<?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>2026-07-26T11:42:19.760+00:00</pubDate>
        <ttl>60</ttl>
        <item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1854873</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1854873</link>
        <title><![CDATA[Do language families matter? Evaluating LLMs for sentiment analysis through a hierarchical cross-lingual lens]]></title>
        <pubdate>2026-07-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Muhamet Kastrati</author><author>Abdul Manaf</author><author>Ali Shariq Imran</author><author>Zenun Kastrati</author><author>Sher Muhammad Daudpota</author><author>Marenglen Biba</author>
        <description><![CDATA[Social media sentiment analysis has become one of the most significant instruments for understanding the opinion of the population in the spheres of healthcare, politics, and education. Yet, large language models (LLMs) remain unevenly distributed in their linguistic coverage, failing to adequately serve a large portion of the world's languages. This study evaluates five state-of-the-art LLMs: GPT-4o, Gemini 2.0 Flash, DeepSeek-V3, Mistral Large, and Claude 3.7 Sonnet on three-class sentiment classification across 36 datasets spanning 36 languages, with emphasis on low- and medium-resource settings, using zero-shot and few-shot prompting without task-specific fine-tuning. In addition to the traditional measures of performance per language, the study presents a hierarchical analysis of languages based on a genealogical tree of Indo-European, Afro-Asiatic, Niger-Congo, Turkic, Austronesian, and English Creole language families, so that it is possible to identify the systematic patterns of performance superiority and inferiority among the language families. The findings show that few-shot prompting improves the results of a vast majority of languages, with several models approaching or surpassing the performance of the state-of-the-art benchmark of task-specific models. The GPT-4o and Claude achieved the highest performance in the high-resource and medium-resource settings, and Gemini is a competent trade-off that allows balancing the performance and the computational cost. Although it has lower zero-shot performance, Mistral benefits the most from few-shot prompting and becomes highly competitive in the few-shot setting. Despite these developments, the level of performance on low-resource languages, such as Oromo, Xitsonga, Azerbaijani, and Twi, remains significantly lower, underscoring that progress in multilingual LLMs requires moving beyond English-centric evaluation toward genuinely representative and globally inclusive benchmarks.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1803320</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1803320</link>
        <title><![CDATA[Mapping seasonal dynamics of forage and cereal crops in a hyper-arid environment using Sentinel-1 and Sentinel-2 time series]]></title>
        <pubdate>2026-07-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Areej Alwahas</author><author>Kasper Johansen</author><author>Jorge Rodriguez</author><author>Matthew F. McCabe</author>
        <description><![CDATA[IntroductionIn arid and hyper-arid regions, agriculture depends heavily on irrigation, making crop type monitoring important for water allocation, monitoring crop management policies, and providing the information required to forecast food supply. However, field labels are often scarce, and crop calendars can shift due to locally managed planting, harvest, and irrigation decisions, complicating mapping at field-scale.MethodsWe present a seasonal crop type mapping approach applied to Wadi Al-Dawasir, Saudi Arabia, generating maps for 2020–2024 from biweekly optical and radar satellite time series. The method learns representations from unlabeled imagery through self-supervised pretraining and fine-tunes a segmentation model using a small set of field observations with pseudo-label augmentation from unsupervised clustering. We mapped four classes: fallow, cereal, vegetables, and forage, and evaluated performance using overall accuracy, F1-score, and mean intersection-over-union (mIoU).ResultsThe configuration combining self-supervised pretraining, pseudo-label augmentation, and optical-radar inputs achieved an mIoU of 0.80, an F1-score of 0.88, and an overall accuracy of 0.97. In contrast, using optical information alone substantially reduced accuracy, with an mIoU of 0.35, an F1-score of 0.32, and an overall accuracy of 0.64. Fallow and forage produced the highest mapping accuracies, with mIoU values of 0.97 and 0.85, respectively, while vegetables had the lowest accuracy among the four crop groups, with an mIoU of 0.52.DiscussionThese results show that self-supervised temporal pretraining combined with pseudo-label augmentation can support efficient multi-season, field-scale crop mapping using limited labels in irrigation-driven arid regions. The lower accuracy for vegetables highlights the continued challenge of mapping heterogeneous crop groups with overlapping phenological patterns.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1826465</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1826465</link>
        <title><![CDATA[AI-based secure event-driven serverless architecture for scalable digital civic participation platform]]></title>
        <pubdate>2026-07-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Aizhan Kassymova</author><author>Abdul Razaque</author><author>Raissa Uskenbayeva</author><author>Zhuldyz Kalpeyeva</author><author>Aizhan Anartayeva</author>
        <description><![CDATA[IntroductionWith the growing digitalization of urban governance and the increasing demand for transparency, sustainability and secure decision-making, the need for scalable and intelligent digital civic platforms has been raised. However, current e-participation systems are often plagued by challenges related to scalability, regulatory compliance, digital sovereignty and secure citizen authentication. The challenges are tackled in this paper by proposing an AI-enabled serverless architecture for next generation digital civic engagement.MethodsThis study proposes an AI-based Secure Event-driven Serverless Participation Architecture (SESPA) for digital e-participation services based on the Citizen Participation Event Model (CPEM), where each citizen interaction is treated as an event within a continuous decision-making process. Architecture employs an event-driven serverless computing paradigm integrated with artificial intelligence modules for biometric citizen verification and anomaly detection. To satisfy the digital sovereignty requirements of the Republic of Kazakhstan, a hybrid data localization model is introduced that separates personally identifiable information from anonymized analytical events. The proposed dual-loop architecture stores sensitive citizen data within national infrastructure while enabling cloud-based processing of anonymized event streams for scalable analytics without violating regulatory requirements.ResultsAn experimental prototype was evaluated under workloads of up to 10,000 concurrent users. The results demonstrated stable latency across p50, p75, p95, and p99 percentile metrics, efficient scalability through provisioned concurrency, and reduced total cost of ownership compared with an equivalent Kubernetes-based deployment. Moreover, the addition of AI modules to the event-processing pipeline added little latency overhead and allowed for precise detection of anomalous and suspicious participation behavior in controlled experimental workloads.Discussion and conclusionThe proposed SESPA architecture effectively combines event-driven serverless computing, AI-assisted security mechanisms and hybrid data localization to provide a secure, scalable and regulation-compliant digital participation platform. The results demonstrate that the proposed framework offers a good technology foundation for next generation smart city applications by supporting high citizen engagement, regulatory compliance, digital sovereignty and intelligent decision-making, while maintaining high system performance and cost efficiency.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1760912</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1760912</link>
        <title><![CDATA[From mechanistic models to artificial intelligence: exploring the potential of digital twins in geriatric oncology]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Panagiotis Karampelesis</author><author>Spyros Denazis</author><author>Odysseas Koufopavlou</author><author>Evangelia I. Zacharaki</author>
        <description><![CDATA[This survey explores how machine learning and artificial intelligence (AI) can be integrated with mechanistic models to create more accurate, dynamic, predictive, and personalized representations of biological systems, commonly referred to as digital twins (DTs). Mechanistic models, such as pathway-based Boolean or differential equation frameworks, provide interpretable insights into biological processes; however, calibrating these models to represent individual variability across large, heterogeneous cohorts remains a significant challenge, as their physically constrained structures often lack the flexibility to capture complex, non-mechanistic nuances in patient data. Focusing on elderly cancer patients–a vulnerable population underrepresented in clinical research–we discuss how hybrid DTs can bridge the gap between interpretable mechanistic frameworks and flexible, predictive AI approaches, enabling continuous monitoring, risk stratification, and adaptive treatment planning. To illustrate these principles, we present a proof-of-concept case study involving a synthetic breast cancer dataset in which comprehensive geriatric assessment, clinical tests, and quality of life measures inform dosing decisions for older patients via a Markov Decision Process. By combining a synthesis of current literature with the application of a sequential decision-making framework optimized using longitudinal data, this work provides a foundational understanding for researchers and clinicians interested in leveraging DTs to improve personalization and outcomes in geriatric oncology.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1821929</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1821929</link>
        <title><![CDATA[HIDANet: a lightweight deep learning framework for Vannamei post-larval stage classification and morphometric estimation with background bias validation]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Sugunapriya A</author><author>Markkandan S</author>
        <description><![CDATA[IntroductionQuality control of hatchery production relies on accurate developmental staging of the Pacific white shrimp Litopenaeus vannamei post-larvae (PL), but current methods rely on subjective manual visual evaluation that leads to observer bias and inconsistency.MethodsIn this study, the Hierarchical Isotropic Dense Attention Network (HIDANet) has been introduced, a lightweight convolutional neural network with 0.033M parameters that learns to classify seven post-larval stages (PL5–PL12) in a digitally obtained image of a defined larva. Its architecture uses multibranch isotropic depthwise separable convolution, channel-wise concatenation, and a Convolutional Block Attention Module (CBAM) to recalibrate sequential channel and spatial features. The background color bias was removed using an aggressive augmentation scheme that included strong color jittering, stochastic grayscale conversion, random auto-contrast, and geometric perturbations with inverse-frequency weighted sampling to address class imbalance.ResultsHIDANet trained on 5,835 collected hatchery images reached a test accuracy of 98.44% with color inputs and 96.89% with grayscale inputs, with a macro-averaged F1-score of 0.99, which confirms strong morphological learning that is independent of chromatic background signals. An end-to-end workflow combining CLAHE image enhancement, Gaussian adaptive thresholding, and skeleton-based morphometric filtering offers automated larva counting, area and length measurements of the population, and density categorized into three levels (Low, Medium, and High).DiscussionThe suggested framework allows for classifying the stages of PL and assessing larval numbers in the same image simultaneously, which is a scalable and low-cost method of quality monitoring in commercial aquaculture systems in real time.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1867175</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1867175</link>
        <title><![CDATA[An explainable end-to-end computer vision pipeline for detection, segmentation, and reconstruction of occluded weapons in forensic imagery]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Vaibhav Rohella</author><author>Kumar Anurag</author><author>Aditya Kumar</author><author>Manjula V.</author><author>Shanthi P.</author>
        <description><![CDATA[IntroductionImages from crime scenes often show partially concealed weapons due to obstructions such as hands and clothing, as well as surveillance camera limitations, which affect the efficacy of traditional detection methods. This work proposes an explainable forensic pipeline for occluded weapons detection, segmentation, and reconstruction.MethodsThe proposed framework integrates RT-DETR-L, a transformer-based weapon detection model; MobileSAM for zero-shot segmentation of visible weapon regions; Stable Diffusion Inpainting for reconstruction of occluded regions; and a bilateral filter with Canny edge detection for forensic sketch generation. The detector was trained using synthetic occlusion augmentation on five weapon categories, such as firearm, grenade, knife, pistol, and rocket, and fire as an additional class for the environmental hazard indicator.ResultsThe RT-DETR-L model achieves a mAP@50 of 0.86 and a mAP@50-95 of 0.65, averaged over the five weapon classes, on a dataset with synthetic occlusion augmentation. Reconstruction quality meets minimum quality thresholds (SSIM > 0.54, PSNR > 16 dB, Region IoU > 0.80) up to approximately 50% occlusion. An interactive Streamlite application demonstrates the pipeline’s feasibility in 42–46 seconds as a controlled laboratory prototype.DiscussionThe proposed explainable forensic pipeline highlights the potential of combining transformer-based detection with generative AI to assist forensic weapon analysis under challenging occlusion scenarios. The generated reconstructions and sketches are intended as visual aids for expert review and must undergo independent forensic validation before any operational use.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1882814</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1882814</link>
        <title><![CDATA[The VIBE-HI framework: a conceptual model for evaluating vibe coding appropriateness, quality, and safety in health informatics]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ahmed Alqheedan</author><author>Saleh Alzughaibi</author>
        <description><![CDATA[BackgroundVibe coding—generating software through natural-language prompts to large language models without reviewing the underlying code—has moved rapidly from consumer technology into peer-reviewed clinical applications. By early 2026, clinicians had published vibe-coded teaching tools, a validated clinical nomogram, and an end-to-end omics platform built in under 10 minutes for under two dollars. Collins Dictionary named vibe coding its 2025 Word of the Year. No governance framework currently addresses the practice in healthcare.ObjectiveTo introduce VIBE-HI, a health-informatics-specific framework for evaluating the appropriateness, quality, and safety of vibe coding across clinical contexts, and to specify its decision logic, quality constructs, and regulatory mapping in operational detail.MethodsVIBE-HI was developed as a conceptual framework through a structured, theory-informed narrative synthesis of three literatures—emerging biomedical vibe-coding reports, empirical software-engineering and security research on AI-generated code and established sociotechnical health-informatics theory and software-quality standards—following recognized conceptual-framework methodology. It was refined through illustrative application to four published clinician-built tools. This is a conceptual contribution; it is not a systematic review or a consensus (Delphi) study, and formal empirical validation is identified as the next step.ResultsVIBE-HI organizes governance into three sequential layers. (1) Risk and Role Stratification assign one of four risk tiers—Green, Yellow, Orange, Red—and a matched clinician-developer role, from prototype to requirements analyst, using four criteria combined by an explicit dominant-criterion rule. (2) Quality and Validation extend ISO/IEC 25010:2023 with three measurable constructs—Code Provenance Transparency, Comprehension Coverage, and Hallucination Resilience—each with defined indicators and tier-dependent thresholds. (3) Compliance and Governance maps HIPAA, IEC 62304, FDA SaMD criteria, and the EU AI Act onto each tier and binds a named accountability owner. The framework treats comprehension abdication—the structural surrender of understanding to a generative system—as the core sociotechnical hazard distinguishing vibe coding from prior AI-assisted development, grounded in the automation-bias, responsibility-gap, and sociotechnical-systems literatures.ConclusionClinical vibe coding needs risk-stratified governance now, before largely invisible adoption outpaces the field’s capacity to assess it. VIBE-HI offers an architecture institutions can apply immediately and provides a clear pathway for empirical validation, beginning with a modified-Delphi consensus study and stakeholder review.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1848216</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1848216</link>
        <title><![CDATA[A four-module neural architecture for the automatic extraction and classification of causal relations in text]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Roman Taberkhan</author><author>Nurbolat Tasbolatuly</author><author>Madina Sambetbayeva</author><author>Saule Tazhibayeva</author><author>Nurmira Zhumay</author><author>Bayangali Abdygalym</author><author>Mira Kaldarova</author>
        <description><![CDATA[This article presents a four-module system for the automatic extraction and classification of causal relationships from texts in the Kazakh language, based on the fine-tuning of the KazBERT transformer language model. The proposed architecture includes four specialized modules: recognition of lexical causality markers (Token Classification, B/I-MARKER); segmentation of cause-effect clauses (Token Classification, B/I-CAUSE · B/I-EFFECT); classification of Tv forms of markers (Sequence Classification, 16 classes); determination of the type of the marker’s syntactic construction—Model Group (Sequence Classification: SYNTHETIC/ANALYTIC/ANALYTICO-SYNTHETIC). The training was conducted using an original annotated corpus consisting of 3,223 sentences in the Kazakh language. The architecture is supplemented by a deterministic positional inversion algorithm for explanatory markers (sebebi, öitkenı, sondyqtan, etc.), which automatically restores the correct CAUSE-EFFECT argument order. Experiments have demonstrated that KazBERT outperforms the baseline models XLM-RoBERTa and mBERT: macro-F1 scores were 0.901 (tags), 0.865 (clauses), 0.884 (Tv-form), and 0.927 (construction type). The scientific novelty lies in the first publicly released four-level annotated corpus of Kazakh causal constructions, the operationalization of the established Turkological synthetic/analytic distinction—extended with a corpus-attested ANALYTICO-SYNTHETIC class—as a four-module annotation target, and a deterministic positional-inversion post-processor that corrects systematic argument-order errors for analytic markers.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1881404</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1881404</link>
        <title><![CDATA[Hybrid fuzzy C-means and deep learning framework for intelligent fault classification in solar PV systems]]></title>
        <pubdate>2026-07-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>V. Vignesh</author><author>R. Senthil Kumar</author><author>G. Suganeshwari</author>
        <description><![CDATA[Photovoltaic (PV) systems have proven themselves to be a viable alternative energy source; however, there are multiple faults related to PV systems which cause energy losses and low efficiencies. Manual or rule-based algorithms are traditionally used for fault diagnosis, which are not efficient and unsuitable for real-time applications. In this paper, a novel hybrid intelligent classification system for PV fault detection is proposed by integrating Fuzzy C-Means (FCM) clustering and Deep Learning (DL) techniques such as Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM) and Gated Recurrent Unit (GRU). The dataset consists of 102,400 samples collected from a real-time 10 kW solar PV system operating under varying irradiance conditions ranging from 600 W/m2 to 1,000 W/m2 and temperature conditions ranging from 25 °C to 40 °C. The FCM technique is used to enhance the extracted features by clustering the membership functions, and the obtained features are used for model training. The performance of proposed models is evaluated using the classification metrics and confusion matrices. The proposed FCM + GRU model achieved 91.13% accuracy 0.79 precision, 0.76 recall, and F1-score of 0.78. The obtained results confirm the effectiveness of the proposed hybrid framework by improving fault classification performance under various operating environments.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1881767</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1881767</link>
        <title><![CDATA[How U.S. Federal Artificial Intelligence (AI) policy is shaping agrifood systems: an integrative review]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Policy and Practice Reviews</category>
        <author>Cole Baerlocher</author><author>Sarah McCord</author><author>Elizabeth Tabares</author><author>Arturo España</author><author>E. Alex Keasler</author><author>Rafael Landaverde</author>
        <description><![CDATA[Artificial intelligence (AI) is increasingly shaping how agrifood systems function in the United States, yet the role of federal policy in guiding its use, oversight, and broader consequences is still not well defined. This study explores how current U.S. federal AI policies support-or limit-the advancement of agrifood systems by synthesizing evidence from publicly available policy documents. Using a focused search strategy and qualitative content analysis, we reviewed nine federal policy documents released through September 2025 to identify key priorities and overlooked areas relevant to agriculture. Our analysis revealed six recurring themes: environment, precision agriculture, workforce development, governance, technological infrastructure, and partnership. The findings show that federal AI policy places considerable emphasis on building infrastructure, strengthening workforce capacity, and establishing governance frameworks. At the same time, less attention is given to environmental trade-offs, equitable access for small- and mid-scale producers, and the place specific conditions that shape agricultural practice. Notably, tensions emerge between policies that promote rapid expansion of AI infrastructure and those aimed at protecting environmental resources and strengthening climate resilience. Taken together, the results suggest that although agriculture is increasingly recognized within the national AI agenda, the lack of a coordinated, agriculture specific policy framework may lead to uneven adoption and unintended outcomes across the agrifood system. This study offers a synthesized policy foundation to support future research, inform decision making, and engage stakeholders in aligning AI innovation with more sustainable and equitable agrifood systems.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1914515</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1914515</link>
        <title><![CDATA[Enterprise AI applications and stock price crash risk]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Zhao Yang</author>
        <description><![CDATA[With Chinese A-share listed enterprises (2015–2024), this paper studies the influence of artificial intelligence applications on stock price crash risk. Our conclusion includes that AI applications significantly reduce the crash risk. The mechanism analysis reveals that artificial intelligence works by improving internal controls quality (ICQ) and decision-making efficiency (DME). Heterogeneity tests show the effect is stronger for enterprises with digital expertise among CEO and located in eastern China. This paper provides a theoretical basis for enterprises to optimize their development strategies and prevent capital market risk.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1932998</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1932998</link>
        <title><![CDATA[Correction: The legalization of international instruments: a hybrid RAG scoring framework based on chain-of-thought prompting]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Correction</category>
        <author>Yan Chen</author><author>Zihua Zeng</author><author>Muhamad Sayuti Hassan</author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1842584</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1842584</link>
        <title><![CDATA[Application of artificial intelligence models in the identification of severe scrub typhus]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Chunyu Chen</author><author>Xiangling Liu</author><author>Ce Yang</author><author>Chuncheng Ma</author><author>Qingxia Zhou</author><author>Junying Zeng</author><author>Yinan Chen</author><author>Peipei Ding</author><author>Ran Li</author><author>Wencong Liang</author><author>Zhiyong Hong</author><author>Chuanbo Qin</author><author>Gang He</author>
        <description><![CDATA[This retrospective study enrolled 492 patients with scrub typhus in Jiangmen from 2013 to 2025. Clinical and laboratory data were analyzed using univariate logistic regression and LASSO regression to identify risk factors for severe illness. Seven machine-learning models, including logistic regression, support vector machine, random forest, XGBoost, Naive Bayes, k-nearest neighbor, and decision tree, were constructed and externally validated. Variable importance was ranked using SHAP analysis. Mechanical ventilation represented the strongest predictor of severe disease, indicating that respiratory failure is the core indicator of disease progression. Elevated bilirubin, prolonged coagulation time, and thrombocytopenia were also closely associated with severe scrub typhus. XGBoost achieved the best predictive performance, with AUC values of 0.888 and 0.928 in the training and validation cohorts, respectively. Owing to limited positive cases, mortality prediction yielded lower AUCs (0.856 and 0.814). These findings demonstrate that artificial intelligence models can effectively stratify the severity of scrub typhus and present favorable potential in prognostic assessment.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1869343</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1869343</link>
        <title><![CDATA[Climate-control ChatGPT for EFL writing and AI literacy]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Hesham Aldamen</author><author>Salwa Mohammad Alawneh</author><author>Mohamad Almashour</author><author>Mutasim Al-Deaibes</author><author>Rami Alsharefeen</author>
        <description><![CDATA[ChatGPT is increasingly used in EFL writing, yet applied linguistics lacks robust methods for measuring whether human-AI dialogue improves learners’ AI literacy, harm-sensitive reasoning, and independent argumentative writing. This mixed-methods quasi-experimental study examines whether climate-control ChatGPT, a prompted interactional design that requires learners to justify, qualify, counterexample, and re-author their own claims, produces stronger gains than standard ChatGPT use and no-ChatGPT instruction. The study involved 120 undergraduate EFL learners in six intact second-year academic writing sections at a public university in Jordan. Participants were assigned to no-ChatGPT control, standard ChatGPT, or climate-control ChatGPT conditions. Data included no-AI diagnostic, pretest, posttest, and delayed transfer essays; ChatGPT dialogue logs; first and final drafts; learner reflections; uptake traces; stimulated-recall interviews; and 9,113 HARM-CAL annotated language units. HARM-CAL, Harm-sensitive Reasoning and Anthropomorphism Calibration in Language, classified learner language for naïve relativism, moral imposition, harm-sensitive reasoning, AI anthropomorphism, calibrated AI literacy, and reflective uncertainty. Results showed that the climate-control condition produced the largest gains in harm-sensitive AI literacy and argumentative writing quality, with advantages maintained on delayed no-AI transfer. It also reduced direct uptake, increased self-repair and authorial control, and strengthened calibrated distinctions between AI output competence and consciousness. The study contributes a classroom-tested intervention and a transparent computational framework that combines reproducible annotation procedures, a fully specified TF-IDF baseline, and archived transformer-family classifier outputs for analyzing responsible AI literacy in second-language argumentation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1869820</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1869820</link>
        <title><![CDATA[On the fragility of neural architecture search: the role of overfitting and task complexity in medical image analysis]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>A. Gómez</author><author>M. Desco</author><author>M. Abella</author>
        <description><![CDATA[IntroductionNeural Architecture Search (NAS) effectively automates Deep Learning pipeline design but is prone to validation overfitting when applied to complex tasks, such as medical image analysis. To mitigate this and enhance generalization, researchers frequently integrate Deep Ensemble Learning (DEL) and data augmentation into the NAS workflow. However, the assumption that these methodologies do not negatively interfere in high-overfitting scenarios remains unproven.MethodsWe evaluated NAS, combined with DEL and data augmentation pipelines, across both CIFAR-10 and a biomedical CT dataset, assessing the influence of task complexity and data scarcity in their interaction. Using an ablative experimental design, we isolated the contributions of DEL and NAS and mapped data augmentation sensitivity landscapes at both global and local scales.ResultsWhile synergistic on large, well defined datasets, statistical analysis revealed that DEL failed to significantly enhance the generalization capabilities of NAS-generated populations in data-constrained regimes. Moreover, we identified local roughness within the data augmentation sensitivity landscapes.DiscussionOur findings challenge the prevailing assumption of unconditional methodological synergy that guides joint architecture exploration, ensemble pruning and data augmentation optimization.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1934012</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1934012</link>
        <title><![CDATA[Correction: Performance of large language models in neonatal resuscitation assessments versus healthcare providers: an exploratory study]]></title>
        <pubdate>2026-07-22T00:00:00Z</pubdate>
        <category>Correction</category>
        <author>Frontiers Production Office </author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1836641</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1836641</link>
        <title><![CDATA[Quantum-enhanced generative artificial intelligence: a critical review of classical limitations, complexity barriers, and hybrid quantum–classical architectures]]></title>
        <pubdate>2026-07-21T00:00:00Z</pubdate>
        <category>Review</category>
        <author>Diljot Singh</author><author>Omana J.</author><author>Smrithy G. S.</author>
        <description><![CDATA[The rapid commercialization of generative artificial intelligence (AI), along with the maturation of quantum technologies has raised a question: can quantum-powered neural networks become the next major shift in large language model (LLM) technology? This naturally leads to another misconception that quantum systems will replace classical LLMs. In this study, both architectures are compared in a contrastive manner in terms of mathematics. The data reveals that identical dynamics that help classical systems learn natural language distributions constrain its ability to use efficient sampling of quantum-mechanical spaces. Performing complexity-theoretic separations (i.e., the widely believed but unproven conjecture that BPP ⊆ BQP) and a 2025 preprint reporting experimental demonstrations of quantum advantage for generative tasks we conclude that quantum utility is unlikely to lie in tasks involving natural language processing under current architectures, but rather in certain computational subroutines. We then suggest a hybrid quantum-classical architecture as the best direction to take in the future, as it has the advantages of both paradigms. This is done by studying a case study that optimizes retrieval-augmented generation (RAG) pipelines with Grover's search algorithm.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1858838</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1858838</link>
        <title><![CDATA[Explainable pulmonary fibrosis detection using edge-strengthened dilated holistic edge detection-based lung segmentation and ResNet-V2 classification]]></title>
        <pubdate>2026-07-21T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>K. Mahapackialakshmi</author><author>G. Jaffino</author>
        <description><![CDATA[IntroductionPulmonary fibrosis (PF) is a progressive interstitial lung disease that requires accurate and early detection to improve patient survival and treatment planning.MethodsThis study proposes an explainable deep learning framework for pulmonary fibrosis detection from chest X-ray images by integrating an edge-strengthened dilated holistic edge detection (ES-D-HED) segmentation network with a fine-tuned ResNet152V2 classification model. Unlike the conventional HED-based approaches, the proposed ES-D-HED architecture incorporates an additional dilated intermediate-output branch and enhanced multi-scale edge fusion to improve contextual boundary modeling and fibrosis-related structural edge continuity. The framework combines edge-aware lung segmentation, fibrosis classification, and Grad-CAM-based explainability to provide interpretable clinical decision support. The model was evaluated on a curated subset of the publicly available NIH Chest X-ray dataset using patient-level five-fold cross-validation. Since the NIH dataset does not contain fibrosis segmentation masks, representative PF regions were retrospectively annotated by a clinical expert radiologist for quantitative validation.Results and DiscussionExperimental results demonstrated a classification accuracy of 98.6%, sensitivity of 98.0%, specificity of 99.2%, and F1-score of 98.5%. The proposed segmentation model achieved Dice similarity coefficients of 0.904 for normal lung segmentation and 0.843 for PF region segmentation, indicating strong structural alignment with expert annotations. Grad-CAM visualization also confirmed that the model successfully identified abnormal lung areas associated with fibrosis. The proposed framework shows the potential application of edge-strengthened explainable deep learning in the PF screening system based on chest X-ray imaging. Future studies will involve the classification of multi-class interstitial lung disease, grading the severity of fibrosis, and the multi-center clinical validation for real-world applicability.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1856972</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1856972</link>
        <title><![CDATA[A multi-model prediction of a stage-specific prognosis for colorectal cancer using attention-driven deep ensemble learning on genomic profiling data]]></title>
        <pubdate>2026-07-21T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>K. Supriya</author><author>A. Anitha</author>
        <description><![CDATA[IntroductionOver the decades, shifts in human lifestyle have led to alterations in dietary habits. The consumption of diets low in fiber and high in fat and sugar results in the production of carcinogenic metabolites during digestion. The food we consume undergoes a complex series of processes involving digestion and excretion, engaging various internal organs within the human body. The DNA and MiRNA present in food are crucial for sustaining human health. Damage to human organs can lead to the development of cancer cells. Among various cancers, Colorectal Cancer (CRC) is the 3rd most common cancer that contributes to the increase in the mortality rate worldwide.MethodsOne of the better ways for early CRC diagnosis is through Genomic Profiling. Frequent mutations in the genes APC, TP53, KRAS, PIK3CA, and SMAD4 are observed in the collected samples, contributing to CRC. An effort has been made in the proposed work to improve classification and prediction by using a deep ensemble learning approach based on a self-attention-based stacked bidirectional LSTM, optimized with the Parallel-Whale Optimization Algorithm (WOA) for global convergence, following a feature selection process. Furthermore, the survival analysis of CRC patients is performed using the DeepSurv technique to assess treatment effectiveness and patient care.ResultsThe performance of the proposed model is evaluated using various metrics for stage-based colorectal cancer prediction, and a comparative analysis is conducted to validate against benchmarking techniques, resulting in improved classification and prediction accuracy.DiscussionThis study may help physicians detect CRC earlier and improve patient management.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/frai.2026.1881434</guid>
        <link>https://www.frontiersin.org/articles/10.3389/frai.2026.1881434</link>
        <title><![CDATA[Ontology-based approaches for multi-destination tourism planning: a systematic literature review]]></title>
        <pubdate>2026-07-20T00:00:00Z</pubdate>
        <category>Systematic Review</category>
        <author>Ummi Maisarah Izani</author><author>Nur Syadhila Che Lah</author>
        <description><![CDATA[Tourism planning is becoming increasingly complex as travel behavior shifts from single-destination visits to multi-destination itineraries. However, many tourism information systems still rely on point-of-interest data and recommendation algorithms that lack the semantic structures needed to represent relationships between destinations. This limitation is important in smart tourism environments that require interoperable, data-integrated systems to support meaningful travel planning. This study conducts a systematic literature review of ontology and knowledge graph-based approaches for tourism planning to assess their ability to support itinerary modelling. Using PRISMA guidelines, relevant studies were retrieved from Scopus, ScienceDirect and IEEE Xplore, with 21 articles meeting all eligibility criteria. The review identifies six themes: ontology and semantic modelling, knowledge graph construction and enrichment, tourism recommendation and personalization, semantic retrieval and question answering, route planning and decision support, and itinerary-level and multi-destination representation. The analysis shows that ontology and knowledge graphs are widely used to organize tourism knowledge and enhance recommendation systems. However, most studies remain focused on entity-level modelling, while explicit semantic modelling of itinerary-level constructs remains limited. This gap highlights the need for ontology frameworks that can formally represent travel sequences and inter-destination dependencies to better support smart-tourism planning.]]></description>
      </item>
      </channel>
    </rss>