Classification Techniques Research Articles

Proactive deadlock prevention based on traffic classification sub-graphs for triplet-based NoC TriBA-cNoC

Network topology and routing algorithms stand as pivotal decision points that profoundly impact the performance of Network-on-Chip (NoC) systems. As core counts rise, so does the inherent competition for shared resources, spotlighting the critical need for meticulously designed routing algorithms that circumvent deadlocks to ensure optimal network efficiency. This research capitalizes on the Triplet-Base Architecture (TriBA) and its Distributed Minimal Routing Algorithm (DM4T) to overcome the limitations of previous approaches. While DM4T exhibits performance advantages over previous routing algorithms, its deterministic nature and potential for circular dependencies during routing can lead to deadlocks and congestion. Therefore, this work addresses these vulnerabilities while leveraging the performance benefits of TriBA and DM4T. This work introduces a novel approach that merges a proactive deadlock prevention mechanism with Intermediate Adjacent Shortest Path Routing (IASPR). This combination guarantees both deadlock-free and livelock-free routing, ensuring reliable communication within the network. The key to this integration lies in a flow model-based data transfer categorization technique. This technique prevents the formation of circular dependencies. Additionally, it reduces redundant distance calculations during the routing process. By addressing these challenges, the proposed approach achieves improvements in both routing latency and throughput. To rigorously assess the performance of TriBA network topologies under varying configurations, extensive simulations were undertaken. The investigation encompassed both TriBA networks comprising 9 nodes and those with 27 nodes, employing DM4T, IASPR routing algorithms, and the proactive deadlock prevention method. The gem5 simulator, operating under the Garnet 3.0 network model using a standalone protocol for synthetic traffic patterns, was utilized for simulations at high injection rates, spanning diverse synthetic traffic patterns and PARSEC benchmark suite applications. Simulations rigorously quantified the effectiveness of the proposed approach, revealing reductions in average latency 40.17% and 34.05% compared to the lookup table and DM4T, respectively. Additionally, there were notable increases in average throughput of 7.48% and 5.66%.

Relasi Penafsiran Amin Al-Khuli tentang Puasa dalam Al-Quran dengan Kondisi Sosiologi dan Psikologi

The literary exegesis of the Qur'an developed by Amin Al-Khuli offers a new approach to interpreting the sacred text as a response to critiques of classical exegesis and modern cultural and intellectual developments. This study aims to explore the evolution of Qur'anic interpretation in the contemporary era, with a focus on Amin Al-Khuli’s literary exegesis of the verses on fasting and its relevance to the sociological and psychological conditions of Muslims who fast. The research uses a qualitative approach with library research methods, where data is gathered through literature review and analyzed using techniques of inventory, classification, and interpretation. The findings reveal that Al-Khuli’s literary exegesis provides a deeper and more contextual understanding of the fasting verses, linking the aspects of language and Qur'anic meaning with contemporary social and psychological conditions. The conclusion of this research is that the literary exegesis approach not only enriches the understanding of the text but also offers practical solutions to the challenges faced by Muslims in fasting practices. Recommendations for future research include conducting field studies to collect more empirical data and exploring the integration of literary exegesis with other interpretive approaches for broader application in Muslims' daily lives.

Class imbalanced data handling with cyberattack classification using Hybrid Salp Swarm Algorithm with deep learning approach

Cyberattack classification involves applying deep learning (DL) and machine learning (ML) models to categorize digital threats based on their features and behaviors. These models examine system logs, network traffic, or other associated data patterns to discriminate between standard activities and malicious actions. Efficient cyberattack classification is vital for on-time threat detection and response, permitting cybersecurity specialists to categorize and reduce potential risks to a system. Handling class-imbalanced data in cyberattack classification using DL is critical for achieving exact and robust models. In cybersecurity databases, instances of normal behavior frequently significantly outnumber instances of cyberattacks, foremost due to biased methods that may complete poorly on minority classes. To address this issue approaches such as oversampling the lesser class, undersampling the popular class, or using more advanced systems can be used. These plans defend that the DL technique is more complex when determining cyberattacks, so it increases complete performance and adapts the effect of the imbalance class on the classification results. This study presents a novel Hybrid Salp Swarm Algorithm with a DL Approach for Cyberattack Classification (HSSADL-CAC) technique. The HSSADL-CAC method intends to resolve class imbalance data handling with an optimum DL model for the recognition of cyberattacks. At first, the HSSADL-CAC method experiences data normalization as a pre-processing stage. The HSSADL-CAC technique uses the ADASYN approach to handle class imbalance problems. In addition, the HSSADL-CAC technique applies an HSSA-based feature selection approach. The HSSADL-CAC technique detects cyberattacks using a deep extreme learning machine (DELM) model. Finally, the hyperparameter tuning of the ELM model takes place by utilizing the beluga whale optimization (BWO) model. The performance analysis of the HSSADL-CAC technique employs a benchmark database. The comprehensive comparison research indicates the superior performance of the HSSADL-CAC technique in the cyberattack detection procedure.

Enhancing source code classification effectiveness via prompt learning incorporating knowledge features

Researchers have investigated the potential of leveraging pre-trained language models, such as CodeBERT, to enhance source code-related tasks. Previous methodologies have relied on CodeBERT’s ‘[CLS]’ token as the embedding representation of input sequences for task performance, necessitating additional neural network layers to enhance feature representation, which in turn increases computational expenses. These approaches have also failed to fully leverage the comprehensive knowledge inherent within the source code and its associated text, potentially limiting classification efficacy. We propose CodeClassPrompt, a text classification technique that harnesses prompt learning to extract rich knowledge associated with input sequences from pre-trained models, thereby eliminating the need for additional layers and lowering computational costs. By applying an attention mechanism, we synthesize multi-layered knowledge into task-specific features, enhancing classification accuracy. Our comprehensive experimentation across four distinct source code-related tasks reveals that CodeClassPrompt achieves competitive performance while significantly reducing computational overhead.

Migrating birds optimization-based feature selection for text classification

Text classification tasks, particularly those involving a large number of features, pose significant challenges in effective feature selection. This research introduces a novel methodology, MBO-NB, which integrates Migrating Birds Optimization (MBO) approach with naïve Bayes as an internal classifier to address these challenges. The motivation behind this study stems from the recognized limitations of existing techniques in efficiently handling extensive feature sets. Traditional approaches often fail to adequately streamline the feature selection process, resulting in suboptimal classification accuracy and increased computational overhead. In response to this need, our primary objective is to propose a scalable and effective solution that enhances both computational efficiency and classification accuracy in text classification systems. To achieve this objective, we preprocess raw data using the Information Gain algorithm, strategically reducing the feature count from an average of 62,221 to 2,089. Through extensive experiments, we demonstrate the superior effectiveness of MBO-NB in feature reduction compared to other existing techniques, resulting in significantly improved classification accuracy. Furthermore, the successful integration of naïve Bayes within MBO offers a comprehensive and well-rounded solution to the feature selection problem. In individual comparisons with Particle Swarm Optimization (PSO), MBO-NB consistently outperforms by an average of 6.9% across four setups. This research provides valuable insights into enhancing feature selection methods, thereby contributing to the advancement of text classification techniques. By offering a scalable and effective solution, MBO-NB addresses the pressing need for improved feature selection methods in text classification, thereby facilitating the development of more robust and efficient classification systems.

System development for enhancing social media advertisement engagement through XLNet-based personality classification

This research focuses on addressing the challenge of implementing personalized advertisements in the retail industry, where existing methods often face complexities that hinder their swift and large-scale adoption. The primary objective of this study was to develop a scalable and efficient social media advertisement personalization system by employing advanced personality classification techniques. The system utilizes the myPersonality dataset, grounded in the Big 5 OCEAN traits theory, to accurately classify user personalities. By integrating the XLNet model, optimized for personality classification, the system achieves a classification accuracy of 97.47 %, with precision, recall, and F1-Score values of 0.95, 0.94, and 0.94, respectively. The findings demonstrate that personalized advertisements, driven by accurately classified personality traits, significantly enhance user interaction rates, showing a 24 % improvement over generalized advertisements. This improvement in engagement suggests that the system can effectively personalize advertisements to resonate more deeply with users, fostering stronger connections between users and the advertised content. The proposed system's high accuracy and improved interaction rates make it a valuable addition to current marketing strategies, enhancing both engagement and conversion rates. This innovative approach has the potential to transform personalized advertising, making it more effective and widely adoptable within the marketing sector

Seagrass classification using unsupervised curriculum learning (UCL)

Seagrass ecosystems are pivotal in marine environments, serving as crucial habitats for diverse marine species and contributing significantly to carbon sequestration. Accurate classification of seagrass species from underwater images is imperative for monitoring and preserving these ecosystems. This paper introduces Unsupervised Curriculum Learning (UCL) to seagrass classification using the DeepSeagrass dataset. UCL progressively learns from simpler to more complex examples, enhancing the model's ability to discern seagrass features in a curriculum-driven manner. Experiments employing state-of-the-art deep learning architectures, convolutional neural networks (CNNs), show that UCL achieved overall 90.12 % precision and 89 % recall, which significantly improves classification accuracy and robustness, outperforming some traditional supervised learning approaches like SimCLR, and unsupervised approaches like Zero-shot CLIP. The methodology of UCL involves four main steps: high-dimensional feature extraction, pseudo-label generation through clustering, reliable sample selection, and fine-tuning the model. The iterative UCL framework refines CNN's learning of underwater images, demonstrating superior accuracy, generalization, and adaptability to unseen seagrass and background samples of undersea images. The findings presented in this paper contribute to the advancement of seagrass classification techniques, providing valuable insights into the conservation and management of marine ecosystems. The code and dataset are made publicly available and can be assessed here: https://github.com/nabid69/Unsupervised-Curriculum-Learning—UCL.

Applications of Machine Learning in Textile Industry

It is a question to all of us that what is machine learning and its importance for business sectors. Smart textile when using AI and computer vision try to act on knitting machines and to reduce production rate of defective types with real-time defects along with irregularities to identify with the objective to bring defective production at zero level. When ML and SC technique, artificial neural network, genetic algorithm, and fuzzy logic have preponderance over all others like Bibliographic analysis revealed a drastic rise in number of studies since 2015. It is proposed that demand forecasting is quite volatile and sensitive to several factors. Data mining studies with classification and clustering techniques, ML algorithms were implemented in textile industry. In a study it was found that machine productivity is having negative effect on quality which was confirmed by the management and confirmed a huge important in quality processing in industrial sectors. Demand forecasting by ML driven techniques has proved minimization of forecasting errors by 50%. ML is being used by Adidas to forecast demand in products since 2016. Ridge Regression estimated the correlation between predictor and observation variables.

Transit-oriented development (TOD) or development-oriented transit (DOT): how does it matter for sustainable urban growth?

ABSTRACT Urban growth patterns occur in various ways with tangible practical consequences. A key dilemma with regard to development and transportation revolves around whether the former follows the latter or vice versa. This study assessed this nexus with transit-oriented development (TOD) and development-oriented development (DOT) cases by using the land use/land cover (LU/LC) classification technique in a developing country example of a city in Turkey. Spatial data included high-resolution Sentinel 2A, Rapid Eye, and two SPOT satellite images and CORONA air photos acquired in 2021, 2013, 2000, 1989, and 1979, respectively, within three buffer zones (500, 750, and 1000 m) for both development forms. While the study acknowledges unorthodox TOD selection criteria and generalizability concerns, the results demonstrate that the TOD drastically reduced green areas (as opposed to ideal TOD requirements), while the DOT excessively grew commercial land uses and street networks. Both developments commonly impaired agricultural, vacant, and water land use. Such transformations direct local governments, transit agencies, and city planners to resolve ideal sustainable urban growth patterns by using remote sensing data and quantitative analysis.

Comparative analysis of features and classification techniques in breast cancer detection for Biglycan biomarker images.

Breast cancer (BC) is considered the world's most prevalent cancer. Early diagnosis of BC enables patients to receive better care and treatment, hence lowering patient mortality rates. Breast lesion identification and classification are challenging even for experienced radiologists due to the complexity of breast tissue and variations in lesion presentations. This work aims to investigate appropriate features and classification techniques for accurate breast cancer detection in 336 Biglycan biomarker images. The Biglycan biomarker images were retrieved from the Mendeley Data website (Repository name: Biglycan breast cancer dataset). Five features were extracted and compared based on shape characteristics (i.e., Harris Points and Minimum Eigenvalue (MinEigen) Points), frequency domain characteristics (i.e., The Two-dimensional Fourier Transform and the Wavelet Transform), and statistical characteristics (i.e., histogram). Six different commonly used classification algorithms were used; i.e., K-nearest neighbours (k-NN), Naïve Bayes (NB), Pseudo-Linear Discriminate Analysis (pl-DA), Support Vector Machine (SVM), Decision Tree (DT), and Random Forest (RF). The histogram of greyscale images showed the best performance for the k-NN (97.6%), SVM (95.8%), and RF (95.3%) classifiers. Additionally, among the five features, the greyscale histogram feature achieved the best accuracy in all classifiers with a maximum accuracy of 97.6%, while the wavelet feature provided a promising accuracy in most classifiers (up to 94.6%). Machine learning demonstrates high accuracy in estimating cancer and such technology can assist doctors in the analysis of routine medical images and biopsy samples to improve early diagnosis and risk stratification.

Identification and functional analysis of genes selected using different chemometric techniques on multiarray expression data of liver from high-fat diet treated mice

The prevalence of obesity is increasing, and this lifestyle disease is related to a high-fat diet, a surplus in caloric intake, and increased inflammation. This study aimed to use a publicly available dataset of microarray gene expression data from the liver of high-fat diet fed mice (GSE39549) to determine the functional importance of small subsets of the overall genes. The regulatory aspects of the chosen mice genes were extrapolated to human genes for the determination of potential diagnostic and therapeutic targets. The chemometric approaches of principal component analysis (PCA), random forest (RF), and genetic algorithm (GA) were used as data reduction techniques to select 50 genes from a total of 15,000 genes to differentiate liver samples from high-fat diet and normal diet-fed mice. A subset of 30 genes from each of the techniques were processed with classification techniques of k-nearest neighbor and support vector machines. The results showed that random forest was best at differentiating the samples and GA was the least accurate. The results of functional annotation and protein-protein interactions showed that genes selected by PCA and RF were more associated with obesity as they identified functions related to inflammatory processes, as well as lipid and cholesterol metabolic processes. The genes selected by GA identified processes related to cilium and cell projection. The proteins identified by RF, such as Msmo and Sqle, had roles in cholesterol metabolic and biosynthetic processes. The results showed that combining the genes selected by PCA and RF allowed a better understanding of the overall functional protein modules. The crosstalk genes such as Abcg5 as well as Abcg8 that relate cholesterol metabolic and biosynthetic process to glutathione metabolic process were identified. Various miRNAs-gene interactions are present in humans for most of the genes identified by PCA, RF, or GA. Some genes that showed fewer interactions with human miRNAs are CIDEA, PLIN4, and NME8. The results suggest the use of different chemometric analyses in combination with functional genomics to identify different sets of targets for diagnostic, therapeutic, and future research.

A New Metaheuristic Approach to Diagnosis of Parkinson’s Disease Through Audio Signals

Parkinson’s disease is accepted as one of the most important diseases in the world. Parkinson’s disease can be diagnosed in various conventional techniques. Recently, these techniques have been replaced by artificial intelligence systems. This study proposes a feature selection and classification technique for Parkinson’s disease based on speech signals using a meta-heuristic algorithm. The proposed method selects the features from the data set including speech signal data that most accurately represent the problem using the efficient search strategies of the immune plasma algorithm (IPA). The experimental results are promising compared to other competing methods for diagnosing Parkinson’s disease in the literature.

A Novel Approach for Detection, Segmentation, and Classification of Brain Tumors in MRI Images Using Neural Network and Special C Means Fuzzy Clustering Techniques

Brain tumors present a significant health challenge globally, necessitating advanced techniques for accurate detection, segmentation, and classification. This paper presents a comprehensive study focused on the development and evaluation of innovative methodologies for brain tumor analysis using medical imaging data. The primary objective of this research is to enhance the accuracy and efficiency of brain tumor detection, segmentation, and classification processes. To achieve this goal, a multi-step approach is proposed, integrating various computational techniques and machine learning algorithms. First, the study explores novel methods for preprocessing medical imaging data to enhance image quality and reduce noise artifacts. This preprocessing step plays a crucial role in improving the subsequent analysis stages' accuracy and reliability. Next, a robust tumor detection algorithm is developed, leveraging advanced image processing techniques and deep learning models. The proposed algorithm effectively identifies tumor regions within brain images with high accuracy and minimal false positives. Following tumor detection, a segmentation framework is introduced to precisely delineate tumor boundaries from surrounding healthy tissues. The segmentation algorithm combines traditional image processing methods with state-of-the-art deep learning architectures to achieve accurate and efficient tumor delineation. In this paper, we proposed three novel methods for automatic detection, classification, and segmentation of brain tumors. Comprehensive experiments are conducted on the BRATS dataset and show that the proposed model obtains competitive results. The parameters under study are Accuracy rate, Specificity, and Sensitivity. First approach focuses on classifying cancerous and non-cancerous brain tumors in MRI scans. The system first reads the images and employs a novel fusion method to combine information from various modalities (Flair, T1, T1C, T2) for a more comprehensive picture. This enhances the accuracy of tumor characterization. Following this fusion, the images undergo preprocessing, feature extraction, and classification. The preprocessing stage involves grayscale conversion, binarization, wavelet analysis, and region-of-interest (ROI) calculation. Finally, a robust Neural Network (NN) classification method effectively differentiates cancerous and non-cancerous brain tissue. Performance is evaluated by metrics like accuracy, sensitivity, and specificity. Compared to existing methods, this research system demonstrates superior results, achieving a classification accuracy of 96.61%, sensitivity of 96.66%, and specificity of 96.55%. The Second approach combines Backpropagation Neural Networks (BPNN) with Spatial Fuzzy C-Means (SFCM) clustering for brain tumor analysis. The BPNN classifies brains into normal, benign (non-cancerous abnormality), or malignant (cancerous) categories. SFCM helps pinpoint the exact tumor location and size within the MRI scan through segmentation. A dual-tree complex wavelet transform is employed to extract image features efficiently. This method achieved exceptional results: 99.77% classification accuracy, 99.87% sensitivity (correctly identifying tumors), and 98.69% specificity (correctly identifying healthy tissue). Additionally, the over 90% precision demonstrates the effectiveness of this technique in feature extraction and classification. The experiments demonstrate that BPNN-SFCM successfully segment, and extracts brain tumors from MRI scans. The overall accuracy of BPNN-SFCM is better as compared to other proposed methods and existing methods.

Radar Powered Perception: Multiple Approaches for Vehicle Classification

The proliferation of Advanced Driver-Assistance Systems (ADAS) and autonomous vehicles demands robust and reliable techniques for vehicle detection and classification. This paper investigates the potential of automotive radar systems for this critical task, focusing on classifying vehicles into three distinct categories (bus/truck, car/sedan/SUV, bike/bicycle). We explore multiple approaches that leverages rich set of features extracted from radar measurements. These features include target size, point density, and other signature characteristics derived from the radar return signal. By analyzing these features, we explain multiple approach aims to achieve vehicle classification. The effectiveness of the proposed method is evaluated using real-world data, with a particular focus on demonstrating its suitability for real-time applications in ADAS. The paper delves into the critical trade-offs that exist between classification accuracy, computational complexity, and inherent limitations of automotive radar sensors. We discuss the impact of sparse data, such as low point cloud data on classification accuracy. Additionally, we discuss on potential mitigation strategies to ensure reliability in challenging scenarios. Finally, the paper concludes by outlining potential future directions for enhancing vehicle classification using automotive radar systems. These advancements will pave the way for the development of more robust and reliable ADAS and autonomous vehicles, ultimately contributing to improved safety on our roads.

NAKNN: An Efficient Classification of Indonesian News Texts with Nazief-Adriani and KNN

Internet usage in Indonesia has seen a significant increase, reaching 215.63 million users in 2022-2023, or 78.19% of the population. With the ease of internet access, digital news portals like Narasi TV have become a primary source of information for many people. However, the large number of news articles makes manual categorizing challenging. This study aims to classify Indonesian-language news documents from Narasi TV using the Nazief-Adriani algorithm for stemming and the K-Nearest Neighbor (KNN) method for classification. The text mining process begins with preprocessing, which includes case folding, tokenizing, stop-word filtering, and stemming. Using a dataset of 500 news documents, the study demonstrated that with a 90:10 data split, the average accuracy reached 93%, with the highest value being 100%. For the 80:20 data split, the average accuracy was 89%, with the highest value being 93%, and for a 70:30 data split, the average accuracy was 87%, with the highest value being 89%. In conclusion, the combination of the Nazief-Adriani algorithm and the KNN method with optimal k selection and random states obtained high accuracy, obtaining an average accuracy of 93%) in classifying Indonesian-language news documents. These results demonstrate the significant potential of text mining and classification techniques to manage digital news.

Time series land use/land cover mapping and change detection to support policies on sustainable environmental and economic management

The upper Tekeze River Basin is facing challenges of widespread deforestation and natural vegetation cover degradation that could exacerbate the water scarcity, food insecurity and extreme poverty in the region. Using remote sensing and GIS, this study quantified the land use land cover change trend in the last three decades and analyzed the current land use / cover statues in the basin. A hybrid classification technique is applied to obtain better classification accuracy. Moreover, for automated cloud and cloud shadow detection the newly developed Mountainous Fmask is used. Using post classification change detection technique, seven major land use/cover classes were identified. These classes remained the dominant classes during the study period, showing marked changes in the area coverage within them. Based on the error matrix statistical indices, the classification accuracies of each class are found to be strong. The overall accuracy and the kappa coefficient for the 2021 map are 91% and 89%, respectively. The techniques used have contributed to improving the accuracy of the classification process and helped the classified images to practically match the ground truths. The analysis revealed settlement expansion by 570.31% in parallel with the expansion of farmland by 52.32% during the period 1991–2021. In contrast, the forestland decreased significantly, by 75.55%. The environmental degradation and unplanned use of land resources could have contributed to why the upper Tekeze basin is experiencing worsening poverty, water scarcity and food insecurity. Thus, land use/cover time series modeling is essential for various purposes, including land use planning and, managing natural resources. In this regards this study provides basic information for implementing sustainable environmental conservation strategies in the area. Furthermore, the applied methodologies may have practical applications in other similar areas.

Explainable AI (XAI) Techniques for Convolutional Neural Network-Based Classification of Drilled Holes in Melamine Faced Chipboard

The furniture manufacturing sector faces significant challenges in machining composite materials, where quality issues such as delamination can lead to substandard products. This study aims to improve the classification of drilled holes in melamine-faced chipboard using Explainable AI (XAI) techniques to better understand and interpret Convolutional Neural Network (CNN) models’ decisions. We evaluated three CNN architectures (VGG16, VGG19, and ResNet101) pretrained on the ImageNet dataset and fine-tuned on our dataset of drilled holes. The data consisted of 8526 images, divided into three categories (Green, Yellow, Red) based on the drill’s condition. We used 5-fold cross-validation for model evaluation and applied LIME and Grad-CAM as XAI techniques to interpret the model decisions. The VGG19 model achieved the highest accuracy of 67.03% and the lowest critical error rate among the evaluated models. LIME and Grad-CAM provided complementary insights into the decision-making process of the model, emphasizing the significance of certain features and regions in the images that influenced the classifications. The integration of XAI techniques with CNN models significantly enhances the interpretability and reliability of automated systems for tool condition monitoring in the wood industry. The VGG19 model, combined with LIME and Grad-CAM, offers a robust solution for classifying drilled holes, ensuring better quality control in manufacturing processes.

Unbiased Morphometric Assessment of Red Blood Cell Storage Lesion in the Presence of Shear-Induced Stomatocytes

Introduction: Red blood cells (RBCs) undergo progressive biochemical and morphological changes during storage, collectively called storage lesion. The quality of red cell concentrates (RCCs) is typically assessed by quantifying hemolysis. An assessment of morphological changes, associated with low quality RBCs, could give an additional indication of the safety and efficacy of the concentrates. The current standard for determining morphological changes is a manual, laborious, and subjectively biased microscopic process that limits the number of cells that can be examined. When using alternative methods like flow cells, flow and shear-induced morphologies affecting especially stomatocyte morphologies must be taken into account. We already established an automated flow morphometric RBC analysis system as an alternative to manual microscopic evaluation. The goal of the present work is to obtain a robust, automated, morphology-related signal (lesion index) quantifying RBC storage lesion in a laminar flow channel under conditions similar to stasis that is not affected by shear-induced reversible morphology changes. Methods: We use a convolutional neural network (CNN) for high throughput classification of RBCs. We analyzed the morphological changes of 5 RCCs over a period of 12 weeks and classified RBC morphologies, including such that are degradation-induced and reversible. We introduce a lesion index to denote the percentage of irreversible spherical morphologies, known to reduce the post-transfusion survival of erythrocytes. We further addressed shear-induced stomatocyte morphologies in laminar flow and whether these affect CNN-based RBC classification. Results: Our flow morphometry system achieves a high-resolution classification comprising nine morphological classes with an excellent overall accuracy of 92% and F1 scores between 84% and 97%. We generate strong evidence that the morphological lesion index can predict the hemolysis level in RCCs during storage. The power of this new classification technique allowed it, for the first time, to detect and measure the lateral concentration gradient of stomatocytes in a conventional flow chamber. Importantly, we show that reversible shear rate-induced morphologies, typical for microfluidic systems, bear no influence on the lesion index. Conclusion: Flow morphometry combined with evaluation by a CNN allows to reliably assess RBC storage lesion and thus concentrate quality. Additionally, this method reduces the need for complex laboratory procedures.

Exploring Explainable AI Techniques for Text Classification in Healthcare: A Scoping Review.

Text classification plays an essential role in the medical domain by organizing and categorizing vast amounts of textual data through machine learning (ML) and deep learning (DL). The adoption of Artificial Intelligence (AI) technologies in healthcare has raised concerns about the interpretability of AI models, often perceived as "black boxes." Explainable AI (XAI) techniques aim to mitigate this issue by elucidating AI model decision-making process. In this paper, we present a scoping review exploring the application of different XAI techniques in medical text classification, identifying two main types: model-specific and model-agnostic methods. Despite some positive feedback from developers, formal evaluations with medical end users of these techniques remain limited. The review highlights the necessity for further research in XAI to enhance trust and transparency in AI-driven decision-making processes in healthcare.

Cardiovascular disease detection from cardiac arrhythmia ECG signals using artificial intelligence models with hyperparameters tuning methodologies

Cardiovascular disease (CVD) is connected with irregular cardiac electrical activity, which can be seen in ECG alterations. Due to its convenience and non-invasive aspect, the ECG is routinely exploited to identify different arrhythmias and automatic ECG recognition is needed immediately. In this paper, enhancement for the detection of CVDs such as Ventricular Tachycardia (VT), Premature Ventricular Contraction (PVC) and ST Change (ST) arrhythmia using different dimensionality reduction techniques and multiple classifiers are presented. Three-dimensionality reduction methods, such as Local Linear Embedding (LLE), Diffusion Maps (DM), and Laplacian Eigen (LE), are employed. The dimensionally reduced ECG samples are further feature selected with Cuckoo Search (CS) and Harmonic Search Optimization (HSO) algorithms. A publicly available MIT-BIH (Physionet) - VT database, PVC database, ST Change database and NSR database were used in this work. The cardiac vascular disturbances are classified by using seven classifiers such as Gaussian Mixture Model (GMM), Expectation Maximization (EM), Non-linear Regression (NLR), Logistic Regression (LR), Bayesian Linear Discriminant Analysis (BDLC), Detrended Fluctuation Analysis (Detrended FA), and Firefly. For different classes, the average overall accuracy of the classification techniques is 55.65 % when without CS and HSO feature selection, 64.36 % when CS feature selection is used, and 75.39 % when HSO feature selection is used. Also, to improve the performance of classifiers, the hyperparameters of four classifiers (GMM, EM, BDLC and Firefly) are tuned with the Adam and Grid Search Optimization (GSO) approaches. The average accuracy of classification for the CS feature-based classifiers that used GSO and Adam hyperparameter tuning was 79.92 % and 85.78 %, respectively. The average accuracy of classification for the HSO feature-based classifiers that used GSO and Adam hyperparameter tuning was 86.87 % and 93.77 %, respectively. The performance of the classifier is analyzed based on the accuracy parameter for both with and without feature selection methods and with hyperparameter tuning techniques. In the case of ST vs. NSR, a higher accuracy of 98.92 % is achieved for the LLE dimensionality reduction with HSO feature selection for the GMM classifier with Adam's hyperparameter tuning approach. The GMM classifier with the Adam hyperparameter tuning approach with 98.92 % accuracy in detecting ST vs. NSR cardiac disease is outperforming all other classifiers and methodologies.