<?xml version="1.0" encoding="utf-8"?>
    <rss version="2.0">
      <channel xmlns:content="http://purl.org/rss/1.0/modules/content/">
        <title>Frontiers in Artificial Intelligence | New and Recent Articles</title>
        <link>https://www.frontiersin.org/journals/artificial-intelligence</link>
        <description>RSS Feed for Frontiers in Artificial Intelligence | New and Recent Articles</description>
        <language>en-us</language>
        <generator>Frontiers Feed Generator,version:1</generator>
        <pubDate>2026-09-26T20:32:04.975+00:00</pubDate>
        <ttl>60</ttl>
        <item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1884918</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1884918</link>
        <title><![CDATA[Whole-slide deep-learning quantification of fibrosis and pathologist-defined proliferative/neoplastic regions in a thioacetamide-induced rat liver model]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Feng Liu</author><author>Yating Pan</author><author>Xinyi Liao</author><author>Yating Deng</author><author>Deqiang Cheng</author><author>Guanzhen Yu</author><author>Ying Chen</author><author>Yu Cheng</author>
        <description><![CDATA[Automated analysis of whole-slide images may support objective quantification of selected lesions in experimental liver pathology. Using H&E-stained whole-slide images from rats exposed to thioacetamide (TAA), we developed a deep-learning framework for whole-slide segmentation and quantification of two prespecified targets: fibrosis and a single pooled class of pathologist-defined proliferative/neoplastic regions. Benign proliferation, dysplasia/preneoplasia, and malignant neoplasia were not annotated or trained as separate subclasses. The available project archive did not contain an auditable whole-slide- or animal-level training/validation manifest; the available training code implements a patch-level random split. The framework generated fibrosis proportion, proliferative/neoplastic-region proportion, and the count of spatially separated positive regions for each whole-slide image. These model-derived measurements were used to describe changes across experimental time points and to explore differences between the TAA and TAA + H2 groups. The TAA + H2 group showed a lower model-derived burden of the two selected lesion categories. The framework does not assess the full spectrum of TAA-associated liver pathology and should not be interpreted as an independently validated diagnostic classifier. Independent animal-level validation is required before diagnostic or therapeutic claims can be made.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1874928</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1874928</link>
        <title><![CDATA[Higher-dimensional embedding of time-series data for machine learning]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Karan Singh</author><author>Pingal Pratyush Nath</author><author>Urbasi Sinha</author><author>Aninda Sinha</author>
        <description><![CDATA[IntroductionDeep learning has revolutionized image analysis, yet most clinical biosignals, especially multi-lead electrocardiograms (ECGs), remain one-dimensional and awkward for modern vision models.MethodsWe introduce an orthogonal-polynomial imaging (OPI) framework that encodes 12-lead ECGs into a single two-dimensional image through an invertible transformation prior to coarse-graining, producing compact image representations that retain sufficient information for accurate classification. We benchmark the proposed representation against established architectures including ResNet, Transformers, and 1D temporal convolutional models on large-scale PhysioNet ECG datasets.ResultsOur results demonstrate that OPI-based image representations achieve highly competitive classification performance while preserving diagnostically relevant temporal and cross-lead information.DiscussionThe framework bridges raw multi-channel biosignals and standard image-based learning architectures without representation loss prior to coarse-graining, offering an effective, scalable, and physics-inspired representation for automated cardiac diagnostics.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1945529</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1945529</link>
        <title><![CDATA[Vision–language guided semantic-geometric transformer for memory-efficient 3D scene understanding]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Licheng Liu</author><author>Yu Li</author><author>Fuyong Liu</author>
        <description><![CDATA[Recent 3D Transformers have become a dominant framework for point-cloud segmentation by modeling spatial context in sparse 3D scenes. However, geometry and color alone provide limited high-level semantic cues, especially for cluttered boundary regions, visually similar objects, and long-tail categories. To address this issue, we propose a segmentation framework guided by Contrastive Language–Image Pre-training (CLIP) that enriches sparse 3D tokens with vision–language semantic priors. Specifically, dense CLIP features extracted from multi-view RGB images are projected onto 3D points through visibility-aware alignment and view pooling, and are fused with relative geometric offsets and color cues to form semantically aware sparse voxel tokens. To better exploit the aligned CLIP semantics during local token interactions, we build on contextual relative signal encoding (cRSE) and introduce a decoupled CLIP-induced semantic residual that forms semantic-geometric attention biases for local window attention. We further adapt block-wise online softmax computation to generate and consume these biases on the fly. Experiments on ScanNet, ScanNet200, and S3DIS demonstrate competitive segmentation performance, improved instance-level discrimination, and a 25.7% reduction in peak online 3D-stage training memory compared with the materialized attention implementation when cached CLIP features are used.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1943994</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1943994</link>
        <title><![CDATA[A comprehensive study on cervical cancer diagnosis using deep learning and artificial intelligence techniques]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Review</category>
        <author>Nilasree Kannadoss</author><author>Kumar Rangasamy</author>
        <description><![CDATA[Cervical cancer is a significant health problem across the globe, especially in nations with low or middle incomes, where access to early screening is limited and contributes to high mortality rates. Traditional diagnostic methods, including Papanicolaou smears (Pap smears) and Human Papillomavirus testing (HPV testing), often suffer from accessibility constraints, lower accuracy, and scalability issues. Recent advancements in artificial intelligence (AI) and Machine Learning (ML) have exhibited significant efficacy by improving detection accuracy and streamlining the diagnostic procedure. This review discusses AI-driven methodologies, including deep learning models (EfficientNetB0, Cervical Net), hybrid frameworks, and ensemble techniques. This study examines important obstacles related to class imbalance, data privacy, and model interpretability while also presenting a solution known as the Adaptive Synthetic Minority Over-Sampling Technique combined with Tomek Links (Adaptive SMOTE Tomek) for data augmentation and federated learning (FL) for secure, distributed AI model training. The outcomes emphasize AI’s vital role in improving cervical cancer screening and correspond with the World Health Organization’s (WHO) 2030 goals for cervical cancer eradication. Future research should focus on real-world validation, improved generalization, and the integration of AI in clinical settings to ensure widespread usage and reliability. This work is useful for medical professionals in cervical cancer diagnosis, in addition to traditional screening methods.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1801583</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1801583</link>
        <title><![CDATA[Drivers and barriers of AI tool adoption in Jordanian SMEs: a qualitative TOE-based study]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Hamad Alsawalqah</author><author>Yazn Alshamaila</author><author>Ons Al-Shamaileh</author><author>Ibrahim Aljarah</author><author>Ahmad Abadleh</author><author>Nailah Al-Madi</author>
        <description><![CDATA[Artificial intelligence tools are increasingly used in business. However, small and medium-sized enterprises (SMEs) continue to face challenges in deciding whether to adopt them and how to. This study examines the factors that encourage or hinder the adoption of AI tools by SMEs in the Jordanian context. The study uses the technology–organisation–environment framework (TOE) and draws on 15 semi-structured interviews with AI solution providers, AI adopters, and firms considering adoption. The data was analysed thematically using TOE-guided deductive coding alongside inductive coding. The findings show that technological and organisational factors, including relative advantage, compatibility, managerial support, financial resources, and prior experience, influence AI adoption decisions. Environmental factors, such as competitive pressure, regulation, industry conditions, and public awareness, also affect adoption, but their role differs across firm categories. AI solution providers mainly emphasised expertise, market scope, and client readiness. Adopters focused more on organisational readiness and competitive pressure, while firms considering adoption reported uncertainty, cost, and employee acceptance as prominent barriers. Resistance to change and public awareness also appeared as context-sensitive factors within the organisational and environmental dimensions of TOE. The study provides qualitative evidence from Jordanian SMEs and offers practical implications for SMEs, AI solution providers, and support bodies in Jordan and similar regional contexts. The findings should be read as exploratory and context-specific, not as broad evidence for all SMEs in developing countries.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1874897</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1874897</link>
        <title><![CDATA[Views: a hardware-aware recursively labeled graph database model for knowledge representation and reasoning]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yanjun Yang</author><author>Adrian Wheeldon</author><author>Yihan Pan</author><author>Alex Serb</author>
        <description><![CDATA[IntroductionKnowledge representation remains a central challenge for reasoning-centered artificial intelligence, particularly when semantic structures involve relations over relations, contextual annotations, and recursively nested descriptions. This paper introduces Views, a recursively labeled graph database (GDB) model designed to represent such graph-structured knowledge within a uniform graph abstraction while retaining a hardware-aware organization for associative search and traversal.MethodsThe model refactors directed labeled graphs into linked-list-like chains of linknodes, supports recursive labeling of vertices and edges, and admits mappings from Resource Description Framework (RDF)- and labeled property graph (LPG)-style graph representations. We describe the data structure, its hardware-oriented memory mappings under the Associative Chip Architecture (ASOCA), and selected associative operations for retrieval over Views-based GDBs. We then evaluate storage footprint under stated mapping and store boundaries using three scales of the Social Network Benchmark published by the Linked Data Benchmark Council, and characterize directed K-hop traversal on an Associative Memory Chip III (ASOCA3) field-programmable gate array (FPGA) implementation.ResultsThe storage results show that allocation and entry width materially affect the reported footprints rather than establishing an intrinsic advantage for Views, while the K-hop measurements show increasing mean returned-vertex count and mean latency with hop bound over the fixed resident graph image. Worked semantic-reasoning and Copycat-inspired examples further illustrate how retrieval operations can be composed over the model, rather than providing application-level reasoning or cognitive validation.DiscussionTaken together, these results position Views as a model-architecture co-design for associative storage and selected graph traversal under the stated conditions; broader database workloads and end-to-end reasoning applications remain to be evaluated.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1910483</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1910483</link>
        <title><![CDATA[Enhancing ZTMAF through federated trust learning, post quantum cryptography integration and seamless cross domain authentication handover]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Jeevan Yoganand</author><author>Riya Bansal V.</author><author>Dishal L. S.</author><author>Jeipratha P. N.</author>
        <description><![CDATA[Vehicular Fog Computing (VFC) has a major role to play in facilitating low-latency and real-time communication in Intelligent Transportation Systems (ITS). Nevertheless, there exist major issues in trust management, security, and cross-domain authentication, considering the dynamic nature of vehicular networks. The Zero-Trust Mobility-Aware Authentication Framework (ZTMAF) has been proposed to overcome issues in trust management, security, and cross-domain authentication. Nevertheless, there exist major issues in distributed intelligence, trust transmission, and handover, which have been addressed in this proposed framework. The proposed framework extends ZTMAF with three major advancements: Federated Trust Learning, PQC-inspired Secure Transmission, and Cross-Domain Authentication Handover. The proposed framework uses fog nodes to train a trust model in a distributed setting, allowing for scalability and privacy preservation. Additionally, Post-Quantum Cryptography inspired encryption would be utilized for secure trust transmission, as well as to remain compliant with future post-quantum encryption standards. A token-based handover mechanism will allow vehicles to make use of previously verified trust, thus lowering latency in cross-domain authentication. The proposed framework has been validated via multiple simulation scenarios, and its performance has been evaluated against different types of attacks (e.g., stealth, on/off, Sybil). The results of the evaluations show that trust stability has improved, that detection of behavior-based attacks has improved, and that the use of the token-based handover mechanism has reduced the authentication delay by 45%. The proposed framework represents a secure and scalable solution for next-generation vehicular networks, and it sheds light on the issues associated with detecting Sybil attacks as part of the process.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1936035</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1936035</link>
        <title><![CDATA[ANN-driven analysis of Cattaneo–Christov heat transfer and entropy generation in Maxwell hybrid nanofluid flow over a vertical cone]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Bhuvana M</author><author>Durgaprasad P</author>
        <description><![CDATA[IntroductionThis research investigates Cattaneo-Christov non-Fourier heat transfer and entropy generation in Maxwell hybrid nanofluid flow over a non-Darcy porous conical surface using numerical, sensitivity, and Artificial neural networks (ANN) approaches.MethodsA coupled momentum-energy model that includes fluid relaxation, magnetic field effects, nanoparticle loading, porous resistance, and thermal relaxation is transformed using similarity variables and solved by a boundary-value numerical method. Sobol' global sensitivity analysis is employed to quantify the direct and interaction effects of the governing parameters on skin friction, Nusselt number, and entropy generation.ResultsThe analysis identifies the porous-medium parameter as the dominant contributor to skin friction and heat transfer, while its contribution to entropy generation remains substantial at both buoyancy conditions. The total-order indices further reveal increased parameter interactions in entropy generation at higher buoyancy. The artificial neural network, trained using the numerically generated dataset, accurately predicts skin friction, Nusselt number, and entropy generation, with test R2 values of 0.99479, 0.99642, and 0.99033, respectively, along with low prediction errors.DiscussionThe combined numerical-Sobol' sensitivity-ANN framework enables rapid evaluation of nonlinear viscoelastic heat-transfer systems. The results underscore the utility of Maxwell hybrid nanofluids in thermally non-Fourier porous media from an engineering perspective and provide a computational framework for evaluating their thermal performance.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1922367</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1922367</link>
        <title><![CDATA[Photonic AI for deeper understanding of the Planck epoch]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Aleksandr Raikov</author>
        <description><![CDATA[This article addresses the requirements for a photonic artificial intelligence (AI) capable of supporting a deeper understanding of the Planck epoch in the origin of the Universe. The article continues the previous author’s articles, which outlined requirements for AI to support studies of protein evolution on Earth, cosmic strings (CS) detection, and a non-standard view on the photon’s nature. Modern digital generative AI cannot support its solution due to a lack of training datasets and limited energy resources for conducting adequate physical experiments. This article analyses the states of the Planck epoch, when there was no space–time, and when entropy began to emerge. It outlines additional requirements for an AI and proposes an adequate full-analogue photonic AI (PAI) that can be helpful for the research. The main idea of the PAI is to reject digitising the natural analogue signal and to use single-step Fourier convolution on many images simultaneously for representing states of the Planck epoch. To create PAI, a single holographic plate must replace a multilevel digital neural network, thereby eliminating the need for traditional multi-step machine learning. PAI is also expected to provide effective solutions to complex tasks such as Markov chains, genetic algorithms, and Kolmogorov (Fokker-Planck) equations, which are necessary for this research. The author’s convergent methodology helps make the research focused and stable. This methodology is based on fundamental thermodynamics and the inverse-problem-solving method in topological spaces. The PAI project has a starting technology readiness level: methodology, principles, and design.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1894859</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1894859</link>
        <title><![CDATA[Mind versus machine: can biological models of the mind inspire machine intelligence design?]]></title>
        <pubdate>2026-09-25T00:00:00Z</pubdate>
        <category>Hypothesis and Theory</category>
        <author>Michael Sharwood Smith</author>
        <description><![CDATA[Models of human cognition that look biologically plausible might in principle be a source of insights that could prove applicable to machine intelligence. The Modular Cognition Framework fits that description in terms of its biological credentials and its potential to identify the challenges confronting GenAI developers. However, it is an account of cognition in the mind and not in the brain. Accordingly, the account includes a clear delineation of the two intimately related concepts, mind and brain, and a definition of their relationship. If either brain or mind can contribute to this debate, a model of the mind could offer a more immediate promising source of inspiration than its complex neural underpinnings. Key features of the theoretical framework include its fundamentally heterarchical architecture having no single locus of control. In addition, general and domain-specific processing principles permit rapid, highly flexible, adaptable interaction between its network of modular systems as they collaborate in the execution of multiple tasks in a constantly changing environment. Eleven types of knowledge are constructed from experience using a similar number of inherited toolkits. Questions are raised as to whether the adaptable and highly creative character of evolved intelligence is reproducible in artificial systems, the best strategies for deriving benefit from studying biological systems and whether neural organisation is really the only place to look for biological insights.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1946060</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1946060</link>
        <title><![CDATA[Research on liquidity and valuation risk assessment of art financial assets based on deep learning]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ziwen Jiang</author><author>Jingzhou Zhao</author><author>Jiamin Yao</author><author>Zehao Bian</author>
        <description><![CDATA[Art is increasingly recognized as an alternative asset class in wealth management. However, its non-standardized nature renders it difficult to treat on a par with conventional financial assets. Valuation risk arises from subjective pricing and expert bias, while liquidity risk is manifested in high auction failure rates. These challenges are difficult to address adequately with traditional linear econometric models. This study proposes a multimodal deep learning framework that integrates painting images, historical transaction data, artist reputation, and macroeconomic indicators. A ResNet-50 network, initialized on ImageNet and subsequently adapted to the artwork domain, serves as the visual encoder, while a multi-layer perceptron processes structured tabular variables; the two representations are combined in a fusion layer. The framework simultaneously performs two risk-assessment tasks: valuation deviation regression and unsold-lot binary classification. The model is estimated on 96,314 modern and contemporary auction lots offered between 2000 and 2023, of which 27.8% went unsold, using a strictly chronological training, validation, and test partition. The empirical results demonstrate that the proposed multimodal framework outperforms traditional benchmarks, including ordinary least squares (OLS), logistic regression, and random forests. The model achieves an Area Under the Curve (AUC) of 0.891 for liquidity risk prediction. It reduces the Root Mean Squared Error (RMSE) for valuation deviation by 32.9% compared to OLS. Ablation studies confirm that visual embeddings provide substantial incremental information beyond what can be attributed to increased model complexity alone, and permutation importance combined with linear probing shows that the visual representation encodes stylistic period, subject type, chromatic intensity, and compositional complexity—the properties through which visual content is theorized to affect market outcomes. Robustness checks further delineate the model's performance boundaries under macroeconomic shocks. This study translates the aesthetic heterogeneity of non-standard assets into quantifiable risk signals. It offers financial institutions a data-calibrated basis for adjusting dynamic loan-to-value ratios and assists art funds in optimizing portfolio allocation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1940124</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1940124</link>
        <title><![CDATA[An empirical study of augmented analytics adoption and its impact on firms’ agility through AI-based models]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Moath Srahin</author><author>Ala Mughaid</author><author>Mahmoud AlJamal</author><author>Abedalmuhdi Almomany</author><author>Muhammed Sutcu</author>
        <description><![CDATA[Augmented analytics extends business analytics by combining artificial intelligence, machine learning, and natural language processing. However, organizations differ in their ability to adopt these technologies and derive value from them. This study examines the organizational conditions associated with augmented analytics adoption and the role of firm agility in linking adoption to competitive advantage. The proposed model draws on Task-Technology Fit, the Resource-Based View, and Dynamic Capability Theory. It includes six resource-technology fit dimensions as antecedents of augmented analytics adoption and tests firm agility as a mediator. Survey data from a final analytic sample of 287 respondents across several business sectors in Jordan were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The findings indicate that the fit dimensions do not contribute equally to adoption, with data, individual, cultural, and analytics-capability fit showing the clearest relationships. The results also show that the relationship between augmented analytics adoption and competitive advantage operates mainly through firm agility. The study therefore suggests that the value of augmented analytics depends not only on adopting the technology but also on aligning organizational resources and using analytical insights to support timely organizational responses.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1896192</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1896192</link>
        <title><![CDATA[Leakage-aware pre-event machine learning evaluation for horse-race prediction under temporal validation]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Brief Research Report</category>
        <author>Shuichi Sugiura</author>
        <description><![CDATA[Horse-racing prediction was used as a case study to develop and evaluate a leakage-aware machine-learning framework based on information available after race-entry finalization and before outcomes were known. Japanese Racing Association flat-race data obtained through JRA-VAN Data Lab. and exported using TARGET frontier JV were split temporally: 2015–2022 for training, 2023–2024 for validation, and January 5, 2025 to May 10, 2026 for independent testing. The test set contained 63,910 horse-level observations from 4,556 races. Confidence intervals were estimated using 10,000 race-level bootstrap resamples. Two horse-level binary outcome labels were evaluated: a win outcome and a JRA place-rule-compatible place outcome derived from the processed field_size variable. Compared with the augmented current_full_model, the hyperparameter-matched no-theory sensitivity-analysis model matched_no_theory_model performed better for both outcomes. For the win outcome, ROC AUC was 0.7543 (95% CI: 0.7475–0.7609) for matched_no_theory_model and 0.7293 (95% CI: 0.7224–0.7362) for current_full_model. For the JRA place-rule-compatible place outcome, ROC AUC was 0.7513 (95% CI: 0.7469–0.7558) and 0.7164 (95% CI: 0.7118–0.7212), respectively. An exploratory race-level uncertainty diagnostic was evaluated separately and was not incorporated into the horse-level prediction models. These findings indicate that the evaluated theory-score block did not improve temporal test-set performance. Domain-derived feature blocks should be evaluated incrementally through leakage audits, temporal validation, feature-block ablation, and matched sensitivity analyses before deployment.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1873285</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1873285</link>
        <title><![CDATA[ Performance evaluation of ARIMA, VAR, ANN, GRU, and LSTM for multivariate healthcare time-series prediction: a rigorous, reproducible benchmarking study with ablation analysis on MIMIC-III and eICU]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Md. Tazul Islam</author><author>Muhammad Hanif</author><author>Samuel Kim</author>
        <description><![CDATA[IntroductionAccurate forecasting of physiological time-series in the intensive care unit (ICU) is critical for early-warning systems, deterioration detection, and dynamic resource allocation. Yet the comparative literature suffers from five persistent methodological deficiencies: non-unified experimental conditions, heterogeneous preprocessing, univariate classical baselines compared against multivariate neural models, single-run point estimates without variance, and underspecified prediction tasks.MethodsThis paper addresses all five deficiencies through a standardized, reproducible benchmarking study comparing Autoregressive Integrated Moving Average (ARIMA), Vector Autoregression (VAR), Artificial Neural Networks (ANN), Gated Recurrent Units (GRU), and Long Short-Term Memory (LSTM) networks on one-step-ahead (1-h-ahead) prediction of heart rate, respiratory rate, systolic and diastolic blood pressure, and oxygen saturation across two large-scale ICU cohorts (8,000 MIMIC-III and 15,000 eICU admissions). VAR is introduced as a multivariate linear baseline alongside univariate ARIMA, enabling a three-way decomposition of performance gains into multivariate modeling, nonlinear approximation, and gated temporal memory. All experiments are repeated over 10 independent seeds and reported as mean ± 95% confidence interval with Diebold-Mariano, paired t-test, and Bonferroni-corrected ANOVA significance testing. Seven categories of ablation study, a clinical alarm simulation, a leave-one-hospital-out protocol, and a subgroup fairness analysis are additionally conducted.ResultsLSTM achieves the highest performance on MIMIC-III (RMSE 0.082 ± 0.004, R2 0.940 ± 0.008, directional accuracy 90.1%±0.9%), with GRU reaching near-equivalent accuracy at 64% of training cost. The layered decomposition reveals that gated temporal memory provides the largest marginal gain (−47% to −53% RMSE over ANN), followed by nonlinear approximation (−19% over VAR) and multivariate modeling (−13% over ARIMA). Ablation identifies normalization as the single most critical preprocessing step (+79.3% RMSE without it) and a window length of l = 10 as optimal. The clinical alarm simulation demonstrates a 38.4% false alarm rate reduction for LSTM relative to threshold-only monitoring, and the leave-one-hospital-out experiment confirms cross-institutional generalizability with consistent model ranking.DiscussionSubgroup fairness analysis identifies a performance gap for severely ill patients (Sequential Organ Failure Assessment score ≥ 8, ratio 1.27) warranting targeted future work. These results establish a reusable, methodologically rigorous benchmark for ICU physiological forecasting and provide actionable model-selection guidance for clinical deployment.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1922657</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1922657</link>
        <title><![CDATA[A transformer-based reinforcement learning framework for autonomous training of industrial robotic manipulators]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Balnur Kenjayeva</author><author>Alibek Galimov</author>
        <description><![CDATA[IntroductionAutonomous robotic manipulation in modern manufacturing requires control policies that simultaneously achieve precision, adaptation, operational safety, and multi-task capability.MethodsThis paper proposes a Transformer-Based Reinforcement Learning Network (TRL-Net) that combines temporal state encoding, task-conditioned representation, actor-critic reinforcement learning, domain-randomized simulation, and safety-constrained action projection. The evaluation is limited to ten simulated industrial manipulation tasks.ResultsThe reported point estimates are a 98.6% task success rate, 0.81 mm average positioning error, 0.0% collision rate in the evaluated simulation episodes, 0.42 m/s³ trajectory jerk, and 4.2 ms policy inference time.DiscussionThese findings are consistent with improved temporal control and multi-task performance relative to the reported baselines. However, the archived experimental record does not contain real-robot trials, component-wise ablations, seed-level dispersion, complete hyperparameters, or hardware specifications. The conclusions are therefore restricted to the reported simulation setting, and the safety layer is interpreted as enforcing configured simulator constraints rather than guaranteeing safe operation on physical industrial hardware.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1930455</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1930455</link>
        <title><![CDATA[The Generative Shortfall: a comparative analysis of autoregressive vs. extractive architectures in ABSA]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Chathurvedi V. Rama Chandra</author><author>T. Padmavathy</author>
        <description><![CDATA[In natural language processing (NLP), Aspect-Based Sentiment Analysis (ABSA) remains a challenge, particularly when handling ambiguous, domain-specific “neutral” sentiment classifications in cases of severe class imbalance. While there has been a pivot toward deploying generative Small Language Models (SLMs) on resource-limited edge devices, their performance on unevenly distributed data remains under-explored. This paper presents a comprehensive, cross-paradigm evaluation of the following ABSA architectures: a sequence-based BiGRU, a state-of-the-art DeBERTa-v3 encoder, and Alibaba’s Qwen2.5-1.5B generative SLM. To evaluate the feasibility of deploying the SLM on edge devices, we modified the Qwen2.5 model to work on a 6GB VRAM device using BFloat16 quantized low-rank adaptation (QLoRA) and an abstract syntax tree (AST) parser. A closer look at the results shows a fundamental Generative Shortfall. Despite optimizing the SLM for edge devices, its performance was suboptimal (accuracy: 79.67%, macro-F1: 0.71), indicating that the model’s conservative token generation behavior limited its ability to identify sparse and subtle semantic targets within the text. While the DeBERTa-v3 encoder, when balanced with multi-sample dropout and layer-wise learning rate decay (LLRD), achieved a macro-F1 of 0.87, successfully identifying 81.95% of the minority neutral class. Supported by bootstrap resampling and McNemar’s statistical significance test (p<0.001), the results show that although generative SLMs can be efficiently deployed on resource-constrained edge devices, structurally regularized extractive encoders provide significantly better performance for precise opinion mining on highly imbalanced data.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1849408</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1849408</link>
        <title><![CDATA[AGMNet: an explainable gated multimodal framework integrating MRI heterogeneity and dopaminergic asymmetry for Parkinson's disease classification]]></title>
        <pubdate>2026-09-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>T. Gnana Abinaya</author><author>R. Sivashankari</author>
        <description><![CDATA[IntroductionParkinson's Disease (PD) is a neurological disorder that worsens over time and is marked by interhemispheric asymmetry, dopaminergic degeneration, and structural changes in the brain.MethodsThis paper presents a novel approach: AGMNet- Adaptive Gated Multimodel Network, which is an explainable gated multimodal deep learning framework that combines structural magnetic resonance imaging (MRI), dopamine transporter (DaT) imaging, asymmetry-aware representations, and biomarker features for robust PD classification. The Grad-CAM method is incorporated with the proposed AGMNet framework to improve the model interpretability by identifying image regions that influence classification decisions. The performance of AGMNet framework was evaluated using five-fold cross-validation and compared with established deep learning architectures, including VGG16, VGG19, InceptionV3, and Xception.ResultsThe proposed AGMNet has achieved 0.8318 of accuracy, 0.8278 of balanced accuracy, sensitivity of 0.8556, specificity of 0.8000, AUC of 0.711, and Matthews correlation coefficient (MCC) of 0.6458. The proposed AGMNet model demonstrated competitive classification performance relative to the evaluated baseline architectures. The combination of DaT imaging, asymmetry-aware features, and biomarker representations in the proposed model, contributed more strongly to the fused representation than MRI features alone.DiscussionThe findings indicate that adaptive multimodal fusion in the AGMNet framework can improve the integration of complementary structural, functional, asymmetry-related, and biomarker information for PD classification. The learned gate weights further demonstrate the relevance of functional and asymmetry-aware information within the proposed framework, while Grad-CAM provides additional insight into model decision-making. Therefore, the AGMNet framework offers a potentially useful and interpretable framework for multimodal PD classification.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1903775</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1903775</link>
        <title><![CDATA[VMamba-QAG-net: a five-stage pipeline for SYNTAX score computation and decision support in interventional cardiology using X-ray angiography]]></title>
        <pubdate>2026-09-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>G. Sunilkumar</author><author>P. Kumaresan</author>
        <description><![CDATA[Coronary Artery Disease (CAD) remains a significant worldwide health problem, making accurate evaluation of disease severity essential for treatment planning. The SYNTAX score, derived from X-ray coronary angiography (XCA) images, plays a key role in guiding treatment decisions. However, manual scoring is time-consuming, subject and prone to inter-observer variability, necessitating the development of an objective and reliable system. We proposed VMamba-QAG-Net, a novel five-stage pipeline designed for SYNTAX score estimation directly from XCA images. Stage 1 performs binary vessel segmentation to resolve extreme class imbalance, using a preprocessing of Bilateral Filtering, CLAHE, and Unsharp Masking to enhance vessel boundaries, an EfficientSS2D block with parallel five direction visual mamba scanning and Quantum Attention Gate (QAG) to preserve long-range vascular connectivity. Stage 2, executes multi-class anatomical labelling into 27 distinct coronary artery segments using a four-channel guided input, supported by a Vessel Prototype Memory Bank with metric learning and a centreline-weighted skeleton loss for thin distal structures. Stage 3 and 4 extract ordered vessel centrelines via skeletonization, calculates percentage diameter stenosis through perpendicular ray casting, and detects grade stenotic lesions, using Aquila Optimiser. Stage 5 combines the stenosis information to calculate the SYNTAX score and support treatment decisions in revascularisation. Our proposed model was evaluated on the ARCADE dataset and attained a Dice coefficient of 0.9070 and accuracy of 0.9841 for binary segmentation. For 27-class coronary artery segmentation achieved a Dice coefficient of 0.7512, accuracy of 0.9514, and present class Dice of 0.8214. These results demonstrate the effectiveness of the proposed framework and its potential to support cardiologists in clinical decision-making.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1992698</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1992698</link>
        <title><![CDATA[Correction: Machine learning-based early detection of abnormal heart rate in critically ill patients: a real-world clinical dataset study with automated alert system integration]]></title>
        <pubdate>2026-09-23T00:00:00Z</pubdate>
        <category>Correction</category>
        <author>Joao C. Ferreira</author><author>Luis B. Elvas</author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1938382</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1938382</link>
        <title><![CDATA[LEFTNet: low-frequency enhanced feature transmission network for dual-branch collaborative single image dehazing]]></title>
        <pubdate>2026-09-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Beesetti Vanitha</author><author>Chintakindi Balaram Murthy</author>
        <description><![CDATA[Image dehazing is a basic low-level vision task designed to recover high-fidelity images from hazed images. In autonomous driving, aerial surveillance, and outdoor scene recognition, image quality is significantly degraded by light attenuation due to scattering in the propagation medium, especially under hazy, foggy, or smoggy conditions. While remarkable advances have been made with existing deep learning methods, they remain constrained by over-smoothing, ineffective use of haze frequency, and insufficient consideration of multi-scale feature interactions. To meet these challenges, we introduce a novel single-image dehazing framework, called Low-Frequency Enhanced Feature Transmission Network (LEFTNet), which consists of four special modules: the Haze Suppression Block (HSB), the Dual-Branch Block (DBB), the Cross-Scale Fusion Block (CSFB), and the Channel Mining Mechanism (CMM). In the first stage, a hazy input image is decomposed into two complementary low-frequency representations: a haze-related component (IhazeL) and a frequency-domain component (IfreqL), which are jointly processed by the HSB for early, channel-selective haze attenuation. Then, the encoder extracts multi-scale hierarchical features, which are cleaned up at each stage by the Channel Mining module. The DBB's residual refinement enables intra-image global-to-local feature interactions, while the CSFB's learnable fusion coefficients enable inter-scale feature fusion. A dual fully connected bottleneck with Global Average Pooling (GAP) suppresses haze-dominant channels while enhancing structure-sensitive channels. Two symmetric branches ensure complementary representation learning, and a hybrid loss function comprising L1 loss and perceptual LSSIM loss assures both pixel-wise accuracy and perceptual fidelity. Comprehensive experiments on four benchmarks, RESIDE-6K, I-Haze, O-Haze, and NH-Haze, demonstrate that LEFTNet achieves competitive performance across the evaluated Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Average Mean Brightness Error (AMBE), Gradient Magnitude Similarity Deviation (GMSD), and Visual Saliency-Induced Index (VSI) metrics compared with recent dehazing methods.]]></description>
      </item>
      </channel>
    </rss>