Informatics

20 pages, 2193 KB

Open AccessArticle

Tourist Flow Prediction Based on GA-ACO-BP Neural Network Model

by Xiang Yang, Yongliang Cheng, Minggang Dong and Xiaolan Xie

Informatics 2025, 12(3), 89; https://doi.org/10.3390/informatics12030089 (registering DOI) - 3 Sep 2025

Tourist flow prediction plays a crucial role in enhancing the efficiency of scenic area management, optimizing resource allocation, and promoting the sustainable development of the tourism industry. To improve the accuracy and real-time performance of tourist flow prediction, we propose a BP model [...] Read more.

Tourist flow prediction plays a crucial role in enhancing the efficiency of scenic area management, optimizing resource allocation, and promoting the sustainable development of the tourism industry. To improve the accuracy and real-time performance of tourist flow prediction, we propose a BP model based on a hybrid genetic algorithm (GA) and ant colony optimization algorithm (ACO), called the GA-ACO-BP model. First, we comprehensively considered multiple key factors related to tourist flow, including historical tourist flow data (such as tourist flow from yesterday, the previous day, and the same period last year), holiday types, climate comfort, and search popularity index on online map platforms. Second, to address the tendency of the BP model to get easily stuck in local optima, we introduce the GA, which has excellent global search capabilities. Finally, to further improve local convergence speed, we further introduce the ACO algorithm. The experimental results based on tourist flow data from the Elephant Trunk Hill Scenic Area in Guilin indicate that the GA-AC*O-BP model achieves optimal values for key tourist flow prediction metrics such as MAPE, RMSE, MAE, and R², compared to commonly used prediction models. These values are 4.09%, 426.34, 258.80, and 0.98795, respectively. Compared to the initial BP neural network, the improved GA-ACO-BP model reduced error metrics such as MAPE, RMSE, and MAE by 1.12%, 244.04, and 122.91, respectively, and increased the R² metric by 1.85%. Full article

(This article belongs to the Topic The Applications of Artificial Intelligence in Tourism)

23 pages, 2012 KB

Open AccessArticle

Preliminary Design Guidelines for Evaluating Immersive Industrial Safety Training

by André Cordeiro, Regina Leite, Lucas Almeida, Cintia Neves, Tiago Silva, Alexandre Siqueira, Marcio Catapan and Ingrid Winkler

Informatics 2025, 12(3), 88; https://doi.org/10.3390/informatics12030088 - 1 Sep 2025

Abstract

This study presents preliminary design guidelines to support the evaluation of industrial safety training using immersive technologies, with a focus on high-risk work environments such as working at height. Although virtual reality has been widely adopted for training, few studies have explored its [...] Read more.

This study presents preliminary design guidelines to support the evaluation of industrial safety training using immersive technologies, with a focus on high-risk work environments such as working at height. Although virtual reality has been widely adopted for training, few studies have explored its use for behavior-level evaluation, corresponding to Level 3 of the Kirkpatrick Model. Addressing this gap, the study adopts the Design Science Research methodology, combining a systematic literature review with expert focus group analysis to develop a conceptual framework for training evaluation. The results identify key elements necessary for immersive training evaluations, including scenario configuration, ethical procedures, recruitment, equipment selection, experimental design, and implementation strategies. The resulting guidelines are organized into six categories: scenario configuration, ethical procedures, recruitment, equipment selection, experimental design, and implementation strategies. These guidelines represent a DSR-based conceptual artifact to inform future empirical studies and support the structured assessment of immersive safety training interventions. The study also highlights the potential of integrating behavioral and physiological indicators to support immersive evaluations of behavioral change, offering an expert-informed and structured foundation for future empirical studies in high-risk industrial contexts. Full article

(This article belongs to the Special Issue Real-World Applications and Prototyping of Information Systems for Extended Reality (VR, AR, and MR))

► Show Figures

Figure 1

24 pages, 1389 KB

Open AccessArticle

Analysis and Forecasting of Cryptocurrency Markets Using Bayesian and LSTM-Based Deep Learning Models

by Bidesh Biswas Biki, Makoto Sakamoto, Amane Takei, Md. Jubirul Alam, Md. Riajuliislam and Showaibuzzaman Showaibuzzaman

Informatics 2025, 12(3), 87; https://doi.org/10.3390/informatics12030087 - 30 Aug 2025

Viewed by 166

Abstract

The rapid rise of the prices of cryptocurrencies has intensified the need for robust forecasting models that can capture the irregular and volatile patterns. This study aims to forecast Bitcoin prices over a 15-day horizon by evaluating and comparing two distant predictive modeling [...] Read more.

The rapid rise of the prices of cryptocurrencies has intensified the need for robust forecasting models that can capture the irregular and volatile patterns. This study aims to forecast Bitcoin prices over a 15-day horizon by evaluating and comparing two distant predictive modeling approaches: the Bayesian State-Space model and Long Short-Term Memory (LSTM) neural networks. Historical price data from January 2024 to April 2025 is used for model training and testing. The Bayesian model provided probabilistic insights by achieving a Mean Squared Error (MSE) of 0.0000 and a Mean Absolute Error (MAE) of 0.0026 for training data. For testing data, it provided 0.0013 for MSE and 0.0307 for MAE. On the other hand, the LSTM model provided temporal dependencies and performed strongly by achieving 0.0004 for MSE, 0.0160 for MAE, 0.0212 for RMSE, 0.9924 for R² in terms of training data and for testing data, and 0.0007 for MSE with an R² of 0.3505. From the result, it indicates that while the LSTM model excels in training performance, the Bayesian model provides better interpretability with lower error margins in testing by highlighting the trade-offs between model accuracy and probabilistic forecasting in the cryptocurrency markets. Full article

► Show Figures

Figure 1

20 pages, 592 KB

Open AccessReview

The Temporal Evolution of Large Language Model Performance: A Comparative Analysis of Past and Current Outputs in Scientific and Medical Research

by Ishith Seth, Gianluca Marcaccini, Bryan Lim, Jennifer Novo, Stephen Bacchi, Roberto Cuomo, Richard J. Ross and Warren M. Rozen

Informatics 2025, 12(3), 86; https://doi.org/10.3390/informatics12030086 - 26 Aug 2025

Viewed by 371

Abstract

Background: Large language models (LLMs) such as ChatGPT have evolved rapidly, with notable improvements in coherence, factual accuracy, and contextual relevance. However, their academic and clinical applicability remains under scrutiny. This study evaluates the temporal performance evolution of LLMs by comparing earlier model [...] Read more.

Background: Large language models (LLMs) such as ChatGPT have evolved rapidly, with notable improvements in coherence, factual accuracy, and contextual relevance. However, their academic and clinical applicability remains under scrutiny. This study evaluates the temporal performance evolution of LLMs by comparing earlier model outputs (GPT-3.5 and GPT-4.0) with ChatGPT-4.5 across three domains: aesthetic surgery counseling, an academic discussion base of thumb arthritis, and a systematic literature review. Methods: We replicated the methodologies of three previously published studies using identical prompts in ChatGPT-4.5. Each output was assessed against its predecessor using a nine-domain Likert-based rubric measuring factual accuracy, completeness, reference quality, clarity, clinical insight, scientific reasoning, bias avoidance, utility, and interactivity. Expert reviewers in plastic and reconstructive surgery independently scored and compared model outputs across versions. Results: ChatGPT-4.5 outperformed earlier versions across all domains. Reference quality improved most significantly (a score increase of +4.5), followed by factual accuracy (+2.5), scientific reasoning (+2.5), and utility (+2.5). In aesthetic surgery counseling, GPT-3.5 produced generic responses lacking clinical detail, whereas ChatGPT-4.5 offered tailored, structured, and psychologically sensitive advice. In academic writing, ChatGPT-4.5 eliminated reference hallucination, correctly applied evidence hierarchies, and demonstrated advanced reasoning. In the literature review, recall remained suboptimal, but precision, citation accuracy, and contextual depth improved substantially. Conclusion: ChatGPT-4.5 represents a major step forward in LLM capability, particularly in generating trustworthy academic and clinical content. While not yet suitable as a standalone decision-making tool, its outputs now support research planning and early-stage manuscript preparation. Persistent limitations include information recall and interpretive flexibility. Continued validation is essential to ensure ethical, effective use in scientific workflows. Full article

► Show Figures

Figure 1

32 pages, 362 KB

Open AccessArticle

Human-AI Symbiotic Theory (HAIST): Development, Multi-Framework Assessment, and AI-Assisted Validation in Academic Research

by Laura Thomsen Morello and John C. Chick

Informatics 2025, 12(3), 85; https://doi.org/10.3390/informatics12030085 - 25 Aug 2025

Viewed by 526

Abstract

This study introduces the Human-AI Symbiotic Theory (HAIST), designed to guide authentic collaboration between human researchers and artificial intelligence in academic contexts, while pioneering a novel AI-assisted approach to theory validation that transforms educational research methodology. Addressing critical gaps in educational theory and [...] Read more.

This study introduces the Human-AI Symbiotic Theory (HAIST), designed to guide authentic collaboration between human researchers and artificial intelligence in academic contexts, while pioneering a novel AI-assisted approach to theory validation that transforms educational research methodology. Addressing critical gaps in educational theory and advancing validation practices, this research employed a sequential three-phase mixed-methods approach: (1) systematic theoretical synthesis integrating five paradigmatic perspectives across learning theory, cognition, information processing, ethics, and AI domains; (2) development of an innovative validation framework combining three established theory-building approaches with groundbreaking AI-assisted content assessment protocols; and (3) comprehensive theory validation through both traditional multi-framework evaluation and novel AI-based content analysis demonstrating unprecedented convergent validity. This research contributes both a theoretically grounded framework for human-AI research collaboration and a transformative methodological innovation demonstrating how AI tools can systematically augment traditional expert-driven theory validation. HAIST provides the first comprehensive theoretical foundation designed explicitly for human-AI partnerships in scholarly research with applicability across disciplines, while the AI-assisted validation methodology offers a scalable, reliable model for theory development. Future research directions include empirical testing of HAIST principles in live research settings and broader application of the AI-assisted validation methodology to accelerate theory development across educational research and related disciplines. Full article

(This article belongs to the Special Issue Generative AI in Higher Education: Applications, Implications, and Future Directions)

25 pages, 2448 KB

Open AccessArticle

Marketing a Banned Remedy: A Topic Model Analysis of Health Misinformation in Thai E-Commerce

by Kanitsorn Suriyapaiboonwattana, Yuttana Jaroenruen, Saiphit Satjawisate, Kate Hone, Panupong Puttarak, Nattapong Kaewboonma, Puriwat Lertkrai and Siwanath Nantapichai

Informatics 2025, 12(3), 84; https://doi.org/10.3390/informatics12030084 - 18 Aug 2025

Viewed by 839

Abstract

Unregulated herbal products marketed via digital platforms present escalating risks to consumer safety and regulatory effectiveness worldwide. This study positions the case of Jindamanee herbal powder—a banned substance under Thai law—as a lens through which to examine broader challenges in digital health governance. [...] Read more.

Unregulated herbal products marketed via digital platforms present escalating risks to consumer safety and regulatory effectiveness worldwide. This study positions the case of Jindamanee herbal powder—a banned substance under Thai law—as a lens through which to examine broader challenges in digital health governance. Drawing on a dataset of 1546 product listings across major platforms (Facebook, TikTok, Shopee, and Lazada), we applied Latent Dirichlet Allocation (LDA) to identify prevailing promotional themes and compliance gaps. Despite explicit platform policies, 87.6% of listings appeared on Facebook. Medical claims, particularly for pain relief, featured in 77.6% of posts, while only 18.4% included any risk disclosure. These findings suggest a systematic exploitation of regulatory blind spots and consumer health anxieties, facilitated by templated cross-platform messaging. Anchored in Information Manipulation Theory and the Health Belief Model, the analysis offers theoretical insight into how misinformation is structured and sustained within digital commerce ecosystems. The Thai case highlights urgent implications for platform accountability, policy harmonization, and the design of algorithmic surveillance systems in global health product regulation. Full article

(This article belongs to the Section Health Informatics)

► Show Figures

Figure 1

10 pages, 477 KB

Open AccessArticle

Predictive Language Processing in Humans and Large Language Models: A Comparative Study of Contextual Dependencies

by Yifan Zhang and Kuzma Strelnikov

Informatics 2025, 12(3), 83; https://doi.org/10.3390/informatics12030083 - 15 Aug 2025

Viewed by 458

Abstract

Human language comprehension relies on predictive processing; however, the computational mechanisms underlying this phenomenon remain unclear. This study investigates these mechanisms using large language models (LLMs), specifically GPT-3.5-turbo and GPT-4. We conducted a comparison of LLM and human performance on a phrase-completion task [...] Read more.

Human language comprehension relies on predictive processing; however, the computational mechanisms underlying this phenomenon remain unclear. This study investigates these mechanisms using large language models (LLMs), specifically GPT-3.5-turbo and GPT-4. We conducted a comparison of LLM and human performance on a phrase-completion task under varying levels of contextual cues (high, medium, and low) as defined using human performance, thereby enabling direct AI–human comparisons. Our findings indicate that LLMs significantly outperform humans, particularly in medium- and low-context conditions. While success in medium-context scenarios reflects the efficient utilization of contextual information, performance in low-context situations—where LLMs achieved approximately 25% accuracy compared to just 1% for humans—suggests that the models harness deep linguistic structures beyond mere surface context. This discovery implies that LLMs may elucidate previously unknown aspects of language architecture. The ability of LLMs to exploit deep structural regularities and statistical patterns in medium- and low-predictability contexts offers a novel perspective on the computational architecture of the human language system. Full article

(This article belongs to the Section Human-Computer Interaction)

► Show Figures

Figure 1

37 pages, 5086 KB

Open AccessArticle

Global Embeddings, Local Signals: Zero-Shot Sentiment Analysis of Transport Complaints

by Aliya Nugumanova, Daniyar Rakhimzhanov and Aiganym Mansurova

Informatics 2025, 12(3), 82; https://doi.org/10.3390/informatics12030082 - 14 Aug 2025

Viewed by 680

Abstract

Public transport agencies must triage thousands of multilingual complaints every day, yet the cost of training and serving fine-grained sentiment analysis models limits real-time deployment. The proposed “one encoder, any facet” framework therefore offers a reproducible, resource-efficient alternative to heavy fine-tuning for domain-specific [...] Read more.

Public transport agencies must triage thousands of multilingual complaints every day, yet the cost of training and serving fine-grained sentiment analysis models limits real-time deployment. The proposed “one encoder, any facet” framework therefore offers a reproducible, resource-efficient alternative to heavy fine-tuning for domain-specific sentiment analysis or opinion mining tasks on digital service data. To the best of our knowledge, we are the first to test this paradigm on operational multilingual complaints, where public transport agencies must prioritize thousands of Russian- and Kazakh-language messages each day. A human-labelled corpus of 2400 complaints is embedded with five open-source universal models. Obtained embeddings are matched to semantic “anchor” queries that describe three distinct facets: service aspect (eight classes), implicit frustration, and explicit customer request. In the strict zero-shot setting, the best encoder reaches 77% accuracy for aspect detection, 74% for frustration, and 80% for request; taken together, these signals reproduce human four-level priority in 60% of cases. Attaching a single-layer logistic probe on top of the frozen embeddings boosts performance to 89% for aspect, 83–87% for the binary facets, and 72% for end-to-end triage. Compared with recent fine-tuned sentiment analysis systems, our pipeline cuts memory demands by two orders of magnitude and eliminates task-specific training yet narrows the accuracy gap to under five percentage points. These findings indicate that a single frozen encoder, guided by handcrafted anchors and an ultra-light head, can deliver near-human triage quality across multiple pragmatic dimensions, opening the door to low-cost, language-agnostic monitoring of digital-service feedback. Full article

(This article belongs to the Special Issue Practical Applications of Sentiment Analysis)

► Show Figures

Figure 1

21 pages, 1977 KB

Open AccessArticle

A Flexible Profile-Based Recommender System for Discovering Cultural Activities in an Emerging Tourist Destination

by Isabel Arregocés-Julio, Andrés Solano-Barliza, Aida Valls, Antonio Moreno, Marysol Castillo-Palacio, Melisa Acosta-Coll and José Escorcia-Gutierrez

Informatics 2025, 12(3), 81; https://doi.org/10.3390/informatics12030081 - 14 Aug 2025

Viewed by 437

Abstract

Recommendation systems applied to tourism are widely recognized for improving the visitor’s experience in tourist destinations, thanks to their ability to personalize the trip. This paper presents a hybrid approach that combines Machine Learning techniques with the Ordered Weighted Averaging (OWA) aggregation operator [...] Read more.

Recommendation systems applied to tourism are widely recognized for improving the visitor’s experience in tourist destinations, thanks to their ability to personalize the trip. This paper presents a hybrid approach that combines Machine Learning techniques with the Ordered Weighted Averaging (OWA) aggregation operator to achieve greater accuracy in user segmentation and generate personalized recommendations. The data were collected through a questionnaire applied to tourists in the different points of interest of the Special, Tourist and Cultural District of Riohacha. In the first stage, the K-means algorithm defines the segmentation of tourists based on their socio-demographic data and travel preferences. The second stage uses the OWA operator with a disjunctive policy to assign the most relevant cluster given the input data. This hybrid approach provides a recommendation mechanism for tourist destinations and their cultural heritage. Full article

(This article belongs to the Topic The Applications of Artificial Intelligence in Tourism)

► Show Figures

Figure 1

25 pages, 7900 KB

Open AccessArticle

Multi-Label Disease Detection in Chest X-Ray Imaging Using a Fine-Tuned ConvNeXtV2 with a Customized Classifier

by Kangzhe Xiong, Yuyun Tu, Xinping Rao, Xiang Zou and Yingkui Du

Informatics 2025, 12(3), 80; https://doi.org/10.3390/informatics12030080 - 14 Aug 2025

Viewed by 625

Abstract

Deep-learning-based multiple label chest X-ray classification has achieved significant success, but existing models still have three main issues: fixed-scale convolutions fail to capture both large and small lesions, standard pooling is lacking in the lack of attention to important regions, and linear classification [...] Read more.

Deep-learning-based multiple label chest X-ray classification has achieved significant success, but existing models still have three main issues: fixed-scale convolutions fail to capture both large and small lesions, standard pooling is lacking in the lack of attention to important regions, and linear classification lacks the capacity to model complex dependency between features. To circumvent these obstacles, we propose CONVFCMAE, a lightweight yet powerful framework that is built on a backbone that is partially frozen (77.08 % of the initial layers are fixed) in order to preserve complex, multi-scale features while decreasing the number of trainable parameters. Our architecture adds (1) an intelligent global pooling module that is learnable, with

1 \times 1

convolutions that are dynamically weighted by their spatial location, and (2) a multi-head attention block that is dedicated to channel re-calibration, along with (3) a two-layer MLP that has been enhanced with ReLU, batch normalization, and dropout. This module is used to enhance the non-linearity of the feature space. To further reduce the noise associated with labels and the imbalance in class distribution inherent to the NIH ChestXray14 dataset, we utilize a combined loss that combines BCEWithLogits and Focal Loss as well as extensive data augmentation. On ChestXray14, the average ROC–AUC of CONVFCMAE is 0.852, which is 3.97 percent greater than the state of the art. Ablation experiments demonstrate the individual and collective effectiveness of each component. Grad-CAM visualizations have a superior capacity to localize the pathological regions, and this increases the interpretability of the model. Overall, CONVFCMAE provides a practical, generalizable solution to the problem of extracting features from medical images in a practical manner. Full article

(This article belongs to the Section Medical and Clinical Informatics)

► Show Figures

Figure 1

17 pages, 1210 KB

Open AccessArticle

CAMBSRec: A Context-Aware Multi-Behavior Sequential Recommendation Model

by Bohan Zhuang, Yan Lan and Minghui Zhang

Informatics 2025, 12(3), 79; https://doi.org/10.3390/informatics12030079 - 4 Aug 2025

Viewed by 571

Abstract

Multi-behavior sequential recommendation (MBSRec) is a form of sequential recommendation. It leverages users’ historical interaction behavior types to better predict their next actions. This approach fits real-world scenarios better than traditional models do. With the rise of the transformer model, attention mechanisms are [...] Read more.

Multi-behavior sequential recommendation (MBSRec) is a form of sequential recommendation. It leverages users’ historical interaction behavior types to better predict their next actions. This approach fits real-world scenarios better than traditional models do. With the rise of the transformer model, attention mechanisms are widely used in recommendation algorithms. However, they suffer from low-pass filtering, and the simple learnable positional encodings in existing models offer limited performance gains. To address these problems, we introduce the context-aware multi-behavior sequential recommendation model (CAMBSRec). It separately encodes items and behavior types, replaces traditional positional encoding with context-similarity positional encoding, and applies the discrete Fourier transform to separate the high and low frequency components and enhance the high frequency components, countering the low-pass filtering effect. Experiments on three public datasets show that CAMBSRec performs better than five baseline models, demonstrating its advantages in terms of recommendation performance. Full article

► Show Figures

Figure 1

13 pages, 1520 KB

Open AccessArticle

Designing a Patient Outcome Clinical Assessment Tool for Modified Rankin Scale: “You Feel the Same Way Too”

by Laura London and Noreen Kamal

Informatics 2025, 12(3), 78; https://doi.org/10.3390/informatics12030078 - 4 Aug 2025

Viewed by 517

Abstract

The modified Rankin Scale (mRS) is a widely used outcome measure for assessing disability in stroke care; however, its administration is often affected by subjectivity and variability, leading to poor inter-rater reliability and inconsistent scoring. Originally designed for hospital discharge evaluations, the mRS [...] Read more.

The modified Rankin Scale (mRS) is a widely used outcome measure for assessing disability in stroke care; however, its administration is often affected by subjectivity and variability, leading to poor inter-rater reliability and inconsistent scoring. Originally designed for hospital discharge evaluations, the mRS has evolved into an outcome tool for disability assessment and clinical decision-making. Inconsistencies persist due to a lack of standardization and cognitive biases during its use. This paper presents design principles for creating a standardized clinical assessment tool (CAT) for the mRS, grounded in human–computer interaction (HCI) and cognitive engineering principles. Design principles were informed in part by an anonymous online survey conducted with clinicians across Canada to gain insights into current administration practices, opinions, and challenges of the mRS. The proposed design principles aim to reduce cognitive load, improve inter-rater reliability, and streamline the administration process of the mRS. By focusing on usability and standardization, the design principles seek to enhance scoring consistency and improve the overall reliability of clinical outcomes in stroke care and research. Developing a standardized CAT for the mRS represents a significant step toward improving the accuracy and consistency of stroke disability assessments. Future work will focus on real-world validation with healthcare stakeholders and exploring self-completed mRS assessments to further refine the tool. Full article

► Show Figures

Figure 1

24 pages, 756 KB

Open AccessArticle

Designs and Interactions for Near-Field Augmented Reality: A Scoping Review

by Jacob Hobbs and Christopher Bull

Informatics 2025, 12(3), 77; https://doi.org/10.3390/informatics12030077 - 1 Aug 2025

Viewed by 736

Abstract

Augmented reality (AR), which overlays digital content within the user’s view, is gaining traction across domains such as education, healthcare, manufacturing, and entertainment. The hardware constraints of commercially available HMDs are well acknowledged, but little work addresses what design or interactions techniques developers [...] Read more.

Augmented reality (AR), which overlays digital content within the user’s view, is gaining traction across domains such as education, healthcare, manufacturing, and entertainment. The hardware constraints of commercially available HMDs are well acknowledged, but little work addresses what design or interactions techniques developers can employ or build into experiences to work around these limitations. We conducted a scoping literature review, with the aim of mapping the current landscape of design principles and interaction techniques employed in near-field AR environments. We searched for literature published between 2016 and 2025 across major databases, including the ACM Digital Library and IEEE Xplore. Studies were included if they explicitly employed design or interaction techniques with a commercially available HMD for near-field AR experiences. A total of 780 articles were returned by the search, but just 7 articles met the inclusion criteria. Our review identifies key themes around how existing techniques are employed and the two competing goals of AR experiences, and we highlight the importance of embodiment in interaction efficacy. We present directions for future research based on and justified by our review. The findings offer a comprehensive overview for researchers, designers, and developers aiming to create more intuitive, effective, and context-aware near-field AR experiences. This review also provides a foundation for future research by outlining underexplored areas and recommending research directions for near-field AR interaction design. Full article

► Show Figures

Figure 1

23 pages, 1192 KB

Open AccessArticle

Multi-Model Dialectical Evaluation of LLM Reasoning Chains: A Structured Framework with Dual Scoring Agents

by Catalin Anghel, Andreea Alexandra Anghel, Emilia Pecheanu, Ioan Susnea, Adina Cocu and Adrian Istrate

Informatics 2025, 12(3), 76; https://doi.org/10.3390/informatics12030076 - 1 Aug 2025

Viewed by 833

Abstract

(1) Background and objectives: Large language models (LLMs) such as GPT, Mistral, and LLaMA exhibit strong capabilities in text generation, yet assessing the quality of their reasoning—particularly in open-ended and argumentative contexts—remains a persistent challenge. This study introduces Dialectical Agent, an internally developed [...] Read more.

(1) Background and objectives: Large language models (LLMs) such as GPT, Mistral, and LLaMA exhibit strong capabilities in text generation, yet assessing the quality of their reasoning—particularly in open-ended and argumentative contexts—remains a persistent challenge. This study introduces Dialectical Agent, an internally developed modular framework designed to evaluate reasoning through a structured three-stage process: opinion, counterargument, and synthesis. The framework enables transparent and comparative analysis of how different LLMs handle dialectical reasoning. (2) Methods: Each stage is executed by a single model, and final syntheses are scored via two independent LLM evaluators (LLaMA 3.1 and GPT-4o) based on a rubric with four dimensions: clarity, coherence, originality, and dialecticality. In parallel, a rule-based semantic analyzer detects rhetorical anomalies and ethical values. All outputs and metadata are stored in a Neo4j graph database for structured exploration. (3) Results: The system was applied to four open-weight models (Gemma 7B, Mistral 7B, Dolphin-Mistral, Zephyr 7B) across ten open-ended prompts on ethical, political, and technological topics. The results show consistent stylistic and semantic variation across models, with moderate inter-rater agreement. Semantic diagnostics revealed differences in value expression and rhetorical flaws not captured by rubric scores. (4) Originality: The framework is, to our knowledge, the first to integrate multi-stage reasoning, rubric-based and semantic evaluation, and graph-based storage into a single system. It enables replicable, interpretable, and multidimensional assessment of generative reasoning—supporting researchers, developers, and educators working with LLMs in high-stakes contexts. Full article

► Show Figures

Figure 1

26 pages, 5535 KB

Open AccessArticle

Research on Power Cable Intrusion Identification Using a GRT-Transformer-Based Distributed Acoustic Sensing (DAS) System

by Xiaoli Huang, Xingcheng Wang, Han Qin and Zhaoliang Zhou

Informatics 2025, 12(3), 75; https://doi.org/10.3390/informatics12030075 - 21 Jul 2025

Viewed by 786

Abstract

To address the high false alarm rate of intrusion detection systems based on distributed acoustic sensing (DAS) for power cables in complex underground environments, an innovative GRT-Transformer multimodal deep learning model is proposed. The core of this model lies in its distinctive three-branch [...] Read more.

To address the high false alarm rate of intrusion detection systems based on distributed acoustic sensing (DAS) for power cables in complex underground environments, an innovative GRT-Transformer multimodal deep learning model is proposed. The core of this model lies in its distinctive three-branch parallel collaborative architecture: two branches employ Gramian Angular Summation Field (GASF) and Recursive Pattern (RP) algorithms to convert one-dimensional intrusion waveforms into two-dimensional images, thereby capturing rich spatial patterns and dynamic characteristics and the third branch utilizes a Gated Recurrent Unit (GRU) algorithm to directly focus on the temporal evolution features of the waveform; additionally, a Transformer component is integrated to capture the overall trend and global dependencies of the signals. Ultimately, the terminal employs a Bidirectional Long Short-Term Memory (BiLSTM) network to perform a deep fusion of the multidimensional features extracted from the three branches, enabling a comprehensive understanding of the bidirectional temporal dependencies within the data. Experimental validation demonstrates that the GRT-Transformer achieves an average recognition accuracy of 97.3% across three typical intrusion events—illegal tapping, mechanical operations, and vehicle passage—significantly reducing false alarms, surpassing traditional methods, and exhibiting strong practical potential in complex real-world scenarios. Full article

► Show Figures

Figure 1

15 pages, 2948 KB

Open AccessReview

A Comprehensive Review of ChatGPT in Teaching and Learning Within Higher Education

by Samkelisiwe Purity Phokoye, Siphokazi Dlamini, Peggy Pinky Mthalane, Mthokozisi Luthuli and Smangele Pretty Moyane

Informatics 2025, 12(3), 74; https://doi.org/10.3390/informatics12030074 - 21 Jul 2025

Viewed by 1966

Abstract

Artificial intelligence (AI) has become an integral component of various sectors, including higher education. AI, particularly in the form of advanced chatbots like ChatGPT, is increasingly recognized as a valuable tool for engagement in higher education institutions (HEIs). This growing trend highlights the [...] Read more.

Artificial intelligence (AI) has become an integral component of various sectors, including higher education. AI, particularly in the form of advanced chatbots like ChatGPT, is increasingly recognized as a valuable tool for engagement in higher education institutions (HEIs). This growing trend highlights the potential of AI to enhance student engagement and subsequently improve academic performance. Given this development, it is crucial for HEIs to delve deeper into the potential integration of AI-driven chatbots into educational practices. The aim of this study was to conduct a comprehensive review of the use of ChatGPT in teaching and learning within higher education. To offer a comprehensive viewpoint, it had two primary objectives: to identify the key factors influencing the adoption and acceptance of ChatGPT in higher education, and to investigate the roles of institutional policies and support systems in the acceptance of ChatGPT in higher education. A bibliometric analysis methodology was employed in this study, and a PRISMA diagram was used to explain the papers included in the analysis. The findings reveal the increasing adoption of ChatGPT within the higher education sector while also identifying the challenges faced during its implementation, ranging from technical issues to educational adaptations. Moreover, this review provides guidelines for various stakeholders to effectively integrate ChatGPT into higher education. Full article

► Show Figures

Figure 1

26 pages, 2596 KB

Open AccessArticle

DFPoLD: A Hard Disk Failure Prediction on Low-Quality Datasets

by Shuting Wei, Xiaoyu Lu, Hongzhang Yang, Chenfeng Tu, Jiangpu Guo, Hailong Sun and Yu Feng

Informatics 2025, 12(3), 73; https://doi.org/10.3390/informatics12030073 - 16 Jul 2025

Viewed by 649

Abstract

Hard disk failure prediction is an important proactive maintenance method for storage systems. Recent years have seen significant progress in hard disk failure prediction using high-quality SMART datasets. However, in industrial applications, data loss often occurs during SMART data collection, transmission, and storage. [...] Read more.

Hard disk failure prediction is an important proactive maintenance method for storage systems. Recent years have seen significant progress in hard disk failure prediction using high-quality SMART datasets. However, in industrial applications, data loss often occurs during SMART data collection, transmission, and storage. Existing machine learning-based hard disk failure prediction models perform poorly on low-quality datasets. Therefore, this paper proposes a hard disk fault prediction technique based on low-quality datasets. Firstly, based on the original Backblaze dataset, we construct a low-quality dataset, Backblaze-, by simulating sector damage in actual scenarios and deleting 10% to 99% of the data. Time series features like the Absolute Sum of First Difference (ASFD) were introduced to amplify the differences between positive and negative samples and reduce the sensitivity of the model to SMART data loss. Considering the impact of different quality datasets on time window selection, we propose a time window selection formula that selects different time windows based on the proportion of data loss. It is found that the poorer the dataset quality, the longer the time window selection should be. The proposed model achieves a True Positive Rate (TPR) of 99.46%, AUC of 0.9971, and F1 score of 0.9871, with a False Positive Rate (FPR) under 0.04%, even with 80% data loss, maintaining performance close to that on the original dataset. Full article

(This article belongs to the Section Big Data Mining and Analytics)

► Show Figures

Figure 1

24 pages, 1618 KB

Open AccessReview

Design Requirements of Breast Cancer Symptom-Management Apps

by Xinyi Huang, Amjad Fayoumi, Emily Winter and Anas Najdawi

Informatics 2025, 12(3), 72; https://doi.org/10.3390/informatics12030072 - 15 Jul 2025

Viewed by 958

Abstract

Many breast cancer patients follow a self-managed treatment pathway, which may lead to gaps in the data available to healthcare professionals, such as information about patients’ everyday symptoms at home. Mobile apps have the potential to bridge this information gap, leading to more [...] Read more.

Many breast cancer patients follow a self-managed treatment pathway, which may lead to gaps in the data available to healthcare professionals, such as information about patients’ everyday symptoms at home. Mobile apps have the potential to bridge this information gap, leading to more effective treatments and interventions, as well as helping breast cancer patients monitor and manage their symptoms. In this paper, we elicit design requirements for breast cancer symptom-management mobile apps using a systematic review following the PRISMA framework. We then evaluate existing cancer symptom-management apps found on the Apple store according to the extent to which they meet these requirements. We find that, whilst some requirements are well supported (such as functionality to record multiple symptoms and provision of information), others are currently not being met, particularly interoperability, functionality related to responses from healthcare professionals, and personalisation. Much work is needed for cancer patients and healthcare professionals to experience the benefits of digital health innovation. The article demonstrates a formal requirements model, in which requirements are categorised as functional and non-functional, and presents a proposal for conceptual design for future mobile apps. Full article

(This article belongs to the Section Health Informatics)

► Show Figures

Figure 1

20 pages, 1550 KB

Open AccessArticle

Strategy for Precopy Live Migration and VM Placement in Data Centers Based on Hybrid Machine Learning

by Taufik Hidayat, Kalamullah Ramli and Ruki Harwahyu

Informatics 2025, 12(3), 71; https://doi.org/10.3390/informatics12030071 - 15 Jul 2025

Viewed by 903

Abstract

Data center virtualization has grown rapidly alongside the expansion of application-based services but continues to face significant challenges, such as downtime caused by suboptimal hardware selection, load balancing, power management, incident response, and resource allocation. To address these challenges, this study proposes a [...] Read more.

Data center virtualization has grown rapidly alongside the expansion of application-based services but continues to face significant challenges, such as downtime caused by suboptimal hardware selection, load balancing, power management, incident response, and resource allocation. To address these challenges, this study proposes a combined machine learning method that uses an MDP to choose which VMs to move, the RF method to sort the VMs according to load, and NSGA-III to achieve multiple optimization objectives, such as reducing downtime, improving SLA, and increasing energy efficiency. For this model, the GWA-Bitbrains dataset was used, on which it had a classification accuracy of 98.77%, a MAPE of 7.69% in predicting migration duration, and an energy efficiency improvement of 90.80%. The results of real-world experiments show that the hybrid machine learning strategy could significantly reduce the data center workload, increase the total migration time, and decrease the downtime. The results of hybrid machine learning affirm the effectiveness of integrating the MDP, RF method, and NSGA-III for providing holistic solutions in VM placement strategies for large-scale data centers. Full article

(This article belongs to the Section Machine Learning)

► Show Figures

Figure 1

40 pages, 759 KB

Open AccessSystematic Review

Decoding Trust in Artificial Intelligence: A Systematic Review of Quantitative Measures and Related Variables

by Letizia Aquilino, Cinzia Di Dio, Federico Manzi, Davide Massaro, Piercosma Bisconti and Antonella Marchetti

Informatics 2025, 12(3), 70; https://doi.org/10.3390/informatics12030070 - 14 Jul 2025

Viewed by 2012

Abstract

As artificial intelligence (AI) becomes ubiquitous across various fields, understanding people’s acceptance and trust in AI systems becomes essential. This review aims to identify quantitative measures used to measure trust in AI and the associated studied elements. Following the PRISMA guidelines, three databases [...] Read more.

As artificial intelligence (AI) becomes ubiquitous across various fields, understanding people’s acceptance and trust in AI systems becomes essential. This review aims to identify quantitative measures used to measure trust in AI and the associated studied elements. Following the PRISMA guidelines, three databases were consulted, selecting articles published before December 2023. Ultimately, 45 articles out of 1283 were selected. Articles were included if they were peer-reviewed journal publications in English reporting empirical studies measuring trust in AI systems with multi-item questionnaires. Studies were analyzed through the lenses of cognitive and affective trust. We investigated trust definitions, questionnaires employed, types of AI systems, and trust-related constructs. Results reveal diverse trust conceptualizations and measurements. In addition, the studies covered a wide range of AI system types, including virtual assistants, content detection tools, chatbots, medical AI, robots, and educational AI. Overall, the studies show compatibility of cognitive or affective trust focus between theorization, items, experimental stimuli, and level of anthropomorphism of the systems. The review underlines the need to adapt measurement of trust in the specific characteristics of human–AI interaction, accounting for both the cognitive and affective sides. Trust definitions and measurement could be chosen depending also on the level of anthropomorphism of the systems and the context of application. Full article

► Show Figures

Figure 1

21 pages, 5069 KB

Open AccessArticle

A Patent-Based Technology Roadmap for AI-Powered Manipulators: An Evolutionary Analysis of the B25J Classification

by Yujia Zhai, Zehao Liu, Rui Zhao, Xin Zhang and Gengfeng Zheng

Informatics 2025, 12(3), 69; https://doi.org/10.3390/informatics12030069 - 11 Jul 2025

Viewed by 961

Abstract

Technology roadmapping is conducted by systematic mapping of technological evolution through patent analytics to inform innovation strategies. This study proposes an integrated framework combining hierarchical Latent Dirichlet Allocation (LDA) modeling with multiphase technology lifecycle theory, analyzing 113,449 Derwent patent abstracts (2008–2022) across three [...] Read more.

Technology roadmapping is conducted by systematic mapping of technological evolution through patent analytics to inform innovation strategies. This study proposes an integrated framework combining hierarchical Latent Dirichlet Allocation (LDA) modeling with multiphase technology lifecycle theory, analyzing 113,449 Derwent patent abstracts (2008–2022) across three dimensions: technological novelty, functional applications, and competitive advantages. By segmenting innovation stages via logistic growth curve modeling and optimizing topic extraction through perplexity validation, we constructed dynamic technology roadmaps to decode latent evolutionary patterns in AI-powered programmable manipulators (B25J classification) within an innovation trajectory. Key findings revealed: (1) a progressive transition from electromechanical actuation to sensor-integrated architectures, evidenced by 58% compound annual growth in embedded sensing patents; (2) application expansion from industrial automation (72% early stage patents) to precision medical operations, with surgical robotics growing 34% annually since 2018; and (3) continuous advancements in adaptive control algorithms, showing 2.7× growth in reinforcement learning implementations. The methodology integrates quantitative topic modeling (via pyLDAvis visualization and cosine similarity analysis) with qualitative lifecycle theory, addressing the limitations of conventional technology analysis methods by reconciling semantic granularity with temporal dynamics. The results identify core innovation trajectories—precision control, intelligent detection, and medical robotics—while highlighting emerging opportunities in autonomous navigation and human–robot collaboration. This framework provides empirically grounded strategic intelligence for R&D prioritization, cross-industry investment, and policy formulation in Industry 4.0. Full article

► Show Figures

Figure 1

32 pages, 4717 KB

Open AccessArticle

MOGAD: Integrated Multi-Omics and Graph Attention for the Discovery of Alzheimer’s Disease’s Biomarkers

by Zhizhong Zhang, Yuqi Chen, Changliang Wang, Maoni Guo, Lu Cai, Jian He, Yanchun Liang, Garry Wong and Liang Chen

Informatics 2025, 12(3), 68; https://doi.org/10.3390/informatics12030068 - 9 Jul 2025

Viewed by 939

Abstract

The selection of appropriate biomarkers in clinical practice aids in the early detection, treatment, and prevention of disease while also assisting in the development of targeted therapeutics. Recently, multi-omics data generated from advanced technology platforms has become available for disease studies. Therefore, the [...] Read more.

The selection of appropriate biomarkers in clinical practice aids in the early detection, treatment, and prevention of disease while also assisting in the development of targeted therapeutics. Recently, multi-omics data generated from advanced technology platforms has become available for disease studies. Therefore, the integration of this data with associated clinical data provides a unique opportunity to gain a deeper understanding of disease. However, the effective integration of large-scale multi-omics data remains a major challenge. To address this, we propose a novel deep learning model—the Multi-Omics Graph Attention biomarker Discovery network (MOGAD). MOGAD aims to efficiently classify diseases and discover biomarkers by integrating various omics data such as DNA methylation, gene expression, and miRNA expression. The model consists of three main modules: Multi-head GAT network (MGAT), Multi-Graph Attention Fusion (MGAF), and Attention Fusion (AF), which work together to dynamically model the complex relationships among different omics layers. We incorporate clinical data (e.g., APOE genotype) which enables a systematic investigation of the influence of non-omics factors on disease classification. The experimental results demonstrate that MOGAD achieves a superior performance compared to existing single-omics and multi-omics integration methods in classification tasks for Alzheimer’s disease (AD). In the comparative experiment on the ROSMAP dataset, our model achieved the highest ACC (0.773), F1-score (0.787), and MCC (0.551). The biomarkers identified by MOGAD show strong associations with the underlying pathogenesis of AD. We also apply a Hi-C dataset to validate the biological rationality of the identified biomarkers. Furthermore, the incorporation of clinical data enhances the model’s robustness and uncovers synergistic interactions between omics and non-omics features. Thus, our deep learning model is able to successfully integrate multi-omics data to efficiently classify disease and discover novel biomarkers. Full article

► Show Figures

Figure 1

24 pages, 1314 KB

Open AccessArticle

Balancing Accuracy and Efficiency in Vehicular Network Firmware Vulnerability Detection: A Fuzzy Matching Framework with Standardized Data Serialization

by Xiyu Fang, Kexun He, Yue Wu, Rui Chen and Jing Zhao

Informatics 2025, 12(3), 67; https://doi.org/10.3390/informatics12030067 - 9 Jul 2025

Viewed by 596

Abstract

Firmware vulnerabilities in embedded devices have caused serious security incidents, necessitating similarity analysis of binary program instruction embeddings to identify vulnerabilities. However, existing instruction embedding methods neglect program execution semantics, resulting in accuracy limitations. Furthermore, current embedding approaches utilize independent computation across models, [...] Read more.

Firmware vulnerabilities in embedded devices have caused serious security incidents, necessitating similarity analysis of binary program instruction embeddings to identify vulnerabilities. However, existing instruction embedding methods neglect program execution semantics, resulting in accuracy limitations. Furthermore, current embedding approaches utilize independent computation across models, where the lack of standardized interaction information between models makes it difficult for embedding models to efficiently detect firmware vulnerabilities. To address these challenges, this paper proposes a firmware vulnerability detection scheme based on statistical inference and code similarity fuzzy matching analysis for resource-constrained vehicular network environments, helping to balance both accuracy and efficiency. First, through dynamic programming and neighborhood search techniques, binary code is systematically partitioned into normalized segment collections according to specific rules. The binary code is then analyzed in segments to construct semantic equivalence mappings, thereby extracting similarity metrics for function execution semantics. Subsequently, Google Protocol Buffers (ProtoBuf) is introduced as a serialization format for inter-model data transmission, serving as a “translation layer” and “bridging technology” within the firmware vulnerability detection framework. Additionally, a ProtoBuf-based certificate authentication scheme is proposed to enhance vehicular network communication reliability, improve data serialization efficiency, and increase the efficiency and accuracy of the detection model. Finally, a vehicular network simulation environment is established through secondary development on the NS-3 network simulator, and the functionality and performance of this architecture were thoroughly tested. Results demonstrate that the algorithm possesses resistance capabilities against common security threats while minimizing performance impact. Experimental results show that FirmPB delivers superior accuracy with 0.044 s inference time and 0.932 AUC, outperforming current SOTA in detection performance. Full article

► Show Figures

Figure 1

24 pages, 2710 KB

Open AccessArticle

From Innovation to Regulation: Insights from a Bibliometric Analysis of Research Patterns in Medical Data Governance

by Iulian V. Nastasa, Andrada-Raluca Artamonov, Ștefan Sebastian Busnatu, Dana Galieta Mincă and Octavian Andronic

Informatics 2025, 12(3), 66; https://doi.org/10.3390/informatics12030066 - 8 Jul 2025

Viewed by 853

Abstract

This study presents a comprehensive bibliometric analysis of the evolving landscape of data protection in medicine, examining research trends, thematic developments, and scholarly contributions from the 1960s to 2024. By analyzing 2159 publications indexed in the Scopus database using the Bibliometrix R package [...] Read more.

This study presents a comprehensive bibliometric analysis of the evolving landscape of data protection in medicine, examining research trends, thematic developments, and scholarly contributions from the 1960s to 2024. By analyzing 2159 publications indexed in the Scopus database using the Bibliometrix R package (v.4.3.2), based on R (v.4.4.3), this paper maps key research areas, leading journals, and international collaboration patterns. Our findings reveal a significant shift in focus over time, from early concerns centered on data privacy and management to contemporary themes involving advanced technologies such as artificial intelligence, blockchain, and big data analytics. This transition reflects the increasing complexity of balancing data accessibility with security, ethical, and regulatory requirements in healthcare. This analysis also highlights persistent challenges, including fragmented research efforts, disparities in global contributions, and the ongoing need for interdisciplinary collaboration. These insights offer a valuable foundation for future investigations into medical data governance and emphasize the importance of ethical and responsible innovation in an increasingly digital healthcare environment. Full article

► Show Figures

Figure 1

25 pages, 4911 KB

Open AccessArticle

DA OMS-CNN: Dual-Attention OMS-CNN with 3D Swin Transformer for Early-Stage Lung Cancer Detection

by Yadollah Zamanidoost, Matis Rivron, Tarek Ould-Bachir and Sylvain Martel

Informatics 2025, 12(3), 65; https://doi.org/10.3390/informatics12030065 - 7 Jul 2025

Viewed by 822

Abstract

Lung cancer is one of the most prevalent and deadly forms of cancer, accounting for a significant portion of cancer-related deaths worldwide. It typically originates in the lung tissues, particularly in the cells lining the airways, and early detection is crucial for improving [...] Read more.

Lung cancer is one of the most prevalent and deadly forms of cancer, accounting for a significant portion of cancer-related deaths worldwide. It typically originates in the lung tissues, particularly in the cells lining the airways, and early detection is crucial for improving patient survival rates. Computed tomography (CT) imaging has become a standard tool for lung cancer screening, providing detailed insights into lung structures and facilitating the early identification of cancerous nodules. In this study, an improved Faster R-CNN model is employed to detect early-stage lung cancer. To enhance the performance of Faster R-CNN, a novel dual-attention optimized multi-scale CNN (DA OMS-CNN) architecture is used to extract representative features of nodules at different sizes. Additionally, dual-attention RoIPooling (DA-RoIpooling) is applied in the classification stage to increase the model’s sensitivity. In the false-positive reduction stage, a combination of multiple 3D shift window transformers (3D SwinT) is designed to reduce false-positive nodules. The proposed model was evaluated on the LUNA16 and PN9 datasets. The results demonstrate that integrating DA OMS-CNN, DA-RoIPooling, and 3D SwinT into the improved Faster R-CNN framework achieves a sensitivity of 96.93% and a CPM score of 0.911. Comprehensive experiments demonstrate that the proposed approach not only increases the sensitivity of lung cancer detection but also significantly reduces the number of false-positive nodules. Therefore, the proposed method can serve as a valuable reference for clinical applications. Full article

► Show Figures

Figure 1

16 pages, 1535 KB

Open AccessArticle

Clinical Text Classification for Tuberculosis Diagnosis Using Natural Language Processing and Deep Learning Model with Statistical Feature Selection Technique

by Shaik Fayaz Ahamed, Sundarakumar Karuppasamy and Ponnuraja Chinnaiyan

Informatics 2025, 12(3), 64; https://doi.org/10.3390/informatics12030064 - 7 Jul 2025

Viewed by 815

Abstract

Background: In the medical field, various deep learning (DL) algorithms have been effectively used to extract valuable information from unstructured clinical text data, potentially leading to more effective outcomes. This study utilized clinical text data to classify clinical case reports into tuberculosis (TB) [...] Read more.

Background: In the medical field, various deep learning (DL) algorithms have been effectively used to extract valuable information from unstructured clinical text data, potentially leading to more effective outcomes. This study utilized clinical text data to classify clinical case reports into tuberculosis (TB) and non-tuberculosis (non-TB) groups using natural language processing (NLP), a pre-processing technique, and DL models. Methods: This study used 1743 open-source respiratory disease clinical text data, labeled via fuzzy matching with ICD-10 codes to create a labeled dataset. Two tokenization methods preprocessed the clinical text data, and three models were evaluated: the existing Text-CNN, the proposed Text-CNN with t-test, and Bio_ClinicalBERT. Performance was assessed using multiple metrics and validated on 228 baseline screening clinical case text data collected from ICMR–NIRT to demonstrate effective TB classification. Results: The proposed model achieved the best results in both the test and validation datasets. On the test dataset, it attained a precision of 88.19%, a recall of 90.71%, an F1-score of 89.44%, and an AUC of 0.91. Similarly, on the validation dataset, it achieved 100% precision, 98.85% recall, 99.42% F1-score, and an AUC of 0.982, demonstrating its effectiveness in TB classification. Conclusions: This study highlights the effectiveness of DL models in classifying TB cases from clinical notes. The proposed model outperformed the other two models. The TF-IDF and t-test showed statistically significant feature selection and enhanced model interpretability and efficiency, demonstrating the potential of NLP and DL in automating TB diagnosis in clinical decision settings. Full article

► Show Figures

Figure 1

17 pages, 561 KB

Open AccessArticle

Web Accessibility in an Academic Management System in Brazil: Problems and Challenges for Attending People with Visual Impairments

by Mayra Correa, Maria Albeti Vitoriano and Carlos Humberto Llanos

Informatics 2025, 12(3), 63; https://doi.org/10.3390/informatics12030063 - 4 Jul 2025

Viewed by 696

Abstract

Accessibility in web systems is essential to ensure everyone can obtain information equally. Based on the Web Content Accessibility Guidelines (WCAGs), the Electronic Government Accessibility Model (eMAG) was established in Brazil to guide the accessibility of federal government web systems. Based on these [...] Read more.

Accessibility in web systems is essential to ensure everyone can obtain information equally. Based on the Web Content Accessibility Guidelines (WCAGs), the Electronic Government Accessibility Model (eMAG) was established in Brazil to guide the accessibility of federal government web systems. Based on these guidelines, this research sought to understand the reasons behind the persistent gaps in web accessibility in Brazil, even after 20 years of eMAG. To this end, the accessibility of the Integrated Academic Activities Management System (SIGAA), used by 39 higher education institutions in Brazil, was evaluated. The living lab methodology was used to carry out accessibility and usability tests based on students’ experiences with visual impairments during interaction with the system. Furthermore, IT professionals’ knowledge of eMAG/WCAG guidelines, the use of accessibility tools, and their beliefs about accessible systems were investigated through an online questionnaire. Additionally, the syllabuses of training courses for IT professionals at 20 universities were analyzed through document analysis. The research confirmed non-compliance with the guidelines in the software researched, gaps in the knowledge of IT professionals regarding software accessibility practices, and inadequacy of accessibility content within training courses. It is concluded, therefore, that universities should incorporate mandatory courses related to software accessibility into the training programs for IT professionals and that organizations should provide continuous training for IT professionals in software accessibility practices. Furthermore, the current accessibility legislation should be updated, and its compliance should be required within all organizations, whether public or private. Full article

► Show Figures

Figure 1

21 pages, 4241 KB

Open AccessArticle

Federated Learning-Driven Cybersecurity Framework for IoT Networks with Privacy Preserving and Real-Time Threat Detection Capabilities

by Milad Rahmati and Antonino Pagano

Informatics 2025, 12(3), 62; https://doi.org/10.3390/informatics12030062 - 4 Jul 2025

Cited by 1 | Viewed by 1516

Abstract

The rapid expansion of the Internet of Things (IoT) ecosystem has transformed industries but also exposed significant cybersecurity vulnerabilities. Traditional centralized methods for securing IoT networks struggle to balance privacy preservation with real-time threat detection. This study presents a Federated Learning-Driven Cybersecurity Framework [...] Read more.

The rapid expansion of the Internet of Things (IoT) ecosystem has transformed industries but also exposed significant cybersecurity vulnerabilities. Traditional centralized methods for securing IoT networks struggle to balance privacy preservation with real-time threat detection. This study presents a Federated Learning-Driven Cybersecurity Framework designed for IoT environments, enabling decentralized data processing through local model training on edge devices to ensure data privacy. Secure aggregation using homomorphic encryption supports collaborative learning without exposing sensitive information. The framework employs GRU-based recurrent neural networks (RNNs) for anomaly detection, optimized for resource-constrained IoT networks. Experimental results demonstrate over 98% accuracy in detecting threats such as distributed denial-of-service (DDoS) attacks, with a 20% reduction in energy consumption and a 30% reduction in communication overhead, showcasing the framework’s efficiency over traditional centralized approaches. This work addresses critical gaps in IoT cybersecurity by integrating federated learning with advanced threat detection techniques. It offers a scalable, privacy-preserving solution for diverse IoT applications, with future directions including blockchain integration for model aggregation traceability and quantum-resistant cryptography to enhance security. Full article

► Show Figures

Figure 1

25 pages, 2618 KB

Open AccessReview

International Trends and Influencing Factors in the Integration of Artificial Intelligence in Education with the Application of Qualitative Methods

by Juan Luis Cabanillas-García

Informatics 2025, 12(3), 61; https://doi.org/10.3390/informatics12030061 - 4 Jul 2025

Viewed by 1156

Abstract

This study offers a comprehensive examination of the scientific output related to the integration of Artificial Intelligence (AI) in education using qualitative research methods, which is an emerging intersection that reflects growing interest in understanding the pedagogical, ethical, and methodological implications of AI [...] Read more.

This study offers a comprehensive examination of the scientific output related to the integration of Artificial Intelligence (AI) in education using qualitative research methods, which is an emerging intersection that reflects growing interest in understanding the pedagogical, ethical, and methodological implications of AI in educational contexts. Grounded in a theoretical framework that emphasizes the potential of AI to support personalized learning, augment instructional design, and facilitate data-driven decision-making, this study conducts a Systematic Literature Review and bibliometric analysis of 630 publications indexed in Scopus between 2014 and 2024. The results show a significant increase in scholarly output, particularly since 2020, with notable contributions from authors and institutions in the United States, China, and the United Kingdom. High-impact research is found in top-tier journals, and dominant themes include health education, higher education, and the use of AI for feedback and assessment. The findings also highlight the role of semi-structured interviews, thematic analysis, and interdisciplinary approaches in capturing the nuanced impacts of AI integration. The study concludes that qualitative methods remain essential for critically evaluating AI’s role in education, reinforcing the need for ethically sound, human-centered, and context-sensitive applications of AI technologies in diverse learning environments. Full article

(This article belongs to the Section Social Informatics and Digital Humanities)

► Show Figures

Figure 1

35 pages, 1982 KB

Open AccessArticle

Predicting Mental Health Problems in Gay Men in Peru Using Machine Learning and Deep Learning Models

by Alejandro Aybar-Flores and Elizabeth Espinoza-Portilla

Informatics 2025, 12(3), 60; https://doi.org/10.3390/informatics12030060 - 2 Jul 2025

Viewed by 860

Abstract

Mental health disparities among those who self-identify as gay men in Peru remain a pressing public health concern, yet predictive models for early identification remain limited. This research aims to (1) develop machine learning and deep learning models to predict mental health issues [...] Read more.

Mental health disparities among those who self-identify as gay men in Peru remain a pressing public health concern, yet predictive models for early identification remain limited. This research aims to (1) develop machine learning and deep learning models to predict mental health issues in those who self-identify as gay men, and (2) evaluate the influence of demographic, economic, health-related, behavioral and social factors using interpretability techniques to enhance understanding of the factors shaping mental health outcomes. A dataset of 2186 gay men from the First Virtual Survey for LGBTIQ+ People in Peru (2017) was analyzed, considering demographic, economic, health-related, behavioral, and social factors. Several classification models were developed and compared, including Logistic Regression, Artificial Neural Networks, Random Forest, Gradient Boosting Machines, eXtreme Gradient Boosting, and a One-dimensional Convolutional Neural Network (1D-CNN). Additionally, the Shapley values and Layer-wise Relevance Propagation (LRP) heatmaps methods were used to evaluate the influence of the studied variables on the prediction of mental health issues. The results revealed that the 1D-CNN model demonstrated the strongest performance, achieving the highest classification accuracy and discrimination capability. Explainability analyses underlined prior infectious diseases diagnosis, access to medical assistance, experiences of discrimination, age, and sexual identity expression as key predictors of mental health outcomes. These findings suggest that advanced predictive techniques can provide valuable insights for identifying at-risk individuals, informing targeted interventions, and improving access to mental health care. Future research should refine these models to enhance predictive accuracy, broaden applicability, and support the integration of artificial intelligence into public health strategies aimed at addressing the mental health needs of this population. Full article

(This article belongs to the Section Health Informatics)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Informatics, Volume 12, Issue 3 (September 2025) – 33 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI