Informatics

20 pages, 14548 KB

Open AccessArticle

Distribution of Cognitive Load and Related Debates in Students Interacting with an Augmented Reality Application for History Learning

by Santiago Criollo-C, Oswaldo Moscoso-Zea, Yunifa Miftachul Arif and Sergio Luján-Mora

Informatics 2026, 13(8), 126; https://doi.org/10.3390/informatics13080126 - 7 Aug 2026

Abstract

The growing adoption of Augmented Reality (AR) in education raises challenges related to the cognitive load that students experience during learning. Inadequate management of this load can negatively affect information processing during learning activities. Although previous studies have explored AR applications, most have [...] Read more.

The growing adoption of Augmented Reality (AR) in education raises challenges related to the cognitive load that students experience during learning. Inadequate management of this load can negatively affect information processing during learning activities. Although previous studies have explored AR applications, most have focused on usability and learning outcomes, leaving aside the detailed analysis of Cognitive Load Theory (CLT), particularly its dimensions: intrinsic cognitive load (ICL), extraneous cognitive load (ECL) and germane cognitive load (GCL). To address this gap, this study evaluates cognitive load using the HistARium application, designed to support history learning with interactive experiences. A total of 60 students participated and completed a structured questionnaire to measure ICL, ECL, and GCL after using the application. The results show a positive distribution: moderate intrinsic load, low extraneous load, and high germane load. These findings indicate that the HistARium application balances content complexity, reduces unnecessary effort, and promotes cognitive processes associated with information organization and schema education construction. Furthermore, the results suggest that integrating CLT principles into AR applications supports cognitive processing during learning, while also opening new opportunities for the development of immersive and adaptive learning environments. 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

21 pages, 2378 KB

Open AccessArticle

Optimizing the Mammography AI Pipeline: From Data Filtering to Vision-Language Models

by Egor Ushakov, Sofya Zimina, Arsenii Litvinov, Sofia Senotrusova, Kirill Lukianov, Tigran G. Gevorkyan and Evgeny Karpulevich

Informatics 2026, 13(8), 125; https://doi.org/10.3390/informatics13080125 - 31 Jul 2026

Abstract

The performance of AI solutions in mammography is largely determined by data quality, preprocessing methods, and augmentation strategies. However, systematic evaluation of these factors for models trained on aggregated multicenter datasets remains underexplored. This article presents a comparative assessment of the effects of [...] Read more.

The performance of AI solutions in mammography is largely determined by data quality, preprocessing methods, and augmentation strategies. However, systematic evaluation of these factors for models trained on aggregated multicenter datasets remains underexplored. This article presents a comparative assessment of the effects of different stages of the training pipeline on the final diagnostic accuracy. Using a pooled dataset (VinDr-Mammo, INBreast, CMMD, CBIS-DDSM), we evaluated each pipeline step—from filtering to architecture selection (EfficientNet-B3, CLIP). External testing was conducted on the MosMed database. Among the tested preprocessing steps, filtering the darkest 5% of images proved most effective. For EfficientNet-B3, optimal geometric and photometric augmentations increased test AUROC on the prepared MosMed test set from 0.844 to 0.900. Domain-specific pretraining and high resolution yielded the best performance: Mammo-CLIP achieved an AUROC of 0.949 ± 0.012, and EfficientNet-B3 reached 0.934 ± 0.013. Overall, this study developed a standardized pipeline that includes sequential data filtering and harmonization, augmentation optimization, and architecture selection. This approach ensures reliable and reproducible results for automated mammogram classification. Full article

(This article belongs to the Special Issue AI-Enabled Digital Health Technologies for Patient-Centered Care and Sustainable Systems)

► Show Figures

Figure 1

29 pages, 1532 KB

Open AccessArticle

Causal-Pathway-Guided DNN–GBDT Distillation for Interpretable Artificial Intelligence in Intensive Care Units

by Hashim Ali

Informatics 2026, 13(8), 124; https://doi.org/10.3390/informatics13080124 - 30 Jul 2026

Abstract

Artificial intelligence (AI) systems for intensive care units (ICUs) must support early risk prediction while producing explanations that clinicians can inspect, question, and relate to physiological reasoning. Deep neural networks (DNNs) can learn complex temporal patterns from electronic health records (EHRs), but their [...] Read more.

Artificial intelligence (AI) systems for intensive care units (ICUs) must support early risk prediction while producing explanations that clinicians can inspect, question, and relate to physiological reasoning. Deep neural networks (DNNs) can learn complex temporal patterns from electronic health records (EHRs), but their internal representations are often difficult to translate into clinically actionable explanations. Gradient-boosted decision trees (GBDTs) offer more transparent decision rules, yet they may not capture the full temporal and nonlinear structure of high-dimensional ICU data. This paper presents a causal-pathway-guided DNN–GBDT distillation framework for interpretable ICU decision support. The framework first estimates a directed acyclic graph (DAG), denoted by

G

, from multivariate ICU time-series data and then uses the graph to guide representation learning in a DNN teacher model through causal gating. The learned teacher is distilled into a GBDT student model using soft predictive targets and a causal attribution-guided split-selection procedure, so that the final model approximates the teacher predictions while prioritizing tree splits aligned with plausible physiological pathways. Experiments using Medical Information Mart for Intensive Care IV (MIMIC-IV) data evaluate sepsis onset and in-hospital mortality prediction through discrimination, precision–recall performance, calibration-oriented reporting, causal consistency, and clinical utility indicators. The proposed causal-aware distilled GBDT achieves stronger predictive performance than conventional interpretable baselines and substantially higher causal consistency than black-box temporal models. The results suggest that causal structure can serve as an inductive bias for converting complex temporal prediction into interpretable rule-based clinical reasoning. The paper also discusses limitations related to observational causal discovery, unmeasured confounding, temporal stationarity, and clinical deployment, following recent reporting expectations for AI-based clinical prediction models. Full article

(This article belongs to the Section Health Informatics)

► Show Figures

Figure 1

20 pages, 6592 KB

Open AccessArticle

SE-POSTER: Channel-Enhanced Landmark Guided Transformer for Facial Emotion Recognition

by Alpamis Kutlimuratov, Kongratbay Sharipov, Piratdin Allayarov, Sayyora Iskandarova, Ruslan Latyfskiy, Gulchehra Tolibaeva and Fazliddin Makhmudov

Informatics 2026, 13(8), 123; https://doi.org/10.3390/informatics13080123 - 30 Jul 2026

Abstract

Recognizing facial emotions automatically from images/videos (FER) still represents a difficult problem for emotion computing, mainly due to variations in the face pose, lighting, occlusion, facial features, and expression intensity in the wild. Recent CNN–Transformer-based hybrid models like POSTER have leveraged local feature [...] Read more.

Recognizing facial emotions automatically from images/videos (FER) still represents a difficult problem for emotion computing, mainly due to variations in the face pose, lighting, occlusion, facial features, and expression intensity in the wild. Recent CNN–Transformer-based hybrid models like POSTER have leveraged local feature learning, landmark guidance, and global dependency modeling to achieve strong performance. Yet these methods give the main focus to spatial and contextual representations while not really going deep into adaptive channel-wise feature importance over multi-scale representations. As different feature channels represent emotions in varying degrees, it is likely that by treating all feature channels equally, one would limit the ability of the learned features to discriminate effectively. To overcome this weakness, this article presents a ResNet-18–Transformer landmark-guided module called SE-POSTER that fuses lightweight Squeeze-and-Excitation (SE) attention modules into the multi-scale feature pyramid of the baseline POSTER architecture. The proposed method carries out feature channel recalibration adaptively at the level of features before Transformer-based global attention modeling, thus allowing the network to focus on emotionally informative feature channels and suppress less relevant responses. The inclusion of SE attention in the network enhances fine, mid, and global levels of feature representations at a very low cost in terms of computation. On the basis of the RAF-DB, FERPlus, and AffectNet datasets, enormous experiments prove that the SE-POSTER framework proposed is capable of steadily boosting recognition accuracy relative to the baseline POSTER and several state-of-the-art FER methods. Especially, the proposed model delivers 92.78% accuracy on RAF-DB while it also shows better robustness and generalization capability under difficult real-world conditions. Moreover, additional ablation studies reveal that multi-level channel recalibration is effective in improving discriminative emotional feature learning. Full article

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

► Show Figures

Figure 1

21 pages, 2459 KB

Open AccessArticle

A Lightweight 3DMM-CNN Pipeline for Real-Time Single-Image 3D Face Reconstruction: Prototyping Personalised Avatars for Extended Reality Applications

by Qianqian He, Wirapong Chansanam, Lan Thi Nguyen, Kannikar Intawong and Kitti Puritat

Informatics 2026, 13(8), 122; https://doi.org/10.3390/informatics13080122 - 24 Jul 2026

Abstract

Personalised three-dimensional (3D) facial avatars underpin a wide range of immersive virtual, augmented, and mixed reality (XR) experiences, yet conventional 3D capture pipelines remain prohibitively expensive and computationally demanding for prototype-stage XR applications. This study presents and evaluates a lightweight hybrid 3D Morphable [...] Read more.

Personalised three-dimensional (3D) facial avatars underpin a wide range of immersive virtual, augmented, and mixed reality (XR) experiences, yet conventional 3D capture pipelines remain prohibitively expensive and computationally demanding for prototype-stage XR applications. This study presents and evaluates a lightweight hybrid 3D Morphable Model–Convolutional Neural Network (3DMM-CNN) pipeline that reconstructs an animation-ready 3D facial mesh from a single unconstrained RGB photograph and exposes it through an interactive prototype with native export to XR-ready asset formats. A four-channel ResNet-50 backbone fuses RGB pixels with a landmark-mask channel, regresses the 3DMM shape, expression, pose, and illumination parameters, and is refined through a multi-task loss that combines 3D parameter regression, 2D landmark consistency, and image-to-mesh-to-image cycle consistency. The model is trained on a curated 2000-image subset of the LFW-People corpus and evaluated under four yaw-angle strata. The results indicate that on a held-out 400-image test set, the pipeline attains R² = 0.854, MSE = 0.022, Pearson r = 0.92, and MAPE = 10.6%, with a single-frame inference latency of 35 ms on a commodity RTX-class GPU. Robustness to head rotation improves by 29.9% at extreme poses (60–90° yaw) compared with a single-modality baseline. A Blender-integrated prototype successfully exports the reconstructed mesh as a deformation-ready asset for Unity- and Unreal-based XR engines. The proposed pipeline offers a cost-effective, real-time-capable component for XR avatar prototyping, lowering the entry barrier for small studios, immersive-learning developers, and AR/MR telepresence research. On the standard AFLW2000-3D benchmark, the pipeline additionally attains a Normalised Mean Error of 2.47% and a full-vertex reconstruction error of 1.50%, which is competitive with published lightweight baselines while retaining sub-50 ms inference latency. Full article

► Show Figures

Figure 1

18 pages, 282 KB

Open AccessArticle

Digital Product Passports in Online Fashion Resale: Supporting Circular Fashion, Transparency, and Consumer Trust

by Anna Elizabeth Stookey and Chunmin Lang

Informatics 2026, 13(8), 121; https://doi.org/10.3390/informatics13080121 - 24 Jul 2026

Abstract

As online luxury fashion resale platforms continue to emerge and expand, persistent challenges related to trust and product authenticity have become increasingly pronounced. digital product passports (DPPs) offer a potential solution to these trust-related challenges by providing detailed information about a product, thereby [...] Read more.

As online luxury fashion resale platforms continue to emerge and expand, persistent challenges related to trust and product authenticity have become increasingly pronounced. digital product passports (DPPs) offer a potential solution to these trust-related challenges by providing detailed information about a product, thereby enhancing product transparency and traceability throughout the product lifecycle. Drawing on signaling theory, this study investigates how DPP-enabled platforms influence consumers’ purchase intentions toward luxury fashion resale consumption. An online survey generated 307 valid responses. Structural equation modeling (SEM) was employed to test the proposed hypotheses. A multi-group chi-square difference test was also conducted to test the moderating role of DPP usage experience. The results demonstrate that trust and attitude are decisive predictors of consumers’ intention to purchase second-hand luxury fashion products from DPP-enabled platforms. Risk reduction alone does not directly drive intention, highlighting the distinction between eliminating uncertainty and fostering positive motivation. This study focuses on consumer perception and intention regarding DPPs within the luxury fashion resale market, an area that has received limited empirical attention. By integrating signaling theory with fashion resale and digital innovation, this study offers novel insights into the role of technological transparency in driving consumer engagement in circular fashion. Full article

25 pages, 2321 KB

Open AccessArticle

Activity Classification in E-Commerce Product Reviews Using Deep Learning and Transformer Models

by Tinashe Wamambo, Arooj Fatima, Bethwel Kiplagat, Mahdi Maktab Dar Oghaz and Cristina Luca

Informatics 2026, 13(8), 120; https://doi.org/10.3390/informatics13080120 - 23 Jul 2026

Abstract

Existing research on e-commerce product reviews has primarily focused on analysing consumers’ opinions, emotions, sentiments and associated star ratings. Whilst these approaches provide insights into consumers’ perceptions of products, they offer limited understanding of how products are used in real-world contexts. Therefore, they [...] Read more.

Existing research on e-commerce product reviews has primarily focused on analysing consumers’ opinions, emotions, sentiments and associated star ratings. Whilst these approaches provide insights into consumers’ perceptions of products, they offer limited understanding of how products are used in real-world contexts. Therefore, they do little to enhance the e-commerce experience by helping consumers make more informed purchasing decisions based on products’ intended uses without requiring them to read numerous reviews during the decision-making process. To address this problem, this paper investigates the feasibility of automatically identifying and classifying product usage activities from e-commerce reviews. A methodology combining natural language processing, manual activity-level annotation and deep learning-based text classification was developed and evaluated. An initial dataset of 60,000 Amazon product reviews was manually labelled according to six activity classes: run, walk, hike, swim, climb and unknown. Following quality inspection and data cleaning, a final dataset of 50,843 reviews was used for model training and evaluation. Multiple classification approaches were assessed, including CNN, LSTM, hybrid LSTM-CNN architectures and transformer-based models (DistilBERT and DistilBERT-CNN). Experimental evaluation was conducted using multiple random seeds to ensure robustness and reproducibility. The results indicate that activity classification from e-commerce reviews is a challenging task due to ambiguity and overlapping usage descriptions, with all evaluated models achieving comparable performance on the full dataset. Among the evaluated models, the hybrid LSTM-CNN-GloVe architecture achieved the highest performance on the keyword-filtered dataset, whilst the DistilBERT-CNN model also demonstrated strong results. The findings demonstrate the feasibility of extracting activity-oriented information from product reviews and highlight activity classification as a distinct and under-explored natural language processing task that complements traditional sentiment analysis. The proposed methodology provides a foundation for improving product discovery and supporting usage-oriented search and recommendation systems in e-commerce environments. Full article

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

► Show Figures

Figure 1

18 pages, 8774 KB

Open AccessArticle

Role of Anthropomorphic Design in Social Robots for Aged Care: A Case Study of Pepper

by James R. Sadler, Samina Ansari, Aila Khan, Michael Lwin and Omar Mubin

Informatics 2026, 13(7), 119; https://doi.org/10.3390/informatics13070119 - 22 Jul 2026

Abstract

The increasing aging population in Australia necessitates innovative caregiving solutions to address the growing needs of elderly residents. Humanoid robots, with their physical embodiment and human-like attributes, offer a promising technological intervention. This exploratory study investigates the integration of Pepper, a humanoid robot, [...] Read more.

The increasing aging population in Australia necessitates innovative caregiving solutions to address the growing needs of elderly residents. Humanoid robots, with their physical embodiment and human-like attributes, offer a promising technological intervention. This exploratory study investigates the integration of Pepper, a humanoid robot, into aged care facilities, focusing on its potential to meet the needs of older adults and serve as a daily companion, thereby reducing staff workload. The research explores the anthropomorphic features of Pepper, their role in fostering connection and engagement, and the perception and acceptance of the robot as a companion among elderly residents. Findings highlight Pepper’s potential to enhance the quality of care and support in aged care settings while identifying areas for improvement to ensure its successful adoption. Full article

(This article belongs to the Special Issue AI-Enabled Digital Health Technologies for Patient-Centered Care and Sustainable Systems)

► Show Figures

Figure A1

38 pages, 11672 KB

Open AccessArticle

Enhancing 3D MRI-Based Necrotic Core Segmentation in Glioblastoma Using Activation Functions in Deep Learning

by Mushtaq Mahyoob Saleh, Eltahir Mohamed Hussein, Musab Elkheir Salih and Mohamed A. A. Ahmed

Informatics 2026, 13(7), 118; https://doi.org/10.3390/informatics13070118 - 20 Jul 2026

Abstract

Precise brain tumour delineation is vital for therapy protocols and tracking. However, standard Rectified Linear Units (ReLU) struggle to capture subtle necrotic-core variations due to zero-gradient behaviour in the negative domain. To address this, we present a controlled benchmark of 12 activation functions [...] Read more.

Precise brain tumour delineation is vital for therapy protocols and tracking. However, standard Rectified Linear Units (ReLU) struggle to capture subtle necrotic-core variations due to zero-gradient behaviour in the negative domain. To address this, we present a controlled benchmark of 12 activation functions within a fixed Residual 3D U-Net using the Brain Tumour Segmentation (BraTS) 2020 dataset. In the single-run benchmark, Swish achieved the best necrotic-core (NCR) Dice (0.676; +2.0% over ReLU, p < 0.01), while TanhExp attained the highest whole-tumour accuracy (0.879). To test the reliability of these single-run results, the four functions central to our claims were retrained across three random seeds. This analysis confirmed a small but consistent NCR advantage for the smooth and adaptive functions—Swish (0.677 ± 0.003) and PReLU (0.678 ± 0.004) over ReLU (0.661 ± 0.012; pooled p < 0.001)—with Swish among the most stable functions in this region. By contrast, the apparent single-run differences in the enhancing tumour, and the underperformance of PReLU, did not generalise across seeds, indicating that activation-function effects in this task are concentrated in the necrotic core and that single-seed comparisons can be misleading. Crucially, Swish achieved these gains with zero additional trainable parameters and only a ~1% training latency penalty on common hardware. Replacing ReLU with Swish offers a cost-effective, architecture-preserving strategy to improve segmentation reliability and boundary delineation. Ultimately, this zero-cost architectural modification is a promising, preliminary step towards more reliable automated tumour delineation, pending prospective validation on multi-institutional data and expert radiological assessment. Full article

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

► Show Figures

Figure 1

20 pages, 552 KB

Open AccessArticle

Evaluating the Performance of Large Language Models in Evidence-Scarce Scenario: The Diabetic Foot Ulcer Transition Phase

by Kamran Shakir, Hadi Sarlak, Giulia Rogati, Alberto Leardini, Lisa Berti and Paolo Caravaggi

Informatics 2026, 13(7), 117; https://doi.org/10.3390/informatics13070117 - 20 Jul 2026

Abstract

Diabetic foot ulcers (DFUs) impose a substantial burden on people with diabetes and healthcare systems. The post-healing “transition phase” remains clinically challenging with limited guideline support. While large language models (LLMs) are increasingly proposed as clinical decision-support tools, their reliability in evidence-scarce scenarios [...] Read more.

Diabetic foot ulcers (DFUs) impose a substantial burden on people with diabetes and healthcare systems. The post-healing “transition phase” remains clinically challenging with limited guideline support. While large language models (LLMs) are increasingly proposed as clinical decision-support tools, their reliability in evidence-scarce scenarios is largely untested. This exploratory study benchmarked leading LLMs against European clinician consensus for the evidence-scarce scenario of DFU transition-phase management. Six LLMs (ChatGPT-4o, ChatGPT-5.0, Gemini 2.5 Flash, Gemini 2.5 Pro, Claude Sonnet 4.0, and Perplexity) were evaluated for accuracy and hallucination using a two-stage framework. Benchmarks were derived from an online survey of European DFU experts reflecting European-level and national practices (Denmark, Netherlands, UK). Binary outcomes were summarized as proportions with Wilson 95% confidence intervals. Paired within-item comparisons across models were assessed using Cochran’s Q, followed by exact McNemar tests with Holm correction (α = 0.05); inferential results were considered supportive due to the limited number of paired items (n = 8). Across 192 accuracy assessments and 384 reference checks, LLM accuracy ranged from 50–75% and declined when simulating national practices. Hallucination rates exceeded 50% in several models. LLMs rely on generic recommendations, which contrast with clinicians’ contextual, patient-centered reasoning, suggesting current limitations in their suitability for clinical decision support in DFU transition phase clinical care. Full article

(This article belongs to the Section Generative AI)

► Show Figures

Graphical abstract

20 pages, 764 KB

Open AccessArticle

Procedural and Distributive Unfairness in AI Interactions: Are People Less Satisfied with Unfairness from AI Compared to Humans?

by Devon Johnson, Sungyong Chun, Sneha Pandey and Gouher Ahmed

Informatics 2026, 13(7), 116; https://doi.org/10.3390/informatics13070116 - 20 Jul 2026

Abstract

This study investigates consumer perceptions of procedural and distributive fairness/unfairness in service interactions involving AI versus human providers. It also examines how these perceptions are influenced by technology-related discomfort and the perceived responsibility of the service organization. A scenario-based online experiment was conducted [...] Read more.

This study investigates consumer perceptions of procedural and distributive fairness/unfairness in service interactions involving AI versus human providers. It also examines how these perceptions are influenced by technology-related discomfort and the perceived responsibility of the service organization. A scenario-based online experiment was conducted using participants recruited through the Prolific research platform. Participants were asked to review a small business loan application process at a fictional digital bank where the application is processed either by an AI algorithm or a human loan officer. The study found that consumers experienced a higher level of satisfaction when they encountered procedural unfairness from humans compared to AI. The negative effect of procedural unfairness on satisfaction was amplified as algorithm discomfort and perceived company responsibility increased. The findings suggest that AI unfairness requires special attention to address differences in consumer reaction. Full article

(This article belongs to the Special Issue Explainable and Trustworthy Artificial Intelligence for Data-Driven Decision Support)

► Show Figures

Figure 1

11 pages, 242 KB

Open AccessArticle

Reliability and Quality of AI-Generated Information on Newborn Screening Tests: A Comparative Analysis of ChatGPT and Gemini

by Ece Nilüfer, Meserret Aslan and Şehma Şen

Informatics 2026, 13(7), 115; https://doi.org/10.3390/informatics13070115 - 17 Jul 2026

Abstract

Background/Objectives: The increasing use of artificial intelligence (AI) chatbots for obtaining health-related information has raised concerns regarding the quality and reliability of the information they provide. This study aimed to compare the quality and reliability of responses generated by ChatGPT free tier (GPT-4o, [...] Read more.

Background/Objectives: The increasing use of artificial intelligence (AI) chatbots for obtaining health-related information has raised concerns regarding the quality and reliability of the information they provide. This study aimed to compare the quality and reliability of responses generated by ChatGPT free tier (GPT-4o, with GPT-4.1 mini as the fallback model after the usage limit) and Gemini 2.5 Flash (Google, free version) regarding newborn screening tests. Methods: A total of 31 questions were developed based on international and national newborn screening guidelines and were posed to both chatbots. Responses were independently evaluated by two researchers using the DISCERN instrument and the Global Quality Score (GQS), and inter-rater reliability was assessed. Descriptive statistics and non-parametric tests were used to compare chatbot performance. Results: Both evaluators assigned significantly higher DISCERN total scores to Gemini than to ChatGPT free tier. For Evaluator 1, the mean DISCERN scores were 48.8 ± 10.2 for Gemini and 42.4 ± 5.9 for ChatGPT free tier (p < 0.001); for Evaluator 2, the corresponding scores were 53.7 ± 8.9 and 42.4 ± 5.8, respectively (p < 0.001). For the GQS ratings, Evaluator 1 rated Gemini significantly higher than ChatGPT free tier (3.6 ± 0.6 vs. 3.0 ± 0.5, p < 0.001), whereas Evaluator 2 found no statistically significant difference between the two chatbots (3.3 ± 1.3 vs. 3.4 ± 0.8, p = 0.877). Inter-rater reliability for DISCERN scores was excellent for ChatGPT and moderate for Gemini, whereas agreement for GQS ratings was low for both chatbots. Conclusions: Gemini demonstrated consistently higher DISCERN scores than ChatGPT free tier; however, its superiority in overall GQS ratings was not consistently supported across the two evaluators. Neither chatbot consistently achieved the highest levels of information quality. AI chatbots should therefore be considered supplementary sources of health information rather than substitutes for healthcare professionals or official health information resources. Full article

30 pages, 1081 KB

Open AccessArticle

Event-Conditioned Causal Extraction in Saudi Dialect: A Comparative Study of Dialect-Trained BERTs and LLM Prompting

by Mariam Elhussein, Samiha Brahimi, Reem Osman and Suhier Elfaki

Informatics 2026, 13(7), 114; https://doi.org/10.3390/informatics13070114 - 17 Jul 2026

Abstract

Causality extraction is an important task in natural language processing, yet it remains underexplored in informal Arabic social media text, particularly in dialectal contexts. This study investigates causal-reason extraction from Saudi Arabic tweets related to sick-leave requests. A gold-standard dataset was annotated for [...] Read more.

Causality extraction is an important task in natural language processing, yet it remains underexplored in informal Arabic social media text, particularly in dialectal contexts. This study investigates causal-reason extraction from Saudi Arabic tweets related to sick-leave requests. A gold-standard dataset was annotated for multiple causality-related tasks, including cause-presence detection, cause-span extraction, cause-category classification, causal-marker detection, and causal marker text identification. The study compares two modeling paradigms: fine-tuned BERT-based models, represented by SaudiBERT and AraBERT, and prompting-based large language models (LLMs), represented by GPT-4.1-mini and Gemini-2.5-flash. The descriptive analysis showed strong class imbalance, substantial implicit causality, and uneven cause-category distributions. Results showed that SaudiBERT generally outperformed AraBERT when macro-level and minority-class performance were considered. Among LLMs, Gemini-2.5-flash achieved the strongest overall performance, particularly under natural 10-shot single-tweet prompting, while balanced few-shot prompting improved macro-F1 for cause-category classification. However, step-wise prompting did not consistently improve performance and may have introduced error propagation. Overall, the findings show that causality extraction in informal Saudi Arabic remains challenging, especially for implicit causal expression. The study highlights the complementary strengths of dialect-specific transformers and LLM-based prompting for Arabic causality extraction. Full article

(This article belongs to the Special Issue Machine Learning in Social Media Analysis)

► Show Figures

Figure 1

5 pages, 162 KB

Open AccessEditorial

AI and Academic Publishing

by Antony Bryant

Informatics 2026, 13(7), 113; https://doi.org/10.3390/informatics13070113 - 14 Jul 2026

Abstract

MDPI, along with all other academic publishers, recognizes the potential benefits and real challenges posed by the advent of Generative AI [GenAI] [...] Full article

(This article belongs to the Section Generative AI)

17 pages, 773 KB

Open AccessArticle

OpenGluco—Innovative Open-Source Diabetes Care Management System

by Michal Kubascik, Andrej Tupy, Lukas Formanek and Miroslav Chochul

Informatics 2026, 13(7), 112; https://doi.org/10.3390/informatics13070112 - 14 Jul 2026

Abstract

Continuous glucose monitoring (CGM) plays a central role in modern diabetes management, yet CGM data are often confined within proprietary manufacturer ecosystems, limiting interoperability and reuse. This paper presents OpenGluco, an open-source and provider-independent platform designed to unify CGM data from multiple commercial [...] Read more.

Continuous glucose monitoring (CGM) plays a central role in modern diabetes management, yet CGM data are often confined within proprietary manufacturer ecosystems, limiting interoperability and reuse. This paper presents OpenGluco, an open-source and provider-independent platform designed to unify CGM data from multiple commercial systems, including Abbott Freestyle Libre, Dexcom, and Medtronic. OpenGluco implements a modular server architecture that abstracts vendor-specific data access, normalizes heterogeneous glucose time series, and exposes standardized access through an application programming interface designed to support future scalability. The platform supports individual users as well as institutional and research deployments. Performance evaluation demonstrates reliable data ingestion, consistent API responsiveness within the evaluated deployment environment, and robust handling of heterogeneous sampling characteristics. Validation of data normalization shows preservation of clinically relevant metrics such as Time in Range. By emphasizing interoperability, user-authorized data access, and open-source transparency, OpenGluco provides a flexible foundation for clinical monitoring, education, and future analytics-driven decision-support applications in diabetes care. Full article

(This article belongs to the Special Issue Health Data Management in the Age of AI)

► Show Figures

Figure 1

38 pages, 6648 KB

Open AccessArticle

A Data-Driven Informatics Framework for Evaluating Thai Provinces Using an Additive Weighting-Based Variant Assessment Algorithm and Two-Stage DEA

by Pasura Aungkulanon, Roberto Montemanni and Pongchanun Luangpaiboon

Informatics 2026, 13(7), 111; https://doi.org/10.3390/informatics13070111 - 10 Jul 2026

Abstract

In order to evaluate regional sustainability, a comprehensive framework is needed that can integrate a number of economic and environmental variables into a transparent and policy-relevant evaluation approach. The present study presents a data-driven informatics framework for the evaluation of Thai provinces that [...] Read more.

In order to evaluate regional sustainability, a comprehensive framework is needed that can integrate a number of economic and environmental variables into a transparent and policy-relevant evaluation approach. The present study presents a data-driven informatics framework for the evaluation of Thai provinces that utilizes the additive weighting-based variant assessment algorithm (AWVAA) with Charnes–Cooper–Rhode (CCR)-based two-stage data envelopment analysis (DEA). The system allows three interrelated activities: provincial screening, representative decision-making unit selection, and comparative efficiency benchmarking of economic and environmental performance. AWVAA employs global and local simple additive weighting algorithms in screening 77 provinces to find representative units while keeping regional balance and data completeness. In the second phase, the selected provinces are evaluated by a two-stage DEA structure based on CCR to measure their relative efficiency for transforming development-related inputs into intermediate operational factors and ultimate economic and environmental outputs. The analysis starts with investment, tourist arrivals, and newborns as initial inputs, moves through energy use, electricity consumption, number of factories, and number of vehicles as intermediate variables, and ends with gross provincial product and air quality indicators, including ozone, PM10, and PM2.5 as final outputs. The proposed framework selects 16 typical provinces and shows significant variations in overall CCR efficiency and super-efficiency performance over the selected set. The results suggest that provinces with high screening-stage prominence may not necessarily become the strongest DEA-based standards and emphasize the complimentary roles of representative unit selection and formal efficiency assessment. The study combines multi-criteria screening with benchmarking based on DEA to give a transparent and replicable method for regional sustainability monitoring, comparative assessment, and evidence-based policy planning. The results provide an informatics-oriented paradigm for complicated regional evaluation and practical insights for enhancing sustainable provincial development in Thailand. Full article

► Show Figures

Figure 1

24 pages, 799 KB

Open AccessArticle

LAPM-RA: Reward-Adaptive Prompt Learning with LLM Augmentation for Multimodal Sentiment Analysis

by Zhi Zhu, Cheng Kuang and Yin Qian

Informatics 2026, 13(7), 110; https://doi.org/10.3390/informatics13070110 - 10 Jul 2026

Abstract

Few-shot multimodal sentiment analysis (MSA), constrained by limited annotated data, often suffers from large cross-modal semantic alignment gaps and difficulty in capturing fine-grained sentiment cues, making it challenging to fully exploit the complementary information between text and images. Recent advances in large language [...] Read more.

Few-shot multimodal sentiment analysis (MSA), constrained by limited annotated data, often suffers from large cross-modal semantic alignment gaps and difficulty in capturing fine-grained sentiment cues, making it challenging to fully exploit the complementary information between text and images. Recent advances in large language models (LLMs) and prompt learning have shown strong potential for improving label efficiency in low-resource natural language processing tasks; however, their direct application to MSA is hindered by static prompt designs and shallow cross-modal integration. To overcome these limitations, we propose LLM-Augmented Prompt Learning for Multimodal Sentiment Analysis with Reward Adaptation (LAPM-RA), a unified framework integrating LLM-based sentiment-aware augmentation, reward-guided prompt selection, and context-aware multimodal fusion. Specifically, LLMs generates sentiment-consistent and counterfactual text variants to enhance lexical and structural diversity while preserving label fidelity; a supervised policy network adaptively selects optimal prompt templates based on reward signals; and a lightweight gating mechanism integrates textual and visual embeddings contextually. Extensive experiments on multiple benchmarks validate the effectiveness and robustness of LAPM-RA over competitive baselines. Full article

► Show Figures

Figure 1

28 pages, 949 KB

Open AccessArticle

From Data to Behaviour: Understanding the Perceived Smartwatch Value for Physical Activity Through Self-Quantification

by Carmen Bekker, Brenda M. Scholtz and Simone Beets

Informatics 2026, 13(7), 109; https://doi.org/10.3390/informatics13070109 - 9 Jul 2026

Abstract

Smartwatches are commonly used IoT wearable devices, which contain sensors that can support the data management process of smartwatch data. This process is essential for providing data-based insights that can motivate users to achieve their physical activity (PA) and health goals. However, there [...] Read more.

Smartwatches are commonly used IoT wearable devices, which contain sensors that can support the data management process of smartwatch data. This process is essential for providing data-based insights that can motivate users to achieve their physical activity (PA) and health goals. However, there is a lack of understanding of users’ perceptions of the health-related value they derive from smartwatches and of user self-tracking behaviour within a PA context. This self-tracking behaviour is also known as Self-Quantification Behaviour (SQB), which is the process of data preparation, collection, integration, reflection, and action. SQB could be affected by the functions provided by the smartwatch device, which vary by manufacturer, brand and model. The paper aims to investigate user perceptions of SQB and its association with their perceived value of smartwatches. A survey was conducted with professional office workers from South Africa who use smartwatches. The findings highlighted that the respondents enjoyed collecting and engaging with their PA data and that SQB significantly affected all four value constructs: Perceived Usefulness, Pleasure, Goal Pursuit Motivation, and Social Value. Social value, in terms of data sharing and user lifestyle, was the least well-rated. The findings contribute to an enhanced understanding of smartwatches and users’ perceived value of self-quantifying behaviour in the context of PA, thereby ultimately improving the smartwatch usage experience and motivating users to exercise more. Full article

(This article belongs to the Special Issue Health Data Management in the Age of AI)

► Show Figures

Figure 1

38 pages, 5157 KB

Open AccessArticle

A Multilabel Embedding-Based Framework for Predicting Short-Term Cross-Category Complaint Recurrence

by Theng-Jia Law, Choo-Yee Ting, Hu Ng and Hui-Ngo Goh

Informatics 2026, 13(7), 108; https://doi.org/10.3390/informatics13070108 - 8 Jul 2026

Abstract

Public complaints exhibit strong spatiotemporal dependencies, where issues often propagate across categories within short timeframes, yet existing studies largely overlook cross-category recurrence and underutilize embedding representations for structured data. To address this gap, this study proposes a multilabel embedding-based framework to predict short-term [...] Read more.

Public complaints exhibit strong spatiotemporal dependencies, where issues often propagate across categories within short timeframes, yet existing studies largely overlook cross-category recurrence and underutilize embedding representations for structured data. To address this gap, this study proposes a multilabel embedding-based framework to predict short-term cross-category complaint recurrence using structured spatiotemporal data. Using 48,103 real-world complaint records, the framework integrates embedding representations with Machine Learning (ML) and Deep Learning (DL) models to predict the likelihood of multiple complaint categories recurring within the next seven days in the same area. The results indicated that both approaches achieved average label-wise F1-scores of 30.5–32.6%, exceeding 81% for highly recurrent categories. The best ML model, Binary Relevance with Logistic Regression using multilingual-e5-large embeddings, achieved the lowest Hamming loss of 0.138 ± 0.126. Statistical analysis confirmed non-normality with Shapiro–Wilk statistics between 0.796 and 0.819 and p-values below 0.05, and significant differences across models with a Friedman test statistic of 512.531 and p-values below 0.05, although no significant pairwise differences were found with the Nemenyi post hoc test. Furthermore, 95% bootstrap confidence intervals indicate stable performance, with F1 ranging from 74.1% to 74.9% for ML and 73.2% to 74% for DL. Full article

► Show Figures

Figure 1

Journal Description

Informatics

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Further Information

Guidelines

MDPI Initiatives

Follow MDPI