The increase of digital ergonomic tools has fundamentally changed how workstation data is collected, generating rich, high-volume datasets that require expert interpretation to produce actionable recommendations. This paper proposes and evaluates an LLM-based text generation layer as a bridge between the structured quantitative output of such tools and the narrative ergonomic assessment reports required by practitioners. A controlled experiment compares six LLMs—GPT-4, Mistral, and Qwen (cloud-based) and Mistral Nemo, Qwen, and LLaMA (locally deployed via LM Studio)—under three prompt engineering strategies against six quality criteria evaluated by a four-rater panel. The online vs. offline deployment comparison is central given industrial data confidentiality requirements. A Friedman test reached significance under Fine-tuned conditions, and online models significantly outperformed offline counterparts across all prompt types. Mistral Nemo (offline) approached GPT-4 performance under enriched prompting, demonstrating the viability of privacy-compliant local deployment. The paper further discusses integration pathways with digital ergonomic tools and outlines a research agenda for LLM-driven report automation.

Despite advances in Artificial Intelligence (AI) and its significant potential to improve clinical care, its real-world implementation remains limited. Low acceptance among healthcare professionals (HCPs) is a significant contributing factor to this limitation, making it essential to understand their views. To explore HCPs’ perspectives on the use of AI in clinical decision-making, we conducted semi-structured interviews (SSIs) with 14 clinicians across different roles and levels of exposure to AI in five countries. Thematic analysis of the interviews revealed that clinicians’ perspectives can be described through three main themes: motivations for AI use, factors contributing to hesitation, and facilitating conditions. While clinicians recognize AI’s potential to support their work, our findings revealed that the use of AI in clinical practice is not simply accepted or rejected. Acceptance is conditional and negotiated through an interplay among motivations, hesitations, and facilitating conditions. This paper provides a conceptual understanding of the acceptance of AI in clinical decision-making grounded in clinicians’ perspectives.

The systematic integration of subjective patient experiences into health product development remains challenging. Narrative experience reports are context-dependent, emotionally embedded, and often unstructured, while existing approaches lack transparent mechanisms for translating experiential knowledge into development-relevant engineering objectives. This paper presents a rule-based, multi-step framework for translating narrative patient experience into engineering objectives under explicit methodological constraints using large language models (LLMs). The framework decomposes the translation process into five sequential stages, segmentation, dimensional structuring, criticality identification, tension formulation, and abstraction, thereby enabling controlled interpretive reduction while preserving traceability to the original narrative input. The framework was evaluated in a proof-of-concept setting using three heterogeneous experience reports, each analyzed with three different LLMs. The results indicate that constrained prompting supports transparent and contextually grounded outputs without evident hallucinations. However, model-dependent variations in abstraction level and solution neutrality highlight the necessity of methodological governance and human oversight. The contribution lies in formalizing a traceable translation logic that systematically bridges experiential patient knowledge and engineering-relevant objective-level representations.

Blockchain and non-fungible tokens (NFTs) create immutable records of digital ownership, but they also introduce complex governance challenges. This paper explores how Artificial Intelligence (AI) can conceptually support governance, transparency, and rights assurance in blockchain-based ownership systems. Rather than focusing on performance or optimization, we treat AI as a tool for modeling governance structures, enforcing policies, verifying ownership, monitoring compliance, and enabling auditable transparency. We first review foundational concepts of blockchain, NFTs, and decentralized governance. We then examine specific roles for AI in ownership governance, including how machine learning and automated reasoning can model complex policies and detect violations. Ethical and regulatory considerations are discussed, highlighting issues of privacy, fairness, accountability, and legal compliance. Finally, we propose a theoretical framework in which AI modules interact with blockchain components to enhance digital-rights assurance without compromising decentralization or user autonomy.