dataset, demonstrating superior accuracy compared to existing methods. Specifically, we achieved 95% floor prediction accuracy and reduced the average positioning error to just 7.82 m. By combining sequential classification and the DNN-WKNN hybrid model, we achieve better localization in complex indoor environments. This system offers practical improvements for real-time location-based services and other applications requiring precise indoor positioning. [ABSTRACT FROM AUTHOR]

The increasing demand for hazelnut kernels is favoring an upsurge in hazelnut cultivation worldwide, but ongoing climate change threatens this crop, affecting yield decreases and subject to uncontrolled pathogen and parasite attacks. Technical advances in precision agriculture are expected to support farmers to more efficiently control the physio-pathological status of crops. Here, we report a straightforward approach to monitoring hazelnut trees in an open field, using aerial multispectral pictures taken by drones. A dataset of 4112 images, each having 2Mpixel resolution per tree and covering RGB, Red Edge, and near-infrared frequencies, was obtained from 185 hazelnut trees located in two different orchards of the Piedmont region (northern Italy). To increase accuracy, and especially to reduce false negatives, the image of each tree was divided into nine quadrants. For each quadrant, nine different vegetation indices (VIs) were computed, and in parallel, each tree quadrant was tagged as "healthy/unhealthy" by visual inspection. Three supervised binary classification algorithms were used to build models capable of predicting the status of the tree quadrant, using the VIs as predictors. Out of the nine VIs considered, only five (GNDVI, GCI, NDREI, NRI, and GI) were good predictors, while NDVI SAVI, RECI, and TCARI were not. Using them, a model accuracy of about 65%, with 13% false negatives was reached in a way that was rather independent of the algorithms, demonstrating that some VIs allow inferring the physio-pathological condition of these trees. These achievements support the use of drone-captured images for performing a rapid, non-destructive physiological characterization of hazelnut trees. This approach offers a sustainable strategy for supporting farmers in their decision-making process during agricultural practices. [ABSTRACT FROM AUTHOR]

Curcumae Longae Rhizoma (CLRh), Curcumae Radix (CRa), and Curcumae Rhizoma (CRh), derived from the different medicinal parts of the Curcuma species, are blood-activating analgesics commonly used for promoting blood circulation and relieving pain. Due to their certain similarities in chemical composition and pharmacological effects, these three herbs exhibit a high risk associated with mixing and indiscriminate use. The diverse methods used for distinguishing the medicinal origins are complex, time-consuming, and limited to intraspecific differentiation, which are not suitable for rapid and systematic identification. We developed a rapid analysis method for identification of affinis and different medicinal materials using attenuated total reflection-Fourier-transform infrared spectroscopy (ATR-FTIR) combined with machine learning algorithms. The original spectroscopic data were pretreated using derivatives, standard normal variate (SNV), multiplicative scatter correction (MSC), and smoothing (S) methods. Among them, 1D + MSC + 13S emerged as the best pretreatment method. Then, t-distributed stochastic neighbor embedding (t-SNE) was applied to visualize the results, and seven kinds of classification models were constructed. The results showed that support vector machine (SVM) modeling was superior to other models and the accuracy of validation and prediction was preferable, with a modeling time of 127.76 s. The established method could be employed to rapidly and effectively distinguish the different origins and parts of Curcuma species and thus provides a technique for rapid quality evaluation of affinis species. [ABSTRACT FROM AUTHOR]

The purpose of our study was to evaluate the accuracy with which classification algorithms could distinguish among standing postures based on center-of-pressure (CoP) trajectories. We performed a secondary analysis of published data from three studies: Study A) assessment of balance control on firm or foam surfaces with eyes-open or closed, Study B) quantification of postural sway in forward–backward and side-to-side directions during four standing-balance tasks that differed in difficulty, and Study C) an evaluation of the impact of two modes of transcutaneous electrical nerve stimulation on balance control in older adults. Three classification algorithms (decision tree, random forest, and k-nearest neighbor) were used to classify standing postures based on the extracted features from CoP trajectories in both the time and time–frequency domains. Such classifications enable the identification of differences and similarities in control strategy. Our results, especially those involving time–frequency features, demonstrated that distinct CoP trajectories could be identified from the extracted features in all conditions and postures in each study. Although the overall classification accuracy was similar using time–frequency features (~ 86%) for the three studies, there were substantial differences in accuracy across conditions and postures in Studies A and B but not in Study C. Nonetheless, the models were far superior to the published results with conventional metrics in distinguishing between the conditions and postures. Moreover, a Shapley Additive exPlanation analysis was able to identify the most important features that contributed to the classification performance of the models. [ABSTRACT FROM AUTHOR]

Background: The age distribution and diversity of the VA Million Veteran Program (MVP) cohort make it a valuable resource for studying the genetics of Alzheimer's disease (AD) and related dementias (ADRD). Objective: We present and evaluate the performance of several International Classification of Diseases (ICD) code-based classification algorithms for AD, ADRD, and dementia for use in MVP genetic studies and other studies using VA electronic medical record (EMR) data. These were benchmarked relative to existing ICD algorithms and AD-medication-identified cases. Methods: We used chart review of n = 103 MVP participants to evaluate diagnostic utility of the algorithms. Suitability for genetic studies was examined by assessing association with APOE ε4, the strongest genetic AD risk factor, in a large MVP cohort (n = 286 K). Results: The newly developed MVP-ADRD algorithm performed well, comparable to the existing PheCode dementia algorithm (Phe-Dementia) in terms of sensitivity (0.95 and 0.95) and specificity (0.65 and 0.70). The strongest APOE ε4 associations were observed in cases identified using MVP-ADRD and Phe-Dementia augmented with medication-identified cases (MVP-ADRD or medication, p = 3.6 ×10−290; Phe-Dementia or medication, p = 1.4 ×10−290). Performance was improved when cases were restricted to those with onset age ≥60. Conclusions: We found that our MVP-developed ICD-based algorithms had good performance in chart review and generated strong genetic signals, especially after inclusion of medication-identified cases. Ultimately, our MVP-derived algorithms are likely to have good performance in the broader VA, and their performance may also be suitable for use in other large-scale EMR-based biobanks in the absence of definitive biomarkers such as amyloid-PET and cerebrospinal fluid biomarkers. [ABSTRACT FROM AUTHOR]

• CD4+ T cells exhibit different immune responses to different diseases. • The expression profile data of CD4+ T cells on various diseases was deeply analyzed. • Special expression patterns were discovered for different diseases. CD4+ T cells play a pivotal role in the immune system, particularly in adaptive immunity, by orchestrating and enhancing immune responses. CD4+ T cell-related immune responses exhibit diverse characteristics in different diseases. This study utilizes gene expression analysis of CD4+ T cells to classify and understand complex diseases. We analyzed the dataset consisting of samples from various diseases, including cancers, metabolic disorders, circulatory and respiratory diseases, and digestive ailments, as well as 53 healthy controls. Each sample contained expression data for 22,881 genes. Four feature ranking algorithms, incremental feature selection method, synthetic minority oversampling technique, and four classification algorithms were utilized to pinpoint essential genes, extract classification rules and build efficient classifiers. The following analysis focused on genes across rules, such as AK4, CALU, LINC01271 , and RUSC1-AS1. AK4 and CALU show fluctuating levels in diseases like asthma, Crohn's disease, and breast cancer. The analysis results and existing research suggest that they may play a role in these diseases. LINC01271 generally has higher expression in conditions including asthma, Crohn's disease, and diabetes. RUSC1-AS1 is more expressed in chronic diseases like asthma and Crohn's, but less in acute illnesses like tonsillitis and influenza. This highlights the distinct roles of these genes in different diseases. Our approach highlights the potential for developing novel therapeutic strategies based on the transcriptional profiles of CD4+ T cells. [ABSTRACT FROM AUTHOR]

Objective: Acute kidney injury (AKI) is a frequent complication in critically ill patients, affecting up to 50% of patients in the intensive care units. The lack of standardized and open-source tools for applying the Kidney Disease Improving Global Outcomes (KDIGO) criteria to time series, requires researchers to implement classification algorithms of their own which is resource intensive and might impact study quality by introducing different interpretations of edge cases. This project introduces pyAKI, an open-source pipeline addressing this gap by providing a comprehensive solution for consistent KDIGO criteria implementation. Materials and methods: The pyAKI pipeline was developed and validated using a subset of the Medical Information Mart for Intensive Care (MIMIC)-IV database, a commonly used database in critical care research. We constructed a standardized data model in order to ensure reproducibility. PyAKI implements the Kidney Disease: Improving Global Outcomes (KDIGO) guideline on AKI diagnosis. After implementation of the diagnostic algorithm, using both serum creatinine and urinary output data, pyAKI was tested on a subset of patients and diagnostic accuracy was compared in a comparative analysis against annotations by physicians. Results: Validation against expert annotations demonstrated pyAKI's robust performance in implementing KDIGO criteria. Comparative analysis revealed its ability to surpass the quality of human labels with an accuracy of 1.0 in all categories. Discussion: The pyAKI pipeline is the first open-source solution for implementing KDIGO criteria in time series data. It provides a standardized data model and a comprehensive solution for consistent AKI classification in research applications for clinicians and data scientists working with AKI data. The pipeline's high accuracy make it a valuable tool for clinical research and decision support systems. Conclusion: This work introduces pyAKI as an open-source solution for implementing the KDIGO criteria for AKI diagnosis using time series data with high accuracy and performance. [ABSTRACT FROM AUTHOR]

Data classification is an important research direction in machine learning. In order to effectively handle extensive datasets, researchers have introduced diverse classification algorithms. Notably, Kernel Extreme Learning Machine (KELM), as a fast and effective classification method, has received widespread attention. However, traditional KELM algorithms have some problems when dealing with large-scale data, such as the need to adjust hyperparameters, poor interpretability, and low classification accuracy. To address these problems, this paper proposes an Enhanced Adaptive Whale Optimization Algorithm to optimize Kernel Extreme Learning Machine (EAWOA-KELM). Various methods were used to improve WOA. As a first step, a novel adaptive perturbation technique employing T-distribution is proposed to perturb the optimal position and avoid being trapped in a local maximum. Secondly, the WOA's position update formula was modified by incorporating inertia weight ω and enhancing convergence factor α, thus improving its capability for local search. Furthermore, inspired by the grey wolf optimization algorithm, use 3 excellent particle surround strategies instead of the original random selecting particles. Finally, a novel Levy flight was implemented to promote the diversity of whale distribution. Results from experiments confirm that the enhanced WOA algorithm outperforms the standard WOA algorithm in terms of both fitness value and convergence speed. EAWOA demonstrates superior optimization accuracy compared to WOA across 21 test functions, with a notable edge on certain functions. The application of the upgraded WOA algorithm in KELM significantly improves the accuracy and efficiency of data classification by optimizing hyperparameters. This paper selects 7 datasets for classification experiments. Compared with the KELM optimized by WOA, the EAWOA optimized KELM in this paper has a significant improvement in performance, with a 5%-6% lead on some datasets, indicating the effectiveness of EAWOA-KELM in classification tasks. [ABSTRACT FROM AUTHOR]

To enhance the accuracy of motor imagery(MI)EEG signal recognition, two methods, namely power spectral density and wavelet packet decomposition combined with a common spatial pattern, were employed to explore the feature information in depth in MI EEG signals. The extracted MI EEG signal features were subjected to series feature fusion, and the F-test method was used to select features with higher information content. Here regarding the accuracy of MI EEG signal classification, we further proposed the Platt Scaling probability calibration method was used to calibrate the results obtained from six basic classifiers, namely random forest (RF), support vector machines (SVM), Logistic Regression (LR), Gaussian naïve bayes (GNB), eXtreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LightGBM). From these 12 classifiers, three to four with higher accuracy were selected for model fusion. The proposed method was validated on Datasets 2a of the 4th International BCI Competition, achieving an average accuracy of MI EEG data of nine subjects reached 91.46%, which indicates that model fusion was an effective method to improve classification accuracy, and provides some reference value for the research on MI brain-machine interface. [ABSTRACT FROM AUTHOR]

In this study, 4 classification algorithms were employed to characterize the influence of road determinants on roadway crash severity with actual crash data. The crash data were obtained from crash records in Texas, USA, from January 2020 to April 2021. The prediction model of crash severity utilized 12 road-related features—including shoulder types, shoulder width, and curb types—as well as 10 other features—such as weather and illumination conditions—as input features. Three crash severity levels—"Minor Damage," "Moderate Damage," and "Severe Damage"—were used as output features. Decision tree, support vector machines, and multi-layer perceptron were employed to compare their prediction performance with the XGBoost model. The results show that the XGBoost model yields the best performance among the 4 algorithms. The overall accuracy, average precision, average recall, and average F1 score of the XGBoost model were 82.65%, 0.83, 0.82, and 0.82, respectively. Besides, SHapley Additive exPlanations (SHAP) and partial dependence plots were used to interpret the model results. Among the road-related features, the most influential one is the median width. Greater crash severity is related to paved right shoulder and curb. These findings are helpful for the design and planning of road safety. [ABSTRACT FROM AUTHOR]

The American College of Cardiology/American Heart Association introduced the Predicting Risk of Cardiovascular EVENTs (PREVENT™) algorithm to estimate the 10-year risk of developing cardiovascular disease. We aimed to assess the cardiovascular risk (CVR) reclassification among rheumatoid arthritis (RA) patients using traditional CVR algorithms—the 2024 PREVENT™ and the 2013 Atherosclerotic Cardiovascular Disease (ASCVD)—and the presence of carotid plaque (CP). This was a cross-sectional study nested of a RA patients' cohort. A certified radiologist performed a high-resolution B-mode carotid ultrasound to identify the presence of CP. The CVR evaluation was performed by a cardiologist, blinded to carotid ultrasound results, using the PREVENT™ and the ASCVD algorithms. Cohen's kappa (k) coefficient assessed concordance between high-risk classification by CVR algorithms and CP presence. ROC curve analysis evaluated the algorithms' capacity to identify RA patients with CP. The cutoff point was determined by the Youden-Index, with p < 0.05 as statistically significant. A total of 210 RA patients were included. The reclassification of CVR due to CP was 34.3% for the PREVENT™ algorithm and 30.0% for the ASCVD algorithm. Of these, 44.4% and 71.4%, respectively, were initially classified as low risk. Concordance between CVR algorithms and carotid ultrasound showed slight agreement (k = 0.032 and k = 0.130, respectively). The PREVENT™ algorithm did not identify more than one-third of high-CVR RA patients with indication of starting statin therapy based on carotid ultrasound findings. The PREVENT™ and ASCVD algorithms showed poor performance in identifying RA patients with CP. Key Points • The presence of CP was identified in more than a third of the evaluated RA patients (35.7%), classifying them as high CVR. • CVR reclassification by the presence of CP was observed in 34.3% RA patients with the PREVENTTM algorithm and in 30.0% RA patients with the ASCVD algorithm. • Most of the reclassified patients belonged to the low-risk category, 44.4% with the PREVENTTM algorithm and 71.4% with the ASCVD algorithm. • When evaluating the concordance between the ASCVD algorithm and the carotid ultrasound for high-risk classification, a slight agreement was found (k = 0.130). [ABSTRACT FROM AUTHOR]

Kernel-based online classification algorithms, such as the Perceptron, NORMA, and passive-aggressive, are renowned for their computational efficiency but have been criticized for slow convergence. However, the parallel projection algorithm, within the adaptive projected subgradient method framework, exhibits accelerated convergence and enhanced noise resilience. Despite these advantages, a specific sparsification procedure for the parallel projection algorithm is currently absent. Additionally, existing online classification algorithms, including those mentioned earlier, heavily rely on the kernel width parameter, rendering them sensitive to its choices. In an effort to bolster the performance of these algorithms, we propose a two-stage classification algorithm within the Cartesian product space of reproducing kernel Hilbert spaces. In the initial stage, we introduce an online double-kernel classifier with parallel projection. This design aims not only to improve convergence but also to address the sensitivity to kernel width. In the subsequent stage, the component with a larger kernel width remains fixed, while the component with a smaller kernel width undergoes updates. To promote sparsity and mitigate model complexity, we incorporate the projection-along-subspace technique. Moreover, for enhanced computational efficiency, we integrate the set-membership technique into the updates, selectively exploiting informative vectors to improve the classifier. The monotone approximation of the proposed classifier, based on the designed ϵ -insensitive function, is presented. Finally, we apply the proposed algorithm to equalize a nonlinear channel. Simulation results demonstrate that the proposed classifier achieves faster convergence and lower misclassification error with comparable model complexity. [ABSTRACT FROM AUTHOR]

The identification of positive selection has been framed as a classification task, with Convolutional Neural Networks (CNNs) already outperforming summary statistics and likelihood-based approaches in accuracy. Despite the prevalence of CNN-based methods that manipulate the pixels of images representing raw genomic data as a preprocessing step to improve classification accuracy, the efficacy of these pixel-rearrangement techniques remains inadequately examined, particularly in the presence of confounding factors like population bottlenecks, migration and recombination hotspots. We introduce a set of pixel rearrangement algorithms aimed at enhancing CNN classification accuracy in detecting selective sweeps. These algorithms are employed to assess the performance of four CNN models for selective sweep detection. Our findings illustrate that the judicious application of rearrangement algorithms notably enhances the overall classification accuracy of a CNN across various datasets simulating confounding factors. We observed that sorting the columns of the genomic matrices has higher on CNN performance than rearranging the sequences. To some extent, these rearrangement algorithms are more robust to misspecified demographic models compared with the utilization of the default preprocessing algorithm as suggested by the respective authors of each CNN architecture. We provide the data rearrangement algorithms as a distinct package available for download at: https://github.com/Zhaohq96/Genetic-data-rearrangement. • Data rearrangement algorithms can boost the overall classification accuracy of CNNs in identifying selective sweeps. • To some extent, data rearrangement algorithms improve classification robustness to demographic model misspecification. • Suitable rearrangement algorithms per CNN are robust to varying genomic window sizes. [ABSTRACT FROM AUTHOR]

SARS-CoV-2 has posed serious global health challenges not only because of the high degree of virus transmissibility but also due to its severe effects on the respiratory system, such as inducing changes in multiple organs through the ACE2 receptor. This virus makes changes to gene expression at the single-cell level and thus to cellular functions and immune responses in a variety of cell types. Previous studies have not been able to resolve these mechanisms fully, and so our study tries to bridge knowledge gaps about the cellular responses under conditions of infection. We performed single-cell RNA-sequencing of nasopharyngeal swabs from COVID-19 patients and healthy controls. We assembled a dataset of 32,588 cells for 58 subjects for analysis. The data were sorted into eight cell types: ciliated, basal, deuterosomal, goblet, myeloid, secretory, squamous, and T cells. Using machine learning, including nine feature ranking algorithms and two classification algorithms, we classified the infection status of single cells and analyzed gene expression to pinpoint critical markers of SARS-CoV-2 infection. Our findings show distinct gene expression profiles between infected and uninfected cells across diverse cell types, with key indicators such as FKBP4, IFITM1, SLC35E1, CD200R1, MT-ATP6, KRT13, RBM15, and FTH1 illuminating unique immune responses and potential pathways for viral spread and immune evasion. The machine learning methods effectively differentiated between infected and non-infected cells, shedding light on the cellular heterogeneity of SARS-CoV-2 infection. The findings will improve our knowledge of the cellular dynamics of SARS-CoV-2. • The mechanism underlying SARS-CoV-2 has not been fully uncovered. • The single-cell RNA-sequencing data of nasopharyngeal swabs was deeply investigated using machine learning methods. • Some key genes were discovered, illuminating unique immune responses and pathways for viral spread and immune evasion. • Efficient classifiers were constructed for predicting SARS-CoV-2 infected cells. [ABSTRACT FROM AUTHOR]

This paper estimates the impact of family ancestry on the educational outcomes of a cohort of Brazilian students. Based on longitudinal data with student identification, we apply an algorithm of surname classification that assigns the student, based on the surnames of his/her parents and grandparents, to one of the following ancestry groups: Iberian, Japanese, Italian, German, Eastern European and Syrian-Lebanese. Our identification strategy relies on the epidemiological approach, controlling for individual trajectory since birth and the persistence of local institutions established during the Era of Mass Immigration to Brazil in the 19th and 20th centuries. We show that, despite slight or absent differences in preschool attendance rate, students with non-Iberian ancestry obtain statistically and substantively higher promotion rates and scores on 3rd and 5th grade nationwide standardized tests. [ABSTRACT FROM AUTHOR]

In contemporary computer vision, deep learning-based real-time single image super-resolution approaches have gained significant attention for their ability to enhance the resolution of images in real time. These approaches are interconnected with various other computer vision domains, including image segmentation and object detection. Numerous surveys have summarized the state of the image SR domain. However, there is no survey that specifically addresses real-time single image SR on IoT devices. Therefore, in this study, we aim to explore strategies, identify the technical challenges, and outline the future directions of SR research, with a special emphasis on real-time super-resolution techniques. We begin with an overview of the core concepts related to real-time SR, recent challenges, and algorithm classification and delve into potential application scenarios that merit attention. Additionally, we explore the challenges and identify promising research areas related to real-time SR specifically related to IoT devices, highlighting potential advancements, limitations, and opportunities for future innovation in this rapidly evolving field. [ABSTRACT FROM AUTHOR]

Driven by the substantial profits, the evolution of Portable Executable (PE) malware has posed persistent threats. PE malware classification has been an important research field, and numerous classification methods have been proposed. With the development of machine learning, learning-based static classification methods achieve excellent performance. However, most existing methods cannot meet the requirements of industrial applications due to the limited resource consumption and concept drift. In this article, we propose a fast, high-accuracy, and robust FCG-based PE malware classification method. We first extract precise function call relationships through code and data cross-referencing analysis. Then we normalize function names to construct a concise and accurate function call graph. Furthermore, we perform topological analysis of the function call graph using social network analysis techniques, thereby enhancing the program function call features. Finally, we use a series of machine learning algorithms for classification. We implement a prototype system named MalSensor and compare it with nine state-of-the-art static PE malware classification methods. The experimental results show that MalSensor is capable of classifying a malicious file in 0.7 seconds on average with up to 98.35% accuracy, which represents a significant advantage over existing methods. [ABSTRACT FROM AUTHOR]

This study investigates the impact of patient stress on COVID-19 screening. An attempt was made to measure the level of anxiety of individuals undertaking rapid tests for SARS-CoV-2. To this end, a galvanic skin response (GSR) sensor that was connected to a microcontroller was used to record the individual stress levels. GSR data were collected from 51 individuals at SARS-CoV-2 testing sites. The recorded data were then compared with theoretical estimates to draw insights into stress patterns. Machine learning analysis was applied for the optimization of the sensor results. Classification algorithms allowed the automatic reading of the sensor results and individual identification as "stressed" or "not stressed". The findings confirmed the initial hypothesis that there was a significant increase in stress levels during the rapid test. This observation is critical, as heightened anxiety may influence a patient's willingness to participate in screening procedures, potentially reducing the effectiveness of public health screening strategies. [ABSTRACT FROM AUTHOR]

Cardiovascular disease causes more deaths than any other cause in the globe. The present method of illness identification involves electrocardiogram (ECG) analysis, a medical monitoring gadget that captures heart activity. Regrettably, a great deal of medical resources is required to locate specialists in ECG data. Consequently, ML feature detection in ECG is rapidly gaining popularity. Human intervention is required for "feature recognition, complex models, and lengthy training timeframes"--limitations that are inherent to these traditional approaches. Using the "MIT-BIH Arrhythmia" database, this study presents five distinct categories of heartbeats and the efficient and effective deep-learning (DL) classification algorithms that go along with them. The five types of pulse features are classified experimentally using the wavelet self-adaptive threshold denoising method. Models such as AlexNet and CNN are employed in this dataset. For model evaluation use some performance metrics, like recall, accuracy, precision, and f1-score. The suggested Alex Net model achieves an overall classification accuracy of 99.68%, while the recommended CNN model achieves an accuracy of 99.89%. The end findings demonstrate that the suggested models outperform the current model on several performance criteria and are more efficient overall. With its accurate categorization, important medical resources are better preserved, which has a positive effect on the practice of medicine. [ABSTRACT FROM AUTHOR]

Automated understanding of activities in enterprises is challenging due to a lack of domain specifications and a lack of domain ontologies. The goal of this research is to develop a method to extract elements of domain-specific processes from textual documents in unstructured English and form domain dataset ontologies. In order to achieve the goal, the related work on discourse analysis and business process modelling have been considered. The prominent technologies for implementation of the proposed method are machine learning, including classification algorithms and natural language processing using a large language model. The first experimental results are presented, and further research is discussed. Potentially, the method proposed can be implemented as a part of some assisting tool for system analysts and can support an analysis of the domain-specific information by providing contextual information from this and potentially related domains. [ABSTRACT FROM AUTHOR]

Helping Moroccan students choose their high school presents significant challenges influenced by a variety of factors, including academic achievement, potential, and environmental influences. This study addresses these complexities using advanced data analytics and intelligent algorithms. We collected and examined authentic data from various secondary schools across Morocco, using the MASSAR system, a centralized education platform. To ensure robust model evaluation and optimized performance, we implemented 5-fold cross-validation and extensive hyper-parameter tuning for both support vector machine (SVM) and neural network models. Advanced classification algorithms, including hybrid learning techniques with neural networks and SVM algorithms, were applied, resulting in outstanding precision measures: 99.17% accuracy, 99.20% precision, 99.37% recall, 99.28% F1 score, and 0.99 area under the curve (AUC). By integrating this hybrid learning approach, powered by big data technologies such as Hadoop and Hadoop Distributed File System (HDFS), we accurately predict student choices and offer valuable academic advice. The use of a Hadoop cluster accelerated execution time by 40%. This pioneering merger underlines the adaptability and effectiveness of our approach to meeting the real-world educational challenges specific to the Moroccan context. [ABSTRACT FROM AUTHOR]

The challenge of getting big amounts of high-quality labeled data is compounded by the fact that data labeling is often subjective and requires significant human effort. In many cases, the quality of the labeled data depends entirely on the expertise and experience of human annotators, making it challenging to ensure labeling accuracy in large and dynamic datasets. Moreover, there may be a significant delay between the arrival of a new instance and its manual labeling. This paper explores the use of fully unsupervised feature selection algorithms in non-stationary data streams, where the importance of features may change over time. We introduce a novel feature selection algorithm called Online Fast FEa-ture SELection-OFFESEL, which calculates the feature importance scores in each incoming window based on their mean normalized values and without using any class labels. We evaluate OFFESEL on 17 benchmark data streams, both stationary and non-stationary, using popular online classifiers like PerceptronMask, VFDT, Online Boosting, and Linear SVM. We compare OFFESEL to several other feature selection algorithms, including state-of-the-art supervised ones like FIRES and ABFS, as well as popular unsupervised ones like MCFS, LS, and Max Variance, which we adapted to data streams. Our results indicate that OFFESEL outperforms all supervised and unsupervised feature selection algorithms in terms of classification accuracy. Specifically, OFFESEL preserves the accuracy level of the supervised FIRES algorithm, which proved more accurate than ABFS in our experiments, while maintaining the accuracy level achieved by the unsupervised Max Variance algorithm. Moreover, OFFESEL requires even less computation time than Max Variance and shows high stability on stationary datasets. Overall, our study demonstrates the potential benefits of using unlabeled data for feature ranking and selection in dynamic data streams. [ABSTRACT FROM AUTHOR]

Feature selection consider one of the essential pre-processing stage of the classification task in machine learning. The datasets that used in classification contain irrelevant features that may directly affect the performance of the used classifiers. The classification accuracy could be race using appropriate feature selector by reducing the number of the extracted features from the datasets. The common and the powerful algorithms that successfully used for feature selection task is the optimization algorithms. Based on the searching strategy of butterflies, the Butterfly Optimization Algorithm (BOA) is a meta-heuristic swarm intelligence algorithm. Because of its performance, BOA has been applied to a wide range of optimization problems. However, BOA has limitations such as reduced population variety and a tendency to become locked in a local optimum. Besides, it suffers in converges speed, accuracy, and precision of the optimal objective value when optimizing high dimensional problems. Therefore, this paper proposed an accurate algorithm based on BOA and Sine-Cosine Algorithm called BOA-SC. The BOA first improved via the update equations then hybrid with SC to enhance the local search stage for better optimization results. Using the improvement strategy and SC enhance the performance of BOA and solve the lower coverage and local optima issues that BOA suffers from. The performance of the proposed hybrid algorithm is evaluated using two assessments via converges speed, the accuracy, and precision of the optimal objective value. First, 23 benchmark functions used to evaluate proposed algorithm that achieved a high optimization result comparing with six most recent metaheuristic algorithms puzzle optimization algorithm (POA), northern goshawk optimization (NGO), coati optimization algorithm (COA), swarm bipolar algorithm (SBA), apiary organizational-based optimization algorithm (AOOA), and swarm space hopping algorithm (SSHA). The obtained results show that, BOA-SC is better than POA, NGO, COA, SBA, AOOA, and SSHA, in 5, 6, 8, 13, 18, 22, and 23 functions. In the second evaluation, the proposed algorithm compared with four BOA variants algorithms s-shaped binary butterfly optimization algorithm (S-bBOA), dynamic butterfly optimization algorithm (DBOA), chaotic butterfly optimization algorithm (CBOA), and optimization and extension of binary butterfly optimization approaches (OEbBOA) which are employed for feature selections methods. The results of BOA-SC are funnier than S-bBOA, DBOA, CBOA, and OEbBOA in three distinct datasets (Sonar, Waveform, and Spect) by archiving a high classification accuracy 97%, 86%, and 87% as a feature selection algorithm for the classification task. [ABSTRACT FROM AUTHOR]

Different biological data may be used to identify people in this investigation. The system uses complex multimodal fusion, feature extraction, classification, template matching, adjustable thresholding, and more. A trustworthy multimodal feature vector (B) is created using the Multimodal Fusion Algorithm from voice, face, and fingerprint data. The key objectives are weighing, normalizing, and extracting characteristics. Complex feature extraction algorithms improve this vector and ensure its accuracy and reliability. Hamming distance is utilized in template matching for accuracy. Support vector machines to ensure classification accuracy. The adaptive threshold technique adjusts option limits based on the biology score mean and standard deviation when external conditions change. A thorough look at the research shows how algorithms operate together and how vital each aspect is for locating criminals. Change the multimodal fusion weights for optimum results. Thorough research using tables and photographs revealed that the fingerprint approach is optimal. Fast, simple, and precise technologies may enable new unlawful recognition tools. The adaptive thresholding algorithm's multiple adaptation steps allow the system to adjust to diverse study circumstances. The Multimodal Biometric Identification System is a cutting-edge leader in its area and provides a trustworthy, practical, and customizable research choice. This novel strategy is at the forefront of criminal recognition technology and has been supported by ablation research. It affects reliability, accuracy, and adaptability. [ABSTRACT FROM AUTHOR]

Email types sorting is one of the most important tasks in current information systems with the purpose to improve the security of messages, allowing for their sorting into different types. This paper aims at studying the Convolution Neural Network and Long Short-Term Memory (CNN-LSTM), Convolution Neural Network and Gated Recurrent Unit (CNN-GRU) and Long Short-Term Memory (LSTM) deep learning models for the classification of emails into categories such as "Normal", "Fraudulent", "Harassment" and "Suspicious". The architecture of each model is discussed and the results of the models' performance by testing on labelled emails are presented. Evaluation outcomes show substantial gains in precision and throughput to conventional approaches hence inferring to the efficiency of these proposed models for automated email filtration and content evaluation. Last but not the least, the performance of the classification algorithms is evaluated with the help of parameters like Accuracy, precision, recall and F1-Score. From the experiment, the models found out that CNN-LSTM, together with the Term Frequency and Inverse Document Frequency (TF-IDF) feature extraction yielded the highest accuracy. The accuracy, precision, recall and f1-score values are 99. 348%, 99. 5%, 99. 3%, and 99. 2%, respectively. [ABSTRACT FROM AUTHOR]