IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-19 DOI: 10.1016/j.array.2025.100585

Kexin Zhang, Lingwei Kong, Kun Zhang, Chenxi Huang, Yaqin Kuang, Xiaoming Huang, Rongqiang Feng

With the widespread integration of distributed energy resources (DERs), regional power systems are increasingly reliant on coordinated control among transmission, distribution, and microgrid layers. Traditional control strategies face significant challenges in managing the heterogeneity, volatility, and dynamic load responses inherent in such systems. To address these issues, this paper proposes a privacy-aware and ethically aligned deep reinforcement learning (DRL) framework for optimizing multi-agent coordinated control across hierarchical grid components. The proposed approach constructs a multi-layered state–action space encompassing transmission, distribution, and microgrid subsystems. A multi-agent DRL mechanism is integrated to achieve multi-objective optimization, including load shedding mitigation, overload prevention, reverse power flow control, and voltage stability. Importantly, our design incorporates privacy-preserving training protocols and explainable decision-making modules to ensure transparency, accountability, and secure deployment in critical infrastructure settings. Extensive simulations across diverse operational scenarios demonstrate the strategy’s superior responsiveness, improved renewable energy utilization, and robustness under uncertain conditions. The framework not only enhances operational efficiency but also aligns with emerging global demands for secure, transparent, and ethically governed AI deployment in smart grids. This study provides a novel pathway toward intelligent, privacy-preserving, and responsible regional grid control.

随着分布式能源的广泛集成，区域电力系统越来越依赖于输电、配电和微网层之间的协调控制。传统的控制策略在管理此类系统固有的异质性、波动性和动态负载响应方面面临重大挑战。为了解决这些问题，本文提出了一个隐私感知和道德对齐的深度强化学习（DRL）框架，用于优化跨分层网格组件的多智能体协调控制。该方法构建了一个包含输电、配电和微电网子系统的多层状态-行为空间。集成多智能体DRL机制，实现多目标优化，包括减载缓解、过载预防、反向潮流控制和电压稳定。重要的是，我们的设计结合了隐私保护培训协议和可解释的决策模块，以确保在关键基础设施设置中的透明度、问责制和安全部署。在不同操作场景中进行的大量模拟表明，该策略具有卓越的响应能力，提高了可再生能源的利用率，并且在不确定条件下具有鲁棒性。该框架不仅提高了运营效率，而且符合全球对智能电网中安全、透明和道德治理的人工智能部署的新需求。该研究为实现智能、隐私保护和负责任的区域网格控制提供了一条新途径。

{"title":"Privacy-aware multi-agent deep reinforcement learning for ethical coordinated control in smart regional grids","authors":"Kexin Zhang, Lingwei Kong, Kun Zhang, Chenxi Huang, Yaqin Kuang, Xiaoming Huang, Rongqiang Feng","doi":"10.1016/j.array.2025.100585","DOIUrl":"10.1016/j.array.2025.100585","url":null,"abstract":"<div><div>With the widespread integration of distributed energy resources (DERs), regional power systems are increasingly reliant on coordinated control among transmission, distribution, and microgrid layers. Traditional control strategies face significant challenges in managing the heterogeneity, volatility, and dynamic load responses inherent in such systems. To address these issues, this paper proposes a privacy-aware and ethically aligned deep reinforcement learning (DRL) framework for optimizing multi-agent coordinated control across hierarchical grid components. The proposed approach constructs a multi-layered state–action space encompassing transmission, distribution, and microgrid subsystems. A multi-agent DRL mechanism is integrated to achieve multi-objective optimization, including load shedding mitigation, overload prevention, reverse power flow control, and voltage stability. Importantly, our design incorporates privacy-preserving training protocols and explainable decision-making modules to ensure transparency, accountability, and secure deployment in critical infrastructure settings. Extensive simulations across diverse operational scenarios demonstrate the strategy’s superior responsiveness, improved renewable energy utilization, and robustness under uncertain conditions. The framework not only enhances operational efficiency but also aligns with emerging global demands for secure, transparent, and ethically governed AI deployment in smart grids. This study provides a novel pathway toward intelligent, privacy-preserving, and responsible regional grid control.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100585"},"PeriodicalIF":4.5,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145735331","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Photonic crystal fiber-based terahertz biosensor: Identifying Staphylococcus aureus bacteria and skin health implications 基于光子晶体光纤的太赫兹生物传感器：识别金黄色葡萄球菌细菌和皮肤健康的意义

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-17 DOI: 10.1016/j.array.2025.100580

A.H.M. Iftekharul Ferdous, MdChomon Islam, Zannatun Naim Pavel, T.H.M. Sumon Rashid, Md Safiul Islam, Rosni Sayed, Md Asaduzzaman Shobug, Rokieya akter

The widespread Gram-positive bacterium Staphylococcus aureus, often known as S. aureus, is present on human skin and in nasal passages. Although it can occasionally be harmless, it is a dangerous pathogen that can cause skin infections and systemic illnesses. One of the main concerns is its increasing resistance to traditional treatments. To accurately identify S. aureus, we describe a highly sensitive photonic crystal fiber (PCF) biosensor. THz PCF basically the foundation of the sensor, which offers a dependable and selective platform for bacterial identification. One of the main causes of wounds, skin infections, and other serious health problems is S. aureus. The suggested PCF efficiently confines and guides terahertz radiation by utilizing air gaps and high index contrast. Its highest relative sensitivity, when operating at 1.6 THz, is 97.21 % for S. aureus (RI = 1.416) and 94.74 % for basal normal (RI = 1.36). The sensor has a high effective material loss of 0.008678 cm⁻¹ and 0.005573 cm⁻¹ regarding equivalent RI values and low confinement loss of 6.513 × 10⁻⁹ dB/m and 2.4276 × 10⁻⁷ dB/m. Sensor exhibits very low CL as well as negligible effective material loss, guaranteeing dependability with efficiency. Timely treatment and improved results with less harsh medications are made possible by early diagnosis of infections, including skin cancer. It streamlines illness monitoring and enhances patient care. The sensor is appropriate for portable diagnostic equipment because of its small size, excellent accuracy, and versatility. In field and hospital settings, its portability makes it perfect for quick on-site detection. This creative design encourages global healthcare awareness and advances treatment approaches.

广泛存在的革兰氏阳性细菌金黄色葡萄球菌，通常被称为金黄色葡萄球菌，存在于人体皮肤和鼻道中。虽然它有时是无害的，但它是一种危险的病原体，可导致皮肤感染和全身疾病。主要的担忧之一是它对传统治疗方法的抵抗力越来越强。为了准确地识别金黄色葡萄球菌，我们设计了一种高灵敏度的光子晶体光纤（PCF）生物传感器。太赫兹PCF是传感器的基础，为细菌鉴定提供了可靠的、选择性的平台。造成伤口、皮肤感染和其他严重健康问题的主要原因之一是金黄色葡萄球菌。所建议的PCF通过利用气隙和高折射率对比度有效地限制和引导太赫兹辐射。当工作在1.6太赫兹时，其最高相对灵敏度对金黄色葡萄球菌（RI = 1.416）为97.21%，对基础正常菌（RI = 1.36）为94.74%。该传感器的有效材料损耗为0.008678 cm−1和0.005573 cm−1，等效RI值和低约束损耗分别为6.513 × 10−9 dB/m和2.4276 × 10−7 dB/m。传感器具有非常低的CL和可忽略不计的有效材料损耗，保证了可靠性和效率。通过早期诊断感染（包括皮肤癌），可以及时治疗并使用较轻的药物改善结果。它简化了疾病监测并加强了患者护理。该传感器适用于便携式诊断设备，因为它体积小，精度好，多功能性强。在现场和医院设置，它的便携性使其完美的快速现场检测。这种创造性的设计鼓励全球医疗保健意识和先进的治疗方法。

{"title":"Photonic crystal fiber-based terahertz biosensor: Identifying Staphylococcus aureus bacteria and skin health implications","authors":"A.H.M. Iftekharul Ferdous, MdChomon Islam, Zannatun Naim Pavel, T.H.M. Sumon Rashid, Md Safiul Islam, Rosni Sayed, Md Asaduzzaman Shobug, Rokieya akter","doi":"10.1016/j.array.2025.100580","DOIUrl":"10.1016/j.array.2025.100580","url":null,"abstract":"<div><div>The widespread Gram-positive bacterium <em>Staphylococcus aureus</em>, often known as <em>S. aureus</em>, is present on human skin and in nasal passages. Although it can occasionally be harmless, it is a dangerous pathogen that can cause skin infections and systemic illnesses. One of the main concerns is its increasing resistance to traditional treatments. To accurately identify <em>S. aureus</em>, we describe a highly sensitive photonic crystal fiber (PCF) biosensor. THz PCF basically the foundation of the sensor, which offers a dependable and selective platform for bacterial identification. One of the main causes of wounds, skin infections, and other serious health problems is <em>S. aureus</em>. The suggested PCF efficiently confines and guides terahertz radiation by utilizing air gaps and high index contrast. Its highest relative sensitivity, when operating at 1.6 THz, is 97.21 % for <em>S. aureus</em> (RI = 1.416) and 94.74 % for basal normal (RI = 1.36). The sensor has a high effective material loss of 0.008678 cm<sup>−1</sup> and 0.005573 cm<sup>−1</sup> regarding equivalent RI values and low confinement loss of 6.513 × 10<sup>−9</sup> dB/m and 2.4276 × 10<sup>−7</sup> dB/m. Sensor exhibits very low CL as well as negligible effective material loss, guaranteeing dependability with efficiency. Timely treatment and improved results with less harsh medications are made possible by early diagnosis of infections, including skin cancer. It streamlines illness monitoring and enhances patient care. The sensor is appropriate for portable diagnostic equipment because of its small size, excellent accuracy, and versatility. In field and hospital settings, its portability makes it perfect for quick on-site detection. This creative design encourages global healthcare awareness and advances treatment approaches.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100580"},"PeriodicalIF":4.5,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A systematic review of healthcare cyber–physical systems with associated innovative technologies for Alzheimer’s and Parkinson’s Diseases 医疗网络物理系统与相关的创新技术对阿尔茨海默病和帕金森病的系统回顾

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-17 DOI: 10.1016/j.array.2025.100575

Youness Amadiaz, Edgar Alfonso-Lizarazo, Ahmed Nait Sidi Moh

The emergence of cyber–physical systems (CPS) and new healthcare technologies provides innovative ways to diagnose and monitor diseases. This paper presents a systematic review of studies related to healthcare CPS for Alzheimer’s disease (AD) and Parkinson’s disease (PD), two of the most prevalent neurodegenerative diseases worldwide. The studies reviewed are classified into four main application domains: (i) assistance with diagnosis, (ii) symptom monitoring, (iii) prediction of treatment response, and (iv) support for therapy and rehabilitation. Methodological approaches are grouped into descriptive, predictive, and prescriptive analytics, evaluating their contributions to patient care and CPS applications. Most PD studies focus on diagnosis and symptom monitoring, using devices such as wearable sensors, EEG/EMG systems, and video tracking combined with machine learning, deep learning, and optimization algorithms. Most AD studies focus on treatment response and rehabilitation support, employing force or pressure sensors, optical motion trackers, and clinical decision support systems (CDSS), along with artificial intelligence techniques and digital twin models. However, healthcare management along the patient care pathway remains underexplored for both diseases. This review also identifies challenges for real-world use, such as integrating data from multiple sources, making pathway-level decisions, and creating CPS frameworks that are more personalized, explainable, and patient-centered.

网络物理系统（CPS）和新的医疗保健技术的出现为诊断和监测疾病提供了创新的方法。本文对全球最常见的两种神经退行性疾病阿尔茨海默病（AD）和帕金森病（PD）的医疗CPS相关研究进行了系统综述。回顾的研究分为四个主要应用领域：(i)协助诊断，（ii）症状监测，（iii）预测治疗反应，（iv）支持治疗和康复。方法学方法分为描述性、预测性和规范性分析，评估其对患者护理和CPS应用的贡献。大多数PD研究侧重于诊断和症状监测，使用可穿戴传感器，脑电图/肌电图系统等设备，以及结合机器学习，深度学习和优化算法的视频跟踪。大多数AD研究关注治疗反应和康复支持，采用力或压力传感器、光学运动跟踪器、临床决策支持系统（CDSS），以及人工智能技术和数字孪生模型。然而，沿着患者护理途径的医疗保健管理对这两种疾病仍未充分探索。本综述还指出了现实应用中的挑战，例如整合来自多个来源的数据，做出路径级决策，以及创建更加个性化、可解释和以患者为中心的CPS框架。

{"title":"A systematic review of healthcare cyber–physical systems with associated innovative technologies for Alzheimer’s and Parkinson’s Diseases","authors":"Youness Amadiaz, Edgar Alfonso-Lizarazo, Ahmed Nait Sidi Moh","doi":"10.1016/j.array.2025.100575","DOIUrl":"10.1016/j.array.2025.100575","url":null,"abstract":"<div><div>The emergence of cyber–physical systems (CPS) and new healthcare technologies provides innovative ways to diagnose and monitor diseases. This paper presents a systematic review of studies related to healthcare CPS for Alzheimer’s disease (AD) and Parkinson’s disease (PD), two of the most prevalent neurodegenerative diseases worldwide. The studies reviewed are classified into four main application domains: (i) assistance with diagnosis, (ii) symptom monitoring, (iii) prediction of treatment response, and (iv) support for therapy and rehabilitation. Methodological approaches are grouped into descriptive, predictive, and prescriptive analytics, evaluating their contributions to patient care and CPS applications. Most PD studies focus on diagnosis and symptom monitoring, using devices such as wearable sensors, EEG/EMG systems, and video tracking combined with machine learning, deep learning, and optimization algorithms. Most AD studies focus on treatment response and rehabilitation support, employing force or pressure sensors, optical motion trackers, and clinical decision support systems (CDSS), along with artificial intelligence techniques and digital twin models. However, healthcare management along the patient care pathway remains underexplored for both diseases. This review also identifies challenges for real-world use, such as integrating data from multiple sources, making pathway-level decisions, and creating CPS frameworks that are more personalized, explainable, and patient-centered.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100575"},"PeriodicalIF":4.5,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568320","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing active learning through latent space exploration: A k-nearest neighbors approach 通过潜在空间探索增强主动学习：k近邻方法

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-17 DOI: 10.1016/j.array.2025.100584

Sergio Flesca , Domenico Mandaglio , Francesco Scala

Supervised machine learning often requires a significant volume of labeled training data, incurring substantial costs for data annotation. In scenarios with limited labeling budgets, selecting the most informative instances for labeling by an annotation oracle becomes crucial. Active learning addresses this challenge by strategically choosing informative instances for labeling, thereby maximizing model performance with limited labeled data. Existing active learning methods, however, typically do not fully exploit abundant unlabeled data that can be used to extract meaningful features from raw data. While some methods integrate variational autoencoders (VAEs) into active learning, this work introduces a novel framework that does not use VAEs merely to assist in the selection of data for the oracle. Instead, our approach leverages the latent space learned by the VAE to heuristically annotate unlabeled data through a k-nearest neighbor classifier within this space. The proposed approach allows to enhance existing active learning methods without relying solely on an annotation oracle, thus reducing the overall annotation cost. Experiments on benchmark datasets show that our proposal can improve the performance of existing active learning methods by up to 33% in terms of classification accuracy and by up to 0.38 in terms of F1-score when the initial labeled data is extremely limited. We make source code and evaluation data available at https://github.com/Franco7Scala/Laken.

监督式机器学习通常需要大量标记的训练数据，这就需要大量的数据注释成本。在标注预算有限的场景中，选择最具信息量的实例进行标注是至关重要的。主动学习通过策略性地选择信息实例进行标记来解决这一挑战，从而在有限的标记数据下最大化模型性能。然而，现有的主动学习方法通常不能充分利用大量的未标记数据，这些数据可用于从原始数据中提取有意义的特征。虽然有些方法将变分自动编码器（VAEs）集成到主动学习中，但这项工作引入了一个新的框架，该框架不仅仅使用VAEs来帮助oracle选择数据。相反，我们的方法利用由VAE学习的潜在空间，通过该空间内的k近邻分类器启发式地注释未标记的数据。提出的方法可以增强现有的主动学习方法，而不依赖于注释oracle，从而降低了总体注释成本。在基准数据集上的实验表明，在初始标记数据非常有限的情况下，我们的提议可以将现有主动学习方法的分类精度提高33%，f1分数提高0.38。我们在https://github.com/Franco7Scala/Laken上提供源代码和评估数据。

{"title":"Enhancing active learning through latent space exploration: A k-nearest neighbors approach","authors":"Sergio Flesca , Domenico Mandaglio , Francesco Scala","doi":"10.1016/j.array.2025.100584","DOIUrl":"10.1016/j.array.2025.100584","url":null,"abstract":"<div><div>Supervised machine learning often requires a significant volume of labeled training data, incurring substantial costs for data annotation. In scenarios with limited labeling budgets, selecting the most informative instances for labeling by an annotation oracle becomes crucial. Active learning addresses this challenge by strategically choosing informative instances for labeling, thereby maximizing model performance with limited labeled data. Existing active learning methods, however, typically do not fully exploit abundant unlabeled data that can be used to extract meaningful features from raw data. While some methods integrate variational autoencoders (VAEs) into active learning, this work introduces a novel framework that does not use VAEs merely to assist in the selection of data for the oracle. Instead, our approach leverages the latent space learned by the VAE to heuristically annotate unlabeled data through a k-nearest neighbor classifier within this space. The proposed approach allows to enhance existing active learning methods without relying solely on an annotation oracle, thus reducing the overall annotation cost. Experiments on benchmark datasets show that our proposal can improve the performance of existing active learning methods by up to 33% in terms of classification accuracy and by up to 0.38 in terms of F1-score when the initial labeled data is extremely limited. We make source code and evaluation data available at <span><span>https://github.com/Franco7Scala/Laken</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100584"},"PeriodicalIF":4.5,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Rationalizing machine learning models for real-time health prognosis of lithium-ion batteries 优化用于锂离子电池实时健康预测的机器学习模型

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-15 DOI: 10.1016/j.array.2025.100577

Mario Mallea, Àngela Nebot, Francisco Mugica

Accurate, personalized, and real-time health forecasts for lithium-ion batteries are essential to ensure the safety and reliability of energy storage systems. Two critical aspects that must be jointly addressed are the State of Health (SoH) and the Remaining Useful Life (RUL). Current data-driven approaches often rely on deep neural network models, which are computationally demanding and depend on the interplay between SoH tracking and RUL prognosis.

To improve both efficiency and performance, this study introduces a conceptual framework grounded in four key concepts: adaptability, complexity, memory, and synergy. We analyze and refine a state-of-the-art deep transfer learning model, identifying which components are most significant concerning these concepts. Building on this, we propose several deep neural network variants and lightweight Echo State Networks with a novel reservoir structure.

Extensive experiments on three diverse battery datasets demonstrate that our models consistently outperform deep transfer learning approaches for both RUL and SoH. Notably, efficiency and performance gains are achieved through strategies that reinterpret the conventional synergy between degradation dynamics leveraged in deep transfer learning. On the largest dataset, our best models achieve reductions in root mean square error of 40% for RUL and 91% for SoH.

准确、个性化和实时的锂离子电池健康预测是确保储能系统安全性和可靠性的关键。必须共同处理的两个关键方面是健康状况（SoH）和剩余使用寿命（RUL）。目前的数据驱动方法通常依赖于深度神经网络模型，这些模型的计算要求很高，并且依赖于SoH跟踪和RUL预后之间的相互作用。为了提高效率和绩效，本研究引入了一个基于四个关键概念的概念框架：适应性、复杂性、记忆和协同。我们分析和完善了最先进的深度迁移学习模型，确定了哪些组件对这些概念最重要。在此基础上，我们提出了几种深度神经网络变体和具有新型储层结构的轻型回声状态网络。在三个不同的电池数据集上进行的大量实验表明，我们的模型在RUL和SoH方面始终优于深度迁移学习方法。值得注意的是，效率和性能的提高是通过重新解释深度迁移学习中利用的退化动态之间的传统协同作用的策略来实现的。在最大的数据集上，我们最好的模型实现了RUL的均方根误差降低40%，SoH的均方根误差降低91%。

{"title":"Rationalizing machine learning models for real-time health prognosis of lithium-ion batteries","authors":"Mario Mallea, Àngela Nebot, Francisco Mugica","doi":"10.1016/j.array.2025.100577","DOIUrl":"10.1016/j.array.2025.100577","url":null,"abstract":"<div><div>Accurate, personalized, and real-time health forecasts for lithium-ion batteries are essential to ensure the safety and reliability of energy storage systems. Two critical aspects that must be jointly addressed are the State of Health (SoH) and the Remaining Useful Life (RUL). Current data-driven approaches often rely on deep neural network models, which are computationally demanding and depend on the interplay between SoH tracking and RUL prognosis.</div><div>To improve both efficiency and performance, this study introduces a conceptual framework grounded in four key concepts: adaptability, complexity, memory, and synergy. We analyze and refine a state-of-the-art deep transfer learning model, identifying which components are most significant concerning these concepts. Building on this, we propose several deep neural network variants and lightweight Echo State Networks with a novel reservoir structure.</div><div>Extensive experiments on three diverse battery datasets demonstrate that our models consistently outperform deep transfer learning approaches for both RUL and SoH. Notably, efficiency and performance gains are achieved through strategies that reinterpret the conventional synergy between degradation dynamics leveraged in deep transfer learning. On the largest dataset, our best models achieve reductions in root mean square error of 40% for RUL and 91% for SoH.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100577"},"PeriodicalIF":4.5,"publicationDate":"2025-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568366","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Brain tumor detection, classification and segmentation by deep learning models from MRI images: Recent approaches, challenges and future directions 基于MRI图像的深度学习模型的脑肿瘤检测、分类和分割：最新方法、挑战和未来方向

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-14 DOI: 10.1016/j.array.2025.100571

Tahasin Ahmed Fahim , Fatema Binte Alam , Md Azad Hossain

The diagnosis of brain tumors presents a very important challenge in the neuro-oncology field because of the complexities, heterogeneity, and high mortality of the tumors. The latest trends in deep learning have revolutionized the research in the field of medical image analysis. Through these trends, automated and precise brain tumor-detection, classification, and segmentation have been accomplished. This paper provides a systematic review with the focus taxonomy of the development of brain tumor analysis models with the most advanced deep neural networks based on tasks, and methods. It also summarizes the information of commonly available public datasets, preprocessing methods for MRI images, performance evaluation metrics. It reviews in detail on deep learning models available for brain tumor detection,classification, and segmentation in terms of performance metrics and clinical relevances. For representing in organized way, all the works reviewed here are divided into several groups and compared on specific benchmarks. Moreover, it figures out current challenges regarding brain tumor diagnosis and the potential implications of future studies to increase clinical applicability and trustworthiness of AI-driven solutions. This review acts as an informational guide to any researchers and healthcare professionals. It describes recent emerging patterns, current issues, and opportunities of deep learning to transform the diagnosis and treatment of brain tumors.

由于脑肿瘤的复杂性、异质性和高死亡率，脑肿瘤的诊断是神经肿瘤学领域一个非常重要的挑战。深度学习的最新趋势使医学图像分析领域的研究发生了革命性的变化。通过这些趋势，自动化和精确的脑肿瘤检测、分类和分割已经完成。本文对基于任务和方法的最先进的深度神经网络发展的脑肿瘤分析模型的重点分类进行了系统综述。总结了常用的公共数据集信息、MRI图像预处理方法、性能评价指标。它从性能指标和临床相关性方面详细回顾了用于脑肿瘤检测、分类和分割的深度学习模型。为了有组织地展示，这里所有的作品都被分成几个组，并在特定的基准上进行比较。此外，它还指出了当前脑肿瘤诊断方面的挑战以及未来研究的潜在意义，以提高人工智能驱动解决方案的临床适用性和可信度。本综述可作为任何研究人员和医疗保健专业人员的信息指南。它描述了最近出现的模式，当前的问题，以及深度学习改变脑肿瘤诊断和治疗的机会。

{"title":"Brain tumor detection, classification and segmentation by deep learning models from MRI images: Recent approaches, challenges and future directions","authors":"Tahasin Ahmed Fahim , Fatema Binte Alam , Md Azad Hossain","doi":"10.1016/j.array.2025.100571","DOIUrl":"10.1016/j.array.2025.100571","url":null,"abstract":"<div><div>The diagnosis of brain tumors presents a very important challenge in the neuro-oncology field because of the complexities, heterogeneity, and high mortality of the tumors. The latest trends in deep learning have revolutionized the research in the field of medical image analysis. Through these trends, automated and precise brain tumor-detection, classification, and segmentation have been accomplished. This paper provides a systematic review with the focus taxonomy of the development of brain tumor analysis models with the most advanced deep neural networks based on tasks, and methods. It also summarizes the information of commonly available public datasets, preprocessing methods for MRI images, performance evaluation metrics. It reviews in detail on deep learning models available for brain tumor detection,classification, and segmentation in terms of performance metrics and clinical relevances. For representing in organized way, all the works reviewed here are divided into several groups and compared on specific benchmarks. Moreover, it figures out current challenges regarding brain tumor diagnosis and the potential implications of future studies to increase clinical applicability and trustworthiness of AI-driven solutions. This review acts as an informational guide to any researchers and healthcare professionals. It describes recent emerging patterns, current issues, and opportunities of deep learning to transform the diagnosis and treatment of brain tumors.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100571"},"PeriodicalIF":4.5,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145568367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Securing wearable healthcare devices: An active data framework 保护可穿戴医疗设备：一个活动数据框架

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-13 DOI: 10.1016/j.array.2025.100582

Egor Litvinov , Jesus Fernandez-Bermejo , Cristina Bolaños , Javier Dorado , Maria J. Santofimia , Felix J. Villanueva

This paper tackles the pressing security challenges of wearable healthcare devices within the internet of medical things, which are especially vulnerable due to limited processing power and reliance on wireless communication. We introduce the active data framework, a novel paradigm in which data itself is endowed with embedded logic and metadata to autonomously detect anomalies and trigger adaptive security responses based on context. The key strengths of this approach lie in its fine-grained, dynamic, and lightweight protection strategy, which differs from conventional static frameworks by enabling data to react in real time to threats. The framework combines lightweight cryptography, obfuscation techniques, and a context-aware anomaly detection module, all optimized for resource-constrained environments. We validate the approach through a real-world case study on fall detection with automatic alerting, demonstrating that the system maintains high performance and responsiveness while significantly enhancing data integrity and security. These results underline the potential of active data to advance trustworthy, autonomous, and efficient security for future internet of medical things applications.

本文解决了医疗物联网中可穿戴医疗设备面临的紧迫安全挑战，由于有限的处理能力和对无线通信的依赖，这些设备尤其容易受到攻击。我们引入了主动数据框架，这是一种新的范式，其中数据本身被赋予嵌入式逻辑和元数据，以自主检测异常并触发基于上下文的自适应安全响应。这种方法的主要优势在于其细粒度、动态和轻量级的保护策略，与传统的静态框架不同，它使数据能够实时对威胁做出反应。该框架结合了轻量级加密技术、混淆技术和上下文感知异常检测模块，所有这些都针对资源受限的环境进行了优化。我们通过一个具有自动警报的跌倒检测的实际案例研究验证了该方法，证明该系统在保持高性能和响应性的同时显著增强了数据完整性和安全性。这些结果强调了主动数据在推动未来医疗物联网应用的可信、自主和高效安全方面的潜力。

{"title":"Securing wearable healthcare devices: An active data framework","authors":"Egor Litvinov , Jesus Fernandez-Bermejo , Cristina Bolaños , Javier Dorado , Maria J. Santofimia , Felix J. Villanueva","doi":"10.1016/j.array.2025.100582","DOIUrl":"10.1016/j.array.2025.100582","url":null,"abstract":"<div><div>This paper tackles the pressing security challenges of wearable healthcare devices within the internet of medical things, which are especially vulnerable due to limited processing power and reliance on wireless communication. We introduce the active data framework, a novel paradigm in which data itself is endowed with embedded logic and metadata to autonomously detect anomalies and trigger adaptive security responses based on context. The key strengths of this approach lie in its fine-grained, dynamic, and lightweight protection strategy, which differs from conventional static frameworks by enabling data to react in real time to threats. The framework combines lightweight cryptography, obfuscation techniques, and a context-aware anomaly detection module, all optimized for resource-constrained environments. We validate the approach through a real-world case study on fall detection with automatic alerting, demonstrating that the system maintains high performance and responsiveness while significantly enhancing data integrity and security. These results underline the potential of active data to advance trustworthy, autonomous, and efficient security for future internet of medical things applications.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100582"},"PeriodicalIF":4.5,"publicationDate":"2025-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516911","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

MinkUNeXt: Point cloud-based large-scale place recognition using 3D sparse convolutions MinkUNeXt：使用3D稀疏卷积的基于点云的大规模位置识别

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-11 DOI: 10.1016/j.array.2025.100569

Juan José Cabrera , Antonio Santo , Arturo Gil , Carlos Viegas , Luis Payá

This paper presents MinkUNeXt, an effective and efficient architecture for place-recognition from point clouds entirely based on the new 3D MinkNeXt Block, a residual block composed of 3D sparse convolutions that follows the philosophy established by recent Transformers but purely using simple 3D convolutions. Feature extraction is performed at different scales by a U-Net encoder–decoder network and the feature aggregation of those features into a single descriptor is carried out by a Generalized Mean Pooling (GeM). The proposed architecture demonstrates that it is possible to surpass the current state-of-the-art by only relying on conventional 3D sparse convolutions without making use of more complex and sophisticated proposals such as Transformers, Attention-Layers or Deformable Convolutions. A thorough assessment of the proposal has been carried out using the Oxford RobotCar, the In-house, the KITTI and the USyd datasets. As a result, MinkUNeXt proves to outperform other methods in the state-of-the-art. The implementation is publicly available at https://juanjo-cabrera.github.io/projects-MinkUNeXt/.

本文介绍了MinkUNeXt，一种有效且高效的点云位置识别架构，完全基于新的3D MinkNeXt块，这是一个由3D稀疏卷积组成的残差块，遵循最近的变形金刚建立的哲学，但纯粹使用简单的3D卷积。通过U-Net编码器-解码器网络在不同尺度上进行特征提取，并通过广义均值池（GeM）将这些特征聚集到单个描述符中。所提出的架构表明，仅依靠传统的3D稀疏卷积，而不使用更复杂和复杂的方案（如变形金刚、注意力层或可变形卷积），就有可能超越当前最先进的技术。使用Oxford RobotCar、the house、KITTI和USyd数据集对该提案进行了彻底的评估。因此，MinkUNeXt证明在最先进的技术中优于其他方法。该实现可在https://juanjo-cabrera.github.io/projects-MinkUNeXt/上公开获得。

{"title":"MinkUNeXt: Point cloud-based large-scale place recognition using 3D sparse convolutions","authors":"Juan José Cabrera , Antonio Santo , Arturo Gil , Carlos Viegas , Luis Payá","doi":"10.1016/j.array.2025.100569","DOIUrl":"10.1016/j.array.2025.100569","url":null,"abstract":"<div><div>This paper presents MinkUNeXt, an effective and efficient architecture for place-recognition from point clouds entirely based on the new 3D MinkNeXt Block, a residual block composed of 3D sparse convolutions that follows the philosophy established by recent Transformers but purely using simple 3D convolutions. Feature extraction is performed at different scales by a U-Net encoder–decoder network and the feature aggregation of those features into a single descriptor is carried out by a Generalized Mean Pooling (GeM). The proposed architecture demonstrates that it is possible to surpass the current state-of-the-art by only relying on conventional 3D sparse convolutions without making use of more complex and sophisticated proposals such as Transformers, Attention-Layers or Deformable Convolutions. A thorough assessment of the proposal has been carried out using the Oxford RobotCar, the In-house, the KITTI and the USyd datasets. As a result, MinkUNeXt proves to outperform other methods in the state-of-the-art. The implementation is publicly available at <span><span>https://juanjo-cabrera.github.io/projects-MinkUNeXt/</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100569"},"PeriodicalIF":4.5,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516910","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

StatVis: A visual analytics framework for statistical cluster validation in high dimensions StatVis：用于高维统计聚类验证的可视化分析框架

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-11 DOI: 10.1016/j.array.2025.100560

Donia Y. Badawood

High-dimensional datasets are common across scientific and industrial domains, yet interpreting their clustering outcomes remains a complex challenge due to projection distortions and metric ambiguities. Traditional visualization techniques, such as Principal Component Analysis (PCA) or t-Distributed Stochastic Neighbor Embedding (t-SNE), often obscure critical information about cluster validity, separability, and density. This paper introduces StatVis, a statistical manifold learning framework designed to enhance Visual Cluster Validation (VCV) in high-dimensional spaces. The framework combines Dimensionality Reduction (DR) with internal validation metrics and density estimation to generate interpretable, statistically grounded cluster visualizations. StatVis integrates manifold learning techniques (Uniform Manifold Approximation and Projection (UMAP), t-SNE) with clustering algorithms (k-means) and calculates multiple validation metrics, including the Silhouette Coefficient, Davies–Bouldin Index, and Dunn Index. Local densities are estimated using both kernel density estimation and k-nearest neighbor methods. Experiments were conducted on three standard datasets: the UCI Dry Bean, MNIST, and 20 Newsgroups datasets. Visual overlays, projection distortion maps, and density-aware correction routines were evaluated. StatVis demonstrated superior capability in identifying and visually communicating cluster quality and projection inconsistencies. It outperformed standard visualizations in conveying cluster separability and compactness. The density-aware correction routine further improved interpretability in regions of high distortion. StatVis effectively bridges statistical validation and visualization, enabling more reliable and explainable cluster analysis. The framework shows promise as a visual decision support tool for exploring high-dimensional data and incorporating human input into machine learning.

高维数据集在科学和工业领域都很常见，但由于投影失真和度量模糊，解释它们的聚类结果仍然是一个复杂的挑战。传统的可视化技术，如主成分分析（PCA）或t分布随机邻居嵌入（t-SNE），往往模糊了关于聚类有效性、可分离性和密度的关键信息。本文介绍了统计流形学习框架StatVis，该框架旨在增强高维空间中的视觉聚类验证（VCV）。该框架将降维（DR）与内部验证度量和密度估计相结合，以生成可解释的、基于统计的聚类可视化。StatVis将流形学习技术（均匀流形近似和投影（UMAP）， t-SNE）与聚类算法（k-means）集成在一起，并计算多个验证指标，包括Silhouette系数，Davies-Bouldin指数和Dunn指数。局部密度估计使用核密度估计和k近邻方法。实验在三个标准数据集上进行：UCI Dry Bean， MNIST和20新闻组数据集。评估了视觉叠加、投影失真图和密度感知校正程序。StatVis在识别和视觉传达聚类质量和投影不一致性方面表现出卓越的能力。它在传递簇的可分离性和紧凑性方面优于标准可视化。密度感知校正程序进一步提高了高失真区域的可解释性。StatVis有效地连接了统计验证和可视化，实现更可靠和可解释的聚类分析。该框架有望成为探索高维数据和将人类输入纳入机器学习的可视化决策支持工具。

{"title":"StatVis: A visual analytics framework for statistical cluster validation in high dimensions","authors":"Donia Y. Badawood","doi":"10.1016/j.array.2025.100560","DOIUrl":"10.1016/j.array.2025.100560","url":null,"abstract":"<div><div>High-dimensional datasets are common across scientific and industrial domains, yet interpreting their clustering outcomes remains a complex challenge due to projection distortions and metric ambiguities. Traditional visualization techniques, such as Principal Component Analysis (PCA) or t-Distributed Stochastic Neighbor Embedding (t-SNE), often obscure critical information about cluster validity, separability, and density. This paper introduces StatVis, a statistical manifold learning framework designed to enhance Visual Cluster Validation (VCV) in high-dimensional spaces. The framework combines Dimensionality Reduction (DR) with internal validation metrics and density estimation to generate interpretable, statistically grounded cluster visualizations. StatVis integrates manifold learning techniques (Uniform Manifold Approximation and Projection (UMAP), t-SNE) with clustering algorithms (k-means) and calculates multiple validation metrics, including the Silhouette Coefficient, Davies–Bouldin Index, and Dunn Index. Local densities are estimated using both kernel density estimation and k-nearest neighbor methods. Experiments were conducted on three standard datasets: the UCI Dry Bean, MNIST, and 20 Newsgroups datasets. Visual overlays, projection distortion maps, and density-aware correction routines were evaluated. StatVis demonstrated superior capability in identifying and visually communicating cluster quality and projection inconsistencies. It outperformed standard visualizations in conveying cluster separability and compactness. The density-aware correction routine further improved interpretability in regions of high distortion. StatVis effectively bridges statistical validation and visualization, enabling more reliable and explainable cluster analysis. The framework shows promise as a visual decision support tool for exploring high-dimensional data and incorporating human input into machine learning.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100560"},"PeriodicalIF":4.5,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing trust in machines integration with Dirichlet distribution and self-sovereign identity 通过Dirichlet分布和自主身份增强对机器集成的信任

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-11-11 DOI: 10.1016/j.array.2025.100579

Joel Herve Mboussam Emati , Vianney Kengne Tchendji , Mounirah Djam-Doudou

In critical IoT ecosystems, where autonomous devices operate without human oversight, security hinges on robust identity management as the foundation of authentication. Traditional methods struggle to address dynamic threats and scalability demands, creating vulnerabilities in machine-to-machine trust. Self-sovereign identity (SSI) emerges as a transformative paradigm, enabling decentralized, cryptographically assured authentication while ensuring operational resilience. This paper proposes a dual-blockchain framework for IoT authentication, integrating cryptographic identity verification with behavioral trust analytics. The architecture consists of a decentralized identity blockchain for credential issuance and lifecycle management, alongside a transaction blockchain for validating device interactions, both coordinated through smart contracts across trust authorities, devices, and full nodes. The protocol combines static authentication using signed credentials with continuous authentication via a machine learning-enhanced Dirichlet trust model, enabling real-time behavioral anomaly detection. Implemented on Hyperledger family, the framework achieves 92.3% accuracy in detecting behavioral anomalies with only a 7.1% false positive rate, maintaining computational efficiency while resisting forgery and Byzantine attacks.

在关键的物联网生态系统中，自主设备在没有人为监督的情况下运行，安全性取决于强大的身份管理作为身份验证的基础。传统方法难以解决动态威胁和可伸缩性需求，从而在机器对机器信任中产生漏洞。自我主权身份（SSI）作为一种变革范例出现，在确保操作弹性的同时，实现了分散的、加密保证的身份验证。本文提出了一种用于物联网认证的双区块链框架，将加密身份验证与行为信任分析相结合。该体系结构包括用于证书颁发和生命周期管理的去中心化身份区块链，以及用于验证设备交互的事务区块链，两者都通过跨信任机构、设备和完整节点的智能合约进行协调。该协议将使用签名凭证的静态身份验证与通过机器学习增强的Dirichlet信任模型进行的连续身份验证相结合，从而实现实时行为异常检测。该框架在超级账本家族上实现，检测行为异常的准确率达到92.3%，假阳性率仅为7.1%，在保持计算效率的同时，还能抵抗伪造和拜占庭攻击。

{"title":"Enhancing trust in machines integration with Dirichlet distribution and self-sovereign identity","authors":"Joel Herve Mboussam Emati , Vianney Kengne Tchendji , Mounirah Djam-Doudou","doi":"10.1016/j.array.2025.100579","DOIUrl":"10.1016/j.array.2025.100579","url":null,"abstract":"<div><div>In critical IoT ecosystems, where autonomous devices operate without human oversight, security hinges on robust identity management as the foundation of authentication. Traditional methods struggle to address dynamic threats and scalability demands, creating vulnerabilities in machine-to-machine trust. Self-sovereign identity (SSI) emerges as a transformative paradigm, enabling decentralized, cryptographically assured authentication while ensuring operational resilience. This paper proposes a dual-blockchain framework for IoT authentication, integrating cryptographic identity verification with behavioral trust analytics. The architecture consists of a decentralized identity blockchain for credential issuance and lifecycle management, alongside a transaction blockchain for validating device interactions, both coordinated through smart contracts across trust authorities, devices, and full nodes. The protocol combines static authentication using signed credentials with continuous authentication via a machine learning-enhanced Dirichlet trust model, enabling real-time behavioral anomaly detection. Implemented on Hyperledger family, the framework achieves 92.3% accuracy in detecting behavioral anomalies with only a 7.1% false positive rate, maintaining computational efficiency while resisting forgery and Byzantine attacks.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100579"},"PeriodicalIF":4.5,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Array最新文献