Transactions on Machine Intelligence 2821-1693 9 2 2026 06 01 Evidence-Gated Autonomy in Healthcare AI Agents: A Systematic Review of Action Scope, Human-Oversight Effectiveness, and Multi-Step Safety Failures 84 111 248202 10.22034/tmi.2026.248202 EN H. Ghayoumizadeh Associate Professor of Biomedical Engineering, Vali-e-Asr University of Rafsanjan 0000-0002-5390-3938 Kh. Rezaee Department of Biomedical Engineering, Meybod University, Meybod, Iran A. Fayazi Department of Engineering.Vali-e-Asr University of Rafsanjan,Iran O. Rahmani Seryasat Assistant Professor,Faculty of Electrical and Computer Engineering, Shams Gonbad Higher Education Institute, Gorgan, Iran 0000-0002-9289-6128 Journal Article 2025 11 23 Background: Healthcare AI agents increasingly perform multi-step workflows and tool use, but evidence on their operational autonomy, safety, and human oversight remains limited. This review examines whether greater autonomy is supported by empirical safety and oversight evidence. Methods: PubMed/MEDLINE, Scopus, and Web of Science were searched from January 2022 to July 2026. Of 4,096 records, 2,557 remained after deduplication. Fifty priority reports were selected for full-text review; 34 were assessed and 28 met the inclusion criteria. Data on workflow, architecture, autonomy, safety, oversight, and evaluation were narratively synthesized. Results: Among 28 studies, 18 were published in 2026. Thirteen addressed diagnostic or clinical decision support, nine EHR or administrative workflows, four treatment planning or prescribing, and two patient-facing or cross-domain applications. Autonomy levels were L0 in five studies, L1 in 13, L2 in three, L3 in one, and L4 in six; none reached L5. Most L4 systems were evaluated only in sandbox or retrospective settings. Only three studies empirically assessed oversight effectiveness. Conclusions: Current evidence supports tool-using and workflow-capable healthcare agents, but not unrestricted clinical autonomy. Greater autonomy should require stronger evidence of safety, failure containment, and effective human oversight.

https://www.tmachineintelligence.ir/article_248202_3b490b92ec85b3e3cb8258a886ca61e9.pdf

Transactions on Machine Intelligence 2821-1693 9 2 2026 06 01 Childhood Trauma and Marital Infidelity Tendency: A Neuropsychological Model of Avoidant Attachment and EEG Indicators 112 123 252076 10.22034/tmi.2026.252076 EN F. Z. Ghasemifard M.Sc. Student in Clinical Psychology, Birjand University of Medical Sciences, Birjand, Iran. - Researcher, Iliya Pardazesh Shargh Company, R&D Center, resident at Ferdowsi University of Mashhad Science and Technology Park, Mashhad, Iran. Email: iliyacore@gmail.com 0009-0004-4912-5522 F. Norouzi Kakhi MSc Student in Clinical Psychology, Kashmar Branch, Islamic Azad University, Kashmar, Iran. 0009-0003-5447-3663 Journal Article 2025 12 17 Objective: Extramarital infidelity is a multidimensional phenomenon shaped by psychological and neurobiological factors. This study aimed to investigate the relationship between adverse childhood experiences, attachment styles, and the tendency toward extramarital infidelity in Iranian men, with an emphasis on neurophysiological markers of brain activity recorded via electroencephalography (EEG) within a brain–computer interface (BCI) framework. By integrating psychological and neural indicators, this study seeks to provide a deeper understanding of the emotional decision-making processes underlying extramarital relationships. Method: The present study employed a descriptive-correlational design with a quantitative approach. The study population consisted of Iranian men with a history of or tendency toward extramarital relationships, from whom 55 participants were purposively selected as the sample. Adverse childhood experiences were assessed using the Adverse Childhood Experiences (ACE) scale, and attachment styles were evaluated using the Attachment Styles Questionnaire. Furthermore, participants' brain activity was recorded via EEG during tasks involving emotional decision-making in romantic relationships, and patterns of brain activity were analyzed within the context of BCI-related frameworks. Data were analyzed using descriptive statistics and correlation analyses. Findings: The results indicated that higher levels of adverse childhood experiences were associated with an increased tendency toward extramarital infidelity. Insecure attachment styles, particularly avoidant and anxious attachment, also showed a positive correlation with this tendency. EEG data analysis revealed patterns of brain activity related to emotional processing and affective decision-making that were more pronounced in individuals with greater adverse experiences and insecure attachment styles. Conclusion: The findings of this study suggest that adverse childhood experiences and attachment styles, together with brain activity patterns, may play a role in shaping the tendency toward extramarital infidelity. Integrating psychological data with neurophysiological indicators within a BCI framework can contribute to a deeper understanding of the mechanisms underlying emotional decision-making and to the design of more effective therapeutic interventions.

https://www.tmachineintelligence.ir/article_252076_dddb7e98f07463f1780bd15d77f37d8a.pdf

Transactions on Machine Intelligence 2821-1693 9 2 2026 06 01 A Knowledge Graph-Based Neuro-Symbolic Conceptual Architecture for Semantic Learner Modeling in Intelligent Tutoring Systems 124 141 252077 10.22034/tmi.2026.252077 EN M. Bayatani PhD Student, Department of Computer Engineering, Arak Branch, Islamic Azad University, Arak, Iran N. Osati Eraghi Assistant Professor, Department of Computer Engineering, Arak Branch, Islamic Azad University, Arak, Iran 0000-0002-6830-7290 Journal Article 2026 01 27 Intelligent Tutoring Systems (ITSs) require accurate and semantic modeling of learner states to deliver personalized learning experiences. Despite advances in the application of ontologies and knowledge graphs in educational systems, existing approaches are generally based either on symbolic methods, which offer high interpretability but have limited flexibility in handling dynamic data and complex learner interactions, or on data-driven approaches, which are capable of extracting behavioral patterns but often lack sufficient semantic transparency and explainability. This paper proposes a knowledge graph-based neuro-symbolic conceptual architecture for semantic learner modeling in Intelligent Tutoring Systems. In the proposed architecture, the data-driven component is responsible for analyzing learner interactions and extracting patterns related to learning states, whereas the symbolic component employs ontologies and knowledge graphs to organize educational knowledge, maintain semantic consistency, and support interpretable reasoning. The architecture's integration mechanism is designed to use the results of learner data analysis to update and enrich the learner model within the knowledge graph. Subsequently, the symbolic layer employs semantic rules and conceptual relationships to generate adaptive instructional recommendations. The proposed architecture is organized as a five-layer framework, providing a conceptual foundation for developing Intelligent Tutoring Systems with enhanced explainability, semantic consistency, and adaptability. By combining the strengths of data-driven and symbolic approaches, this framework seeks to reduce the gap between learner data analytics and semantic reasoning in intelligent educational environments.

https://www.tmachineintelligence.ir/article_252077_ebe366c14eac998595e9fa23f2a7a278.pdf

Transactions on Machine Intelligence 2821-1693 9 2 2026 06 01 AI-Enabled Anticipatory Capacity Planning for Smart Government: Integrating Artificial Intelligence, Strategic Foresight, and Queueing Theory under Future Demand Uncertainty 142 154 252085 10.22034/tmi.2026.252085 EN K. Yeganegi Assistant Professor, Department of Industrial Engineering, Islamic Azad University, Za.C, Zanjan, Iran 0000-0003-0421-6258 M. Mohebbi Ph.D. in Futures Studies, Director of Administrative Reform and Performance Evaluation, Management and Planning Organization of Zanjan Province, Zanjan, Iran 0000-0001-6697-2302 Journal Article 2026 02 09 The future operating environment of smart government will be increasingly influenced by artificial intelligence (AI), rapidly changing citizen expectations, demographic shifts, digital service adoption, regulatory changes, cyber disruptions, public-health emergencies, and other sources of demand uncertainty. In such an environment, conventional reactive capacity planning may be insufficient to maintain service quality and operational resilience. This study develops an AI-enabled anticipatory capacity-planning framework that integrates artificial intelligence, strategic foresight, scenario construction, and queueing theory to support resilient and adaptive smart government services. The proposed framework uses AI as an analytical and decision-support layer for identifying emerging demand patterns, supporting future-demand estimation, and translating alternative future conditions into quantitative arrival-rate assumptions. These assumptions are subsequently evaluated through an M/M/c queueing model to examine their implications for utilization, queue length, and waiting time. Three alternative capacity configurations and four future-demand scenarios are analyzed to illustrate how AI-supported foresight can be connected to quantitative service-system design. The results demonstrate a strongly nonlinear relationship between system utilization and service performance. At an arrival rate of 40 requests per hour and a service rate of 15 requests per hour per channel, increasing capacity from three to four channels reduces average waiting time from approximately 9.57 minutes to 1.14 minutes. Adding a fifth channel further reduces waiting time but results in lower capacity utilization. Scenario stress testing indicates that a four-channel system performs effectively under stable and accelerated digital adoption; however, congestion increases rapidly as demand approaches total service capacity. The findings suggest that AI should not be viewed solely as a technology for automating public services. When integrated with strategic foresight and queueing analysis, AI can contribute to anticipatory governance by supporting early identification of emerging demand pressures, detection of critical capacity thresholds, scenario-based stress testing, and adaptive resource planning. In this framework, queueing theory provides the quantitative mechanism for translating AI-supported future-demand assumptions into measurable operational consequences, while strategic foresight provides the structure for dealing with uncertainty and alternative futures. The study contributes to smart government, futures studies, artificial intelligence, and operations research by proposing a transparent analytical bridge between AI-supported demand intelligence, qualitative future scenarios, and quantitative capacity planning. The framework can assist public-sector decision-makers in moving from reactive resource allocation toward anticipatory, data-informed, and adaptive capacity management.

https://www.tmachineintelligence.ir/article_252085_b46dc4a276f8a619426bbbbba9e990c7.pdf