{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,5]],"date-time":"2026-05-05T03:00:02Z","timestamp":1777950002245,"version":"3.51.4"},"reference-count":80,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T00:00:00Z","timestamp":1777852800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>We address moral uncertainty in reinforcement learning (RL) by proposing a framework that integrates multiple ethical theories into decision-making. Existing approaches rely on single moral frameworks or handcrafted rewards, limiting scalability and failing to capture moral pluralism. We introduce AMULED, a task-agnostic ethical layer that refines a pre-trained RL agent using large language models (LLMs) to provide multi-perspective moral feedback.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Following initial training, the RL model is fine-tuned using LLM-generated feedback in place of human feedback. Five moral clusters\u2014consequentialist, deontological, virtue, care, and social justice\u2014assign belief values to candidate actions. These beliefs are aggregated using Belief Jensen\u2013Shannon Divergence and Dempster\u2013Shafer Theory to produce probability scores that serve as shaping rewards, while a KL-regularization term constrains deviation from the base policy. The framework is evaluated across two environments (Finding Milk and Driving and Rescuing), multiple LLM backbones, and alternative belief aggregation methods, with 50-run replicates.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>AMULED improves ethical behavior without substantially degrading task performance. In Finding Milk, it increases desirable actions (63.1% more crying babies attended) and reduces undesirable actions (60.3% fewer sleeping babies disturbed), with only a 5.1% increase in path length. In Driving and Rescuing, it balances competing objectives more effectively than baselines, rescuing 38.4% more targets than human-feedback agents while maintaining lower collision rates and reduced policy degradation. Across experiments, BJSD-DST aggregation outperforms standard methods (e.g., voting, averaging) in handling conflicting moral signals and achieves the best overall performance on most metrics.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>AMULED operationalizes moral pluralism through scalable, LLM-based feedback and provides a principled mechanism for resolving conflicting ethical signals. The framework demonstrates robustness across tasks and model variants, though performance depends on LLM reasoning quality and can degrade in spatially complex settings. These results suggest that LLM-driven belief aggregation offers a practical alternative to handcrafted rewards and human supervision for ethical decision-making in RL.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/frai.2026.1754973","type":"journal-article","created":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T05:31:36Z","timestamp":1777872696000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["AMULED: Addressing Moral Uncertainty using Large language models for Ethical Decision-making"],"prefix":"10.3389","volume":"9","author":[{"given":"Rohit K.","family":"Dubey","sequence":"first","affiliation":[{"name":"Computational Social Science","place":["ETH Z\u00fcrich, Zurich, Switzerland"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Damian","family":"Dailisan","sequence":"additional","affiliation":[{"name":"Computational Social Science","place":["ETH Z\u00fcrich, Zurich, Switzerland"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Sachit","family":"Mahajan","sequence":"additional","affiliation":[{"name":"Computational Social Science","place":["ETH Z\u00fcrich, Zurich, Switzerland"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2026,5,4]]},"reference":[{"key":"B1","article-title":"\u201cReinforcement learning as a framework for ethical decision making,\u201d","volume-title":"Workshops at the Thirtieth AAAI Conference on Artificial Intelligence","author":"Abel","year":"2016"},{"key":"B2","first-page":"22","article-title":"\u201cConstrained policy optimization,\u201d","volume-title":"Proceedings of the 34th International Conference on Machine Learning","author":"Achiam","year":"2017"},{"key":"B3","article-title":"\u201cDeontological ethics,\u201d","author":"Alexander","year":"2024","journal-title":"The Stanford Encyclopedia of Philosophy"},{"key":"B4","doi-asserted-by":"publisher","first-page":"251","DOI":"10.1080\/09528130050111428","article-title":"Prolegomena to any future artificial moral agent","volume":"12","author":"Allen","year":"2000","journal-title":"J. Exp. Theoret. Artif. Intellig"},{"key":"B5","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v32i1.11797","article-title":"\u201cSafe reinforcement learning via shielding,\u201d","author":"Alshiekh","year":"2018","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9780511978036","author":"Anderson","year":"2011","journal-title":"Machine ethics"},{"key":"B7","doi-asserted-by":"publisher","first-page":"679","DOI":"10.1512\/iumj.1957.6.56038","article-title":"A markovian decision process","volume":"6","author":"Bellman","year":"1957","journal-title":"J. Math. Mech"},{"key":"B8","doi-asserted-by":"publisher","first-page":"103239","DOI":"10.1016\/j.artint.2020.103239","article-title":"Ethical approaches and autonomous systems","volume":"281","author":"Bench-Capon","year":"2020","journal-title":"Artif. Intellig"},{"key":"B9","first-page":"908","article-title":"\u201cSafe model-based reinforcement learning with stability guarantees,\u201d","author":"Berkenkamp","year":"2017","journal-title":"International Conference on Neural Information Processing Systems"},{"key":"B10","doi-asserted-by":"publisher","first-page":"38","DOI":"10.1109\/MIS.2006.82","article-title":"Toward a general logicist methodology for engineering ethically correct robots","volume":"21","author":"Bringsjord","year":"2006","journal-title":"IEEE Intellig. Syst"},{"key":"B11","first-page":"437","article-title":"\u201cAI alignment and human reward,\u201d","volume-title":"Proceedings of the 2021 AAAI\/ACM Conference on AI, Ethics, and Society, AIES\u201321","author":"Butlin","year":"2021"},{"key":"B12","doi-asserted-by":"publisher","first-page":"629","DOI":"10.1038\/s43588-024-00695-4","article-title":"The lost data: how AI systems censor LGBTQ+ content in the name of safety","volume":"4","author":"Chen","year":"2024","journal-title":"Nat. Comput. Sci"},{"key":"B13","doi-asserted-by":"publisher","first-page":"16483","DOI":"10.1007\/s00500-023-09112-w","article-title":"A novel combination rule for conflict management in data fusion","volume":"27","author":"Chen","year":"2023","journal-title":"Soft Comput"},{"key":"B14","first-page":"809","article-title":"\u201cLyapunov-based safe policy optimization for continuous control,\u201d","volume-title":"Proceedings of the 35th International Conference on Machine Learning","author":"Chow","year":"2018"},{"key":"B15","article-title":"\u201cDeep reinforcement learning from human preferences,\u201d","author":"Christiano","year":"2017","journal-title":"International Conference on Neural Information Processing Systems"},{"key":"B16","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s11948-025-00529-0","article-title":"LLMs, truth, and democracy: an overview of risks","volume":"31","author":"Coeckelbergh","year":"2025","journal-title":"Sci. Eng. Ethics"},{"key":"B17","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v31i1.11140","article-title":"\u201cMoral decision making frameworks for artificial intelligence,\u201d","author":"Conitzer","year":"2017","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"B18","doi-asserted-by":"publisher","first-page":"62","DOI":"10.1109\/MIS.2014.87","article-title":"Man versus machine or man+ machine?","volume":"29","author":"Cummings","year":"2014","journal-title":"IEEE Intellig. Syst"},{"key":"B19","doi-asserted-by":"crossref","first-page":"57","DOI":"10.1007\/978-3-540-44792-4_3","article-title":"\u201cUpper and lower probabilities induced by a multivalued mapping,\u201d","volume-title":"Classic Works of the Dempster-Shafer Theory of Belief Functions","author":"Dempster","year":"2008"},{"key":"B20","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.robot.2015.11.012","article-title":"Formal verification of ethical choices in autonomous systems","volume":"77","author":"Dennis","year":"2016","journal-title":"Robot. Autonom. Syst"},{"key":"B21","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s10676-018-9450-z","article-title":"Ethics in artificial intelligence: introduction to the special issue","volume":"20","author":"Dignum","year":"2018","journal-title":"Ethics Inform. Technol"},{"key":"B22","doi-asserted-by":"publisher","first-page":"63","DOI":"10.1016\/j.ijinfomgt.2019.01.021","article-title":"Artificial intelligence for decision making in the era of big data-evolution, challenges and research agenda","volume":"48","author":"Duan","year":"2019","journal-title":"Int. J. Inform. Managem"},{"key":"B23","article-title":"\u201cReinforcement learning under moral uncertainty,\u201d","author":"Ecoffet","year":"2020","journal-title":"International Conference on Machine Learning"},{"key":"B24","doi-asserted-by":"publisher","first-page":"13080","DOI":"10.1038\/s41598-019-49411-7","article-title":"Human decision-making biases in the moral dilemmas of autonomous vehicles","volume":"9","author":"Frank","year":"2019","journal-title":"Scient. Rep"},{"key":"B25","doi-asserted-by":"publisher","first-page":"1437","DOI":"10.5555\/2789272.2886795","article-title":"The moral turing test: evaluating human-LLM alignment in moral decision-making","volume":"16","author":"Garcia","year":"2024","journal-title":"J. Mach. Learn. Res."},{"key":"B26","first-page":"1437","article-title":"A comprehensive survey on safe reinforcement learning","volume":"16","author":"Garc\u00eda","year":"2015","journal-title":"J. Mach. Learn. Res"},{"key":"B27","doi-asserted-by":"publisher","first-page":"89","DOI":"10.1111\/josp.12050","article-title":"Moral injury and the ethic of care: reframing the conversation about differences","volume":"45","author":"Gilligan","year":"2014","journal-title":"J. Soc. Philosophy"},{"key":"B28","doi-asserted-by":"publisher","first-page":"244","DOI":"10.1017\/CBO9780511978036.019","article-title":"Towards the ethical robot","volume":"1","author":"Gips","year":"2011","journal-title":"Mach. Ethics"},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2205.10330","article-title":"A review of safe reinforcement learning: Methods, theory and applications","author":"Gu","year":"2022","journal-title":"arXiv"},{"key":"B30","doi-asserted-by":"publisher","first-page":"2285","DOI":"10.1098\/rsta.2024.0100","article-title":"Large language models (LLMs) as agents for augmented democracy","volume":"382","author":"Gudi no-Rosero","year":"2024","journal-title":"Philos. T. R. Soc. A: Mathem. Phys. Eng. Sci"},{"key":"B31","first-page":"220","article-title":"\u201cThe off-switch game,\u201d","volume-title":"Workshops at the Thirty-First AAAI Conference on Artificial Intelligence","author":"Hadfield-Menell","year":"2017"},{"key":"B32","doi-asserted-by":"publisher","first-page":"833","DOI":"10.1038\/s43588-023-00527-x","article-title":"Human-like intuitive behavior and reasoning biases emerged in large language models but disappeared in ChatGPT","volume":"3","author":"Hagendorff","year":"2023","journal-title":"Nat. Comput. Sci"},{"key":"B33","doi-asserted-by":"crossref","first-page":"285","DOI":"10.1007\/978-3-030-62330-2_14","article-title":"\u201cSummary: What's wrong with AI? Humanistic technology needed!,\u201d","volume-title":"Next Civilization: Digital Democracy and Socio-Ecological Finance-How to Avoid Dystopia and Upgrade Society by Digital Means","author":"Helbing","year":"2021"},{"key":"B34","doi-asserted-by":"publisher","first-page":"15","DOI":"10.1007\/s10676-024-09756-8","article-title":"Why converging technologies need converging international regulation","volume":"26","author":"Helbing","year":"2024","journal-title":"Ethics Inform. Technol"},{"key":"B35","first-page":"1","article-title":"CleanRL: high-quality single-file implementations of deep reinforcement learning algorithms","volume":"23","author":"Huang","year":"2022","journal-title":"J. Mach. Learn. Res"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.1007\/s43681-025-00814-5","article-title":"Navigating LLM ethics: advancements, challenges, and future directions","author":"Jiao","year":"2024","journal-title":"arXiv"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.1016\/j.trc.2024.104511.","article-title":"Democratizing traffic control in smart cities","author":"Korecki","year":"","journal-title":"Transportation Research Part C: Emerging Technologies"},{"key":"B38","doi-asserted-by":"publisher","first-page":"705164","DOI":"10.3389\/frai.2021.705164","article-title":"Ai systems and respect for human autonomy","volume":"4","author":"Laitinen","year":"2021","journal-title":"Front. Artif. Intellig"},{"key":"B39","unstructured":"Lambert\n              N.\n            \n            \n              Castricato\n              L.\n            \n            \n              von Werra\n              L.\n            \n            \n              Havrilla\n              A.\n            \n          \n          Illustrating reinforcement learning from human feedback (rlhf)\n          \n          2022"},{"key":"B40","year":""},{"key":"B41","doi-asserted-by":"crossref","DOI":"10.1201\/9781315366951","volume-title":"Artificial Intelligence with Uncertainty","author":"Li","year":"2017"},{"key":"B42","doi-asserted-by":"publisher","first-page":"293","DOI":"10.1007\/BF00992699","article-title":"Self-improving reactive agents based on reinforcement learning, planning and teaching","volume":"8","author":"Lin","year":"1992","journal-title":"Mach. Learn."},{"key":"B43","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1801.03160","article-title":"A formalization of kant's second formulation of the categorical imperative","author":"Lindner","year":"2018","journal-title":"arXiv"},{"key":"B44","doi-asserted-by":"crossref","DOI":"10.4337\/9781788972994","volume-title":"The Ethics of Competition: How a Competitive Society is Good for All","author":"L\u00fctge","year":"2019"},{"key":"B45","doi-asserted-by":"crossref","DOI":"10.1093\/oso\/9780198722274.001.0001","volume-title":"Moral Uncertainty","author":"MacAskill","year":"2020"},{"key":"B46","volume-title":"Normative Uncertainty","author":"MacAskill","year":"2014"},{"key":"B47","doi-asserted-by":"publisher","first-page":"1619029","DOI":"10.3389\/frai.2025.1619029","article-title":"Ethical theories, governance models, and strategic frameworks for responsible AI adoption and organizational success","volume":"8","author":"Madanchian","year":"2025","journal-title":"Front. Artif. Intellig"},{"key":"B48","doi-asserted-by":"publisher","first-page":"3953","DOI":"10.1007\/s00146-024-01968-2","article-title":"The executioner paradox: understanding self-referential dilemma in computational systems","volume":"40","author":"Mahajan","year":"2024","journal-title":"AI Society"},{"key":"B49","doi-asserted-by":"crossref","first-page":"1674","DOI":"10.1609\/aies.v8i2.36665","article-title":"\u201cMapping moral reasoning in llms: A multi-dimensional analysis of safety principle conflicts,\u201d","volume":"8","author":"Mahajan","year":"2025","journal-title":"Proceedings of the AAAI\/ACM Conference on AI, Ethics, and Society"},{"key":"B50","doi-asserted-by":"publisher","first-page":"243","DOI":"10.1007\/s10676-015-9367-8","article-title":"Integrating robot ethics and machine morality: the study and design of moral competence in robots","volume":"18","author":"Malle","year":"2016","journal-title":"Ethics Inform. Technol"},{"key":"B51","doi-asserted-by":"publisher","first-page":"335","DOI":"10.1162\/0899766053011528","article-title":"Robust reinforcement learning","volume":"17","author":"Morimoto","year":"2005","journal-title":"Neural Comput"},{"key":"B52","article-title":"\u201cAlgorithms for inverse reinforcement learning,\u201d","author":"Ng","year":"2000","journal-title":"International Conference on Machine Learning"},{"key":"B53","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2304.03442","author":"Park","year":"2023","journal-title":"Generative Agents: Interactive Simulacra of Human Behavior"},{"key":"B54","doi-asserted-by":"crossref","DOI":"10.1093\/acprof:oso\/9780195136364.001.0001","author":"Pogge","year":"2007","journal-title":"John Rawls: His Life and Theory of Justice"},{"key":"B55","doi-asserted-by":"publisher","first-page":"46","DOI":"10.1109\/MIS.2006.77","article-title":"Prospects for a kantian machine","volume":"21","author":"Powers","year":"2006","journal-title":"IEEE Intellig. Syst"},{"key":"B56","doi-asserted-by":"publisher","first-page":"1377011","DOI":"10.3389\/frai.2024.1377011","article-title":"Ethics and responsible ai deployment","volume":"7","author":"Radanliev","year":"2024","journal-title":"Front. Artif. Intellig"},{"key":"B57","doi-asserted-by":"publisher","first-page":"632394","DOI":"10.3389\/frobt.2021.632394","article-title":"A deeper look at autonomous vehicle ethics: an integrative ethical decision-making framework to explain moral pluralism","volume":"8","author":"Rhim","year":"2021","journal-title":"Front. Robot. AI"},{"key":"B58","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1609\/aimag.v36i4.2577","article-title":"Research priorities for robust and beneficial artificial intelligence","volume":"36","author":"Russell","year":"2015","journal-title":"AI Magazine"},{"key":"B59","doi-asserted-by":"publisher","first-page":"15","DOI":"10.3389\/frobt.2018.00015","article-title":"Meaningful human control over autonomous systems: A philosophical account","volume":"5","author":"Santoni de Sio","year":"2018","journal-title":"Front. Robot. AI"},{"key":"B60","first-page":"51778","article-title":"\u201cEvaluating the moral beliefs encoded in LLMs,\u201d","author":"Scherrer","year":"2023","journal-title":"International Conference on Neural Information Processing Systems"},{"key":"B61","doi-asserted-by":"publisher","first-page":"258","DOI":"10.1038\/s42256-022-00458-8","article-title":"Large pre-trained language models contain human-like biases of what is right and wrong to do","volume":"4","author":"Schramowski","year":"2022","journal-title":"Nat. Mach. Intellig"},{"key":"B62","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1707.06347","author":"Schulman","year":"2017","journal-title":"Proximal Policy Optimization Algorithms"},{"key":"B63","volume-title":"Collective Choice and Social Welfare: Expanded Edition","author":"Sen","year":"2017"},{"key":"B64","doi-asserted-by":"publisher","first-page":"641","DOI":"10.1038\/s43588-024-00694-5","article-title":"Defining our future with generative AI","volume":"4","author":"Suri","year":"2024","journal-title":"Nat. Comput. Sci"},{"key":"B65","volume-title":"Reinforcement Learning: An Introduction","author":"Sutton","year":"2018"},{"key":"B66","first-page":"1651","article-title":"\u201cPolicy gradients with variance related risk criteria,\u201d","author":"Tamar","year":"2012","journal-title":"Proceedings of the 29th International Conference on Machine Learning"},{"key":"B67","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3419633","article-title":"Implementations in machine ethics: A survey","volume":"53","author":"Tolmeijer","year":"2020","journal-title":"ACM Comp. Surv"},{"key":"B68","doi-asserted-by":"publisher","first-page":"253","DOI":"10.1093\/scan\/nsab100","article-title":"Neuro-computational foundations of moral preferences","volume":"17","author":"Ugazio","year":"2021","journal-title":"Soc. Cognit. Affect. Neurosci"},{"key":"B69","doi-asserted-by":"publisher","first-page":"672","DOI":"10.1038\/d41586-023-02980-0","article-title":"Ai and science: what 1,600 researchers think","volume":"621","author":"Van Noorden","year":"2023","journal-title":"Nature"},{"key":"B70","author":"Wallach","year":"2008","journal-title":"Moral Machines: Teaching Robots Right from Wrong"},{"key":"B71","doi-asserted-by":"publisher","first-page":"186345","DOI":"10.1007\/s11704-024-40231-1","article-title":"A survey on large language model based autonomous agents","volume":"18","author":"Wang","year":"2024","journal-title":"Front. Comp. Sci"},{"key":"B72","first-page":"24824","article-title":"\u201cChain-of-thought prompting elicits reasoning in large language models,\u201d","author":"Wei","year":"2022","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B73","article-title":"\u201cLivebench: a challenging, contamination-free LLM benchmark,\u201d","author":"White","year":"2024","journal-title":"International Conference on Learning Representations (ICLR)"},{"key":"B74","first-page":"32","article-title":"\u201cA low-cost ethics shaping approach for designing reinforcement learning agents,\u201d","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence","author":"Wu","year":"2018"},{"key":"B75","year":""},{"key":"B76","doi-asserted-by":"publisher","first-page":"23","DOI":"10.1016\/j.inffus.2018.04.003","article-title":"Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy","volume":"46","author":"Xiao","year":"2019","journal-title":"Inform. Fusion"},{"key":"B77","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.findings-naacl.158","article-title":"Fire: Fact-checking with iterative retrieval and verification","author":"Xie","year":"2024","journal-title":"arXiv"},{"key":"B78","first-page":"1696","article-title":"\u201cLLM voting: Human choices and AI collective decision making,\u201d","volume-title":"Proceedings of the AAAI\/ACM Conference on AI, Ethics, and Society","author":"Yang","year":"2024"},{"key":"B79","doi-asserted-by":"publisher","first-page":"109075","DOI":"10.1016\/j.asoc.2022.109075","article-title":"A survey: optimization and applications of evidence fusion algorithm based on Dempster-Shafer theory","volume":"124","author":"Zhao","year":"2022","journal-title":"Appl. Soft Comp"},{"key":"B80","first-page":"1433","author":"Ziebart","year":"2008","journal-title":"Maximum Entropy Inverse Reinforcement Learning"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1754973\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T05:31:44Z","timestamp":1777872704000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1754973\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,4]]},"references-count":80,"alternative-id":["10.3389\/frai.2026.1754973"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1754973","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,4]]},"article-number":"1754973"}}