Frontiers in Artificial Intelligence最新文献

{"title":"Dynamic taxonomy generation for future skills identification using a named entity recognition and relation extraction pipeline.","authors":"Luis Jose Gonzalez-Gomez, Sofia Margarita Hernandez-Munoz, Abiel Borja, Fernando A Arana-Salas, Jose Daniel Azofeifa, Julieta Noguez, Patricia Caratozzolo","doi":"10.3389/frai.2025.1579998","DOIUrl":"10.3389/frai.2025.1579998","url":null,"abstract":"<p><strong>Introduction: </strong>The labor market is rapidly evolving, leading to a mismatch between existing Knowledge, Skills, and Abilities (KSAs) and future occupational requirements. Reports from organizations like the World Economic Forum and the OECD emphasize the need for dynamic skill identification. This paper introduces a novel system for constructing a dynamic taxonomy using Natural Language Processing (NLP) techniques, specifically Named Entity Recognition (NER) and Relation Extraction (RE), to identify and predict future skills. By leveraging machine learning models, this taxonomy aims to bridge the gap between current skills and future demands, contributing to educational and professional development.</p><p><strong>Methods: </strong>To achieve this, an NLP-based architecture was developed using a combination of text preprocessing, NER, and RE models. The NER model identifies and categorizes KSAs and occupations from a corpus of labor market reports, while the RE model establishes the relationships between these entities. A custom pipeline was used for PDF text extraction, tokenization, and lemmatization to standardize the data. The models were trained and evaluated using over 1,700 annotated documents, with the training process optimized for both entity recognition and relationship prediction accuracy.</p><p><strong>Results: </strong>The NER and RE models demonstrated promising performance. The NER model achieved a best micro-averaged F1-score of 65.38% in identifying occupations, skills, and knowledge entities. The RE model subsequently achieved a best micro-F1 score of 82.2% for accurately classifying semantic relationships between these entities at epoch 1,009. The taxonomy generated from these models effectively identified emerging skills and occupations, offering insights into future workforce requirements. Visualizations of the taxonomy were created using various graph structures, demonstrating its applicability across multiple sectors. The results indicate that this system can dynamically update and adapt to changes in skill demand over time.</p><p><strong>Discussion: </strong>The dynamic taxonomy model not only provides real-time updates on current competencies but also predicts emerging skill trends, offering a valuable tool for workforce planning. The high recall rates in NER suggest strong entity recognition capabilities, though precision improvements are needed to reduce false positives. Limitations include the need for a larger corpus and sector-specific models. Future work will focus on expanding the corpus, improving model accuracy, and incorporating expert feedback to further refine the taxonomy.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1579998"},"PeriodicalIF":3.0,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12263674/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144650752","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ethical-legal implications of AI-powered healthcare in critical perspective.","authors":"Mohammad Nasir, Kaif Siddiqui, Samreen Ahmed","doi":"10.3389/frai.2025.1619463","DOIUrl":"10.3389/frai.2025.1619463","url":null,"abstract":"<p><p>The increasing utilization of Artificial Intelligence (AI) systems in the field of healthcare, from diagnosis to medical decision making and patient care, necessitates identification of its potential benefits, risks and challenges. This requires an appraisal of AI use from a legal and ethical perspective. A review of the existing literature on AI in healthcare available on PubMed, Oxford Academic and Scopus revealed several common concerns regarding the relationship between AI, ethics, and healthcare-(i) the question of data: the choices inherent in collection, analysis, interpretation, and deployment of data inputted to and outputted by AI systems; (ii) the challenges to traditional patient-doctor relationships and long-held assumptions about privacy, identity and autonomy, as well as to the functioning of healthcare institutions. The potential benefits of AI's application need to be balanced against the legal-ethical issues emanating from its use-bias, consent, access, privacy and cost-to guard against detrimental effects of uncritical AI use. The authors suggest that a legal framework for AI should adopt a critical and grounded perspective-cognizant of the material political realities of AI and its wider impact on more marginalized communities. The largescale utilization of health datasets often without consent, responsibility or accountability, further necessitates regulation in the field of technology design, given the entwined nature of AI research with advancements in wearables and sensor technology. Taking into account the 'superhuman' and 'subhuman' traits of AI, regulation should aim to encourage the development of AI systems that augment rather than outrightly replace human effort.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1619463"},"PeriodicalIF":3.0,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12263600/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144650753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A fuzzy system for detection of road slipperiness in Arctic snowy conditions using LiDAR.","authors":"Aqsa Rahim, Sushmit Dhar, Fuqing Yuan, Javad Barabady","doi":"10.3389/frai.2025.1600174","DOIUrl":"10.3389/frai.2025.1600174","url":null,"abstract":"<p><p>The advancement of self-driving cars has significantly improved transportation by enhancing safety, efficiency, and mobility. However, their operation in Arctic environments remains challenging due to snow, ice, and slush, which negatively impact traction and road surface perception. To address these challenges, this study integrates LiDAR-based reflected intensity measurements with environmental parameters such as humidity, temperature, and the coefficient of friction to detect road surface slipperiness and roughness. A Fuzzy Logic System is developed to process these features and classify the slipperiness levels. The analysis establishes a strong correlation between LiDAR intensity and the coefficient of friction, enabling reliable detection of surface conditions. The proposed method achieves a testing accuracy of 87% in classifying road slipperiness under Arctic conditions. These findings demonstrate the effectiveness of LiDAR and sensor fusion for real-time road condition monitoring and highlight their potential in enhancing the safety and performance of autonomous vehicles in extreme weather environments.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1600174"},"PeriodicalIF":3.0,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12263566/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144650750","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting the risk of depression in older adults with disability using machine learning: an analysis based on CHARLS data.","authors":"Tongtong Jin, Ayitijiang Halili","doi":"10.3389/frai.2025.1624171","DOIUrl":"10.3389/frai.2025.1624171","url":null,"abstract":"<p><strong>Background: </strong>The advancement of artificial intelligence technologies has opened new avenues for depression prevention and management in older adults with disability (defined by basic or instrumental activities of daily living, BADL/IADL). This study systematically developed machine learning (ML) models to predict depression risk in disabled elderly individuals using longitudinal data from the China Health and Retirement Longitudinal Study (CHARLS), providing a potentially generalizable tool for early screening.</p><p><strong>Methods: </strong>This study utilized longitudinal data from the CHARLS 2011-2015 cohort. A three-stage serial consensus approach feature selection framework (LASSO, Elastic Net, and Boruta) was employed to identify 21 robust predictors from 74 candidate variables. Ten ML algorithms were evaluated: LR, HistGBM, MLP, XGBoost, bagging, DT, LightGBM, RF, SVM, and CatBoost. Temporal external validation was performed using an independent 2018-2020 cohort to assess model generalizability. Performance was comprehensively evaluated using accuracy, AUC, F1-score, precision, and recall metrics. The SHAP framework was employed to interpret feature contribution mechanisms.</p><p><strong>Results: </strong>Results demonstrated that the HistGBM model achieved optimal overall performance on the testing sets (AUC = 0.779, F1-score = 0.735, accuracy = 0.713), with only an 8.5% AUC difference between training and testing sets and a 10% difference between external validation and testing sets, indicating temporal stability. SHAP interpretability analysis revealed that sleep time (mean SHAP value = 0.344) in the health behavior domain and life satisfaction (0.339) and episodic memory (0.220) in the subjective perception domain contributed more significantly to prediction than traditional biomedical indicators.</p><p><strong>Conclusion: </strong>This study developed an AI-based tool for depression risk assessment in older adults with disability through a multi-stage feature selection process and a temporal external validation framework. These findings provide a practical screening instrument and a methodological reference for implementing AI technologies in geriatric mental health applications, thereby facilitating clinical translation of predictive analytics in this field.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1624171"},"PeriodicalIF":3.0,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12263909/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144650754","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI in humanitarian healthcare: a game changer for crisis response.","authors":"Diala Haykal, Mohamad Goldust, Hugues Cartier, Patrick Treacy","doi":"10.3389/frai.2025.1627773","DOIUrl":"10.3389/frai.2025.1627773","url":null,"abstract":"<p><p>Artificial Intelligence (AI) is transforming humanitarian healthcare by providing innovative solutions to critical challenges in crisis response. This review explores peer-reviewed literature and case reports from 2001 to 2025, retrieved from PubMed, Scopus, and Google Scholar, using targeted keywords. Results indicate that AI enhances disaster prediction, disease surveillance, resource allocation, and mental health support through tools such as machine learning, natural language processing, robotics, and blockchain. Prominent applications include AI-powered early warning systems, chatbots for displaced populations, telemedicine platforms, and automated supply chain logistics. Ethical concerns such as data privacy, bias, and access inequities remain critical to responsible deployment. By uniting governments, NGOs, and technology providers, AI serves as a powerful tool to strengthen humanitarian healthcare systems, enhancing resilience and efficiency while ensuring better outcomes for vulnerable populations during crises.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1627773"},"PeriodicalIF":3.0,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12263676/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144650751","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Navigating ethical minefields: a multi-stakeholder approach to assessing interconnected risks in generative AI using grey DEMATEL.","authors":"Sridhar Jonnala, Nisha Mary Thomas, Sarthak Mishra","doi":"10.3389/frai.2025.1611024","DOIUrl":"10.3389/frai.2025.1611024","url":null,"abstract":"<p><p>The rapid advancement of generative artificial intelligence (AI) technologies has introduced unprecedented capabilities in content creation and human-AI interaction, while simultaneously raising significant ethical concerns. This study examined the complex landscape of ethical risks associated with generative AI (GAI) through a novel multi-stakeholder empirical analysis using the grey decision-making-trial-and-evaluation-laboratory methodology to quantitatively analyze the causal relationships between risks and their relative influence on AI deployment outcomes. Through a comprehensive literature review and expert validation across three key stakeholder groups (AI developers, end users, and policymakers), we identified and analyzed 14 critical ethical challenges across the input, training, and output modules, including both traditional and emerging risks, such as deepfakes, intellectual property rights, data transparency, and algorithmic bias. This study analyzed the perspectives of key stakeholders to understand how ethical risks are perceived, prioritized, and interconnected in practice. Using Euclidean-distance analysis, we identified significant divergences in risk perception among stakeholders, particularly in areas of adversarial prompts, data bias, and output bias. Our findings contribute to the development of a balanced ethical risk framework by categorizing risks into four distinct zones: critical enablers, mild enablers, independent enablers, and critical dependents. This categorization promotes technological advancement and responsible AI deployment. This study addressed the current gaps in academic work by providing actionable recommendations for risk-mitigation strategies and policy development while highlighting the need for collaborative approaches among stakeholders in the rapidly evolving field of GAI.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1611024"},"PeriodicalIF":3.0,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12261451/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144643758","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Contestable AI for criminal intelligence analysis: improving decision-making through semantic modeling and human oversight.","authors":"Falk Maoro, Michaela Geierhos","doi":"10.3389/frai.2025.1602998","DOIUrl":"10.3389/frai.2025.1602998","url":null,"abstract":"<p><p>Criminal investigation analysis involves processing large amounts of data, making manual analysis impractical. Artificial intelligence (AI)-driven information extraction systems can assist investigators in handling this data, leading to significant improvements in effectiveness and efficiency. However, the use of AI in criminal investigations also poses significant risks to individuals, requiring the integration of contestability into systems and processes. To meet this challenge, contestability requirements must be tailored to specific contexts. In this work, we analyzed and adapted existing requirements for criminal investigation analysis, focusing on the retrospective analysis of police reports. For this purpose, we introduced a novel information extraction pipeline based on three language modeling tasks, which we refer to as semantic modeling. Building on this concept, we evaluated contestability requirements and integrated them into our system. As a proof of concept, we developed an AI-driven information extraction system that incorporates contestability features and provides multiple functionalities for data analysis. Our findings highlight three key perspectives essential for contestability in AI-driven investigations: information provision, interactive controls, and quality assurance. This work contributes to the development of more transparent, accountable, and adaptable AI systems for law enforcement applications.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1602998"},"PeriodicalIF":3.0,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12259644/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144643757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploring the role of generative AI in international students' sociocultural adaptation: a cognitive-affective model.","authors":"Huajun Ma, Qingnan You, Zhiyuan Jin, Xinglin Liu, Zimeng Chen","doi":"10.3389/frai.2025.1615113","DOIUrl":"10.3389/frai.2025.1615113","url":null,"abstract":"<p><p>Against the backdrop of increasing global educational exchanges, the sociocultural adaptation of international students has attracted significant attention. The rise of Generative Artificial Intelligence has brought new perspectives to research in this field, yet existing studies have insufficiently explored the mechanisms through which GenAI influences the sociocultural adaptation of international students. Drawing on the cognitive-affective personality system theory and conservation of resources theory, this study employed a three-stage time-lagged questionnaire survey to collect 329 valid responses from international students at three universities in North, South, and East China. The research aims to investigate how GenAI use impacts students' sociocultural adaptation, while examining the mediating roles of positive reappraisal and perceived empathy, as well as the moderating effect of AI anthropomorphism. The findings reveal that GenAI use is significantly positively associated with international students' sociocultural adaptation. Positive reappraisal and users' subjective perceived empathy mediate the relationship between GenAI use and sociocultural adaptation. Additionally, the degree of AI anthropomorphism positively moderates the relationships between GenAI use and both positive reappraisal and perceived empathy, enhancing the indirect effects of these mediating variables on the relationship between GenAI use and sociocultural adaptation. This study enriches the technological premises of cross-cultural adaptation for international students and provides GenAI-based intervention strategies for their educational management.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1615113"},"PeriodicalIF":3.0,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12256517/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144638297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A nnU-Net-based automatic segmentation of FCD type II lesions in 3D FLAIR MRI images.","authors":"Shubham Joshi, Millie Pant, Arnav Malhotra, Kusum Deep, Vaclav Snasel","doi":"10.3389/frai.2025.1601815","DOIUrl":"10.3389/frai.2025.1601815","url":null,"abstract":"<p><p>Focal cortical dysplasia (FCD) type II is a common cause of epilepsy and is challenging to detect due to its similarities with other brain conditions. Finding these lesions accurately is essential for successful surgery and seizure control. Manual detection is slow and challenging because the MRI features are subtle. Deep learning, especially convolutional neural networks, has shown great potential in automating image classification and segmentation by learning and extracting features. The nnU-Net framework is known for its ability to adapt its settings, including preprocessing, network design, training, and post-processing, to any new medical imaging task. This study employs an automated slice selection approach that ranks axial FLAIR slices by their peak voxel intensity and retains the five highest-ranked slices per scan, thereby focusing the network on lesion-rich slices and uses nnU-Net to automate the segmentation of FCD type II lesions on 3D FLAIR MRI images. The study was conducted on 85 FCD type II subjects and results are evaluated through 5-fold cross-validation. Using nnU-Net's flexible and robust design, this study aims to improve the accuracy and speed of lesion detection, helping with better presurgical evaluations and outcomes for epilepsy patients.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1601815"},"PeriodicalIF":3.0,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12247529/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144627309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Data stream-pairwise bottleneck transformer for engagement estimation from video conversation.","authors":"Keita Suzuki, Nobukatsu Hojo, Kazutoshi Shinoda, Saki Mizuno, Ryo Masumura","doi":"10.3389/frai.2025.1516295","DOIUrl":"10.3389/frai.2025.1516295","url":null,"abstract":"<p><p>This study aims to assess participant engagement in multiparty conversations using video and audio data. For this task, the interaction among numerous data streams, such as video and audio from multiple participants, should be modeled effectively, considering the redundancy of video and audio across frames. To efficiently model participant interactions while accounting for such redundancy, a previous study proposed inputting participant feature sequences into global token-based transformers, which constrain attention across feature sequences to pass through only a small set of internal units, allowing the model to focus on key information. However, this approach still faces the challenge of redundancy in participant-feature estimation based on standard cross-attention transformers, which can connect all frames across different modalities. To address this, we propose a joint model for interactions among all data streams using global token-based transformers, without distinguishing between cross-modal and cross-participant interactions. Experiments on the RoomReader corpus confirm that the proposed model outperforms previous models, achieving accuracy ranging from 0.720 to 0.763, weighted F1 scores from 0.733 to 0.771, and macro F1 scores from 0.236 to 0.277.</p>","PeriodicalId":33315,"journal":{"name":"Frontiers in Artificial Intelligence","volume":"8 ","pages":"1516295"},"PeriodicalIF":3.0,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12246976/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144627310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}