Human factors engineering in intelligent medical devices: paradigm evolution, core issues and frontier prospects

1 Introduction

Patient safety is a global public health priority. Avoidable risks associated with medical devices can compromise clinical quality and the stability of healthcare systems. Retrospective analyses of adverse events indicate that preventable harm often stems from operational errors, suboptimal interface design, and misaligned clinical workflows, rather than inherent hardware failures (Knisely et al., 2020). Recent regulatory surveillance also suggests that insufficient attention to human factors engineering is frequently associated with device corrections and recalls, highlighting widespread challenges related to usability, workflow integration, and user-centered design (U.S. Food and Drug Administration, 2025).

Advances in artificial intelligence (AI), telemedicine, surgical robotics, and mixed reality (MR) have transformed medical devices from passive tools into semi-autonomous collaborative agents (Fraser et al., 2023; Yip et al., 2023). This transition has introduced new human factors challenges, including potential over-reliance on automated decisions, limited interpretability of algorithms, constrained remote interaction fidelity, and elevated cognitive demands within immersive clinical environments (Elnaggar et al., 2023).

Human factors engineering (HFE), used in this review interchangeably with the term human factors/ergonomics, is formally defined by the International Ergonomics Association (IEA) as “the scientific discipline concerned with the understanding of interactions among humans and other elements of a system, and the profession that applies theory, principles, data and methods to design in order to optimize human wellbeing and overall system performance” (International Ergonomics Association, 2000). This human-centred, system-oriented principle is further formalized for interactive systems by ISO 9241–210:2019 (International Organization for Standardization, 2019) (Ergonomics of human-system interaction—Part 210: Human-centred design for interactive systems), which complements the IEA definition and the manufacturer-facing ANSI/AAMI HE75:2025 guidance adopted later in this review. Targeting intelligent medical devices as the research object of this scoping review, an operational definition of HFE oriented to clinical scenarios is further formulated: it centers on user-centric human–machine interaction analysis, covering usability characteristics, clinicians’ cognitive load, human–AI trust relationships, and cross-agent collaborative workflows.

The core research themes extracted in this work include fluctuations in clinician operational performance, dynamic human–AI trust calibration, cognitive burden induced by remote clinical interaction, and perceptual barriers brought by immersive clinical technologies. Rather than being treated as independent research branches, patient experience and psychological feedback are integrated as key evaluation indicators throughout the analysis, ensuring the whole discussion revolves around real-world clinical operational bottlenecks.

These thematic categories are generated via systematic thematic coding of all included literature in strict accordance with the PRISMA-ScR framework, corresponding to the most frequently discussed human factors pain points of intelligent medical devices in existing research, rather than a complete inventory of all HFE-related topics across all healthcare domains.

HFE and usability engineering are required as core parts of pre-market regulatory evaluation for medical devices in the United States, the European Union, and China. These three jurisdictions mandate usability engineering through distinct instruments: in the United States, via FDA human factors and usability engineering guidance (U.S. Food and Drug Administration, 2016); in the European Union, via the Medical Device Regulation (EU) 2017/745 (MDR), which requires usability-related risk control within the conformity-assessment process (European Parliament and Council, 2017); and in China, via the National Medical Products Administration’s registration review guidance on medical device usability engineering (National Medical Products Administration of China, 2024). Although all three treat usability engineering as a mandatory component of pre-market evaluation, their specific mandatory requirements, implementation workflows and compliance thresholds differ significantly across these regimes. Two core widely applied documents govern medical device human factors practice yet carry distinct functions: IEC 62366–1:2021 stipulates standardized step-by-step usability engineering processes for regulatory submission, while ANSI/AAMI HE75:2009 and its 2025 revision deliver detailed practical human factors design guidance for manufacturers. Together, they form complementary rather than identical frameworks to embed usability engineering across the full medical device product lifecycle (Association for the Advancement of Medical Instrumentation, 2009; Association for the Advancement of Medical Instrumentation, 2025; International Electrotechnical Commission, 2021).

Focusing specifically on these intelligent technologies, this review outlines a multi-decade developmental framework and identifies translational and regulatory considerations particularly relevant to human–AI collaborative systems. This targeted perspective complements broader reviews and enriches the literature on HFE specific to next-generation intelligent medical systems.

Despite progress in academic and regulatory HFE research, many existing reviews organize developments in a primarily chronological manner, with less emphasis on paradigm-oriented comparative analysis. Research on intelligent HFE remains distributed across technical, clinical, and regulatory domains: academic work often emphasizes technical innovation and cognitive modeling, while regulatory efforts focus on safety and usability validation, contributing to a noticeable disconnect between theoretical advances and clinical implementation. A recent critical review of human factors in digital healthcare systems similarly reports that research efforts remain dispersed across subfields and lack a unifying analytical structure (Ogbeyemi et al., 2025).

Limited research has explored the connections between fragmented research output, inconsistent evaluation systems and disciplinary divides, which hinder the establishment of unified assessment architectures for intelligent medical HFE. Moving past chronological descriptive summaries, this review explores drivers of cross-domain separation and puts forward an integrated framework connecting technical evolution, clinical demands and regulatory standards, adding complementary perspectives to available literature.

Disparities persist between theoretical modeling, methodological development, and industrial translation. To help address these disparities, this review structures the evolutionary trajectory of medical HFE, identifies four core tensions within intelligent healthcare delivery—notably the misalignment between technical performance and clinicians’ cognitive capacities—analyzes ongoing theoretical debates and methodological limitations, and proposes contextually relevant directions for future research aligned with digital innovation and global regulatory trends. Clinician expectations and cognitive factors significantly influence the real-world performance of clinical decision support systems (Van Biesen et al., 2022), reinforcing the importance of user-centered design for intelligent medical devices.

This work integrates technical research, clinical practice and international regulatory guidance, reconciles conflicting viewpoints, adopts standard literature screening steps, and carries out thematic analysis grounded in real-world intelligent medical device applications. The resulting integrated framework offers supplementary perspectives for subsequent research, and all analytical procedures follow established scoping review methods: standardized literature retrieval, dual independent screening, quality assessment, and thematic synthesis of evidence.

2 Methodology

This study employs a structured analytical approach commonly used in biomedical engineering and human factors research. Methodological procedures, including literature retrieval, screening, eligibility criteria, data charting, quality appraisal, and thematic synthesis, are described in detail in the sections that follow.

2.1 Literature databases

Literature searches were carried out in four mainstream databases: IEEE Xplore, Web of Science, PubMed, and Scopus, covering biomedical engineering, medical informatics, and human factors research. Relevant industrial standards and regulatory documents were additionally collected from ANSI, AAMI, and IEC.

2.2 Search strategy and search terms

A combined Boolean search strategy was used (HFE OR usability engineering OR human–computer interaction) AND (intelligent medical device OR medical AI OR surgical robot OR wearable medical device OR telemedicine OR MR in healthcare) AND (human–AI trust OR explainable AI OR remote interaction OR immersive technology OR medical device regulation).

Publications published between January 2020 and April 2026 were retrieved to capture state-of-the-art advances in the intelligent transformation of medical devices. To ensure consistent analytical boundaries, only full-text English academic articles were included in the literature screening, while non-English literature was excluded. Authoritative industrial standards and regulatory documents were separately collected and supplemented for thematic analysis.

2.3 Inclusion and exclusion criteria

2.3.1 Inclusion criteria

• Peer-reviewed journal articles, conference papers, official industrial standards, and regulatory documents focused on human factors in intelligent medical devices;
• Studies investigating human–AI interaction, remote medical operation, immersive clinical applications, and medical device usability;
• Literature analyzing industrial development, technical bottlenecks, and regulatory requirements for medical human factors engineering.

2.3.2 Exclusion criteria

• Letters, short communications, abstract-only papers, editorials, and case reports lacking thematic landscape analysis;
• Studies focused solely on traditional medical devices without intelligent functionalities;
• Literature concerning basic medical science, clinical interventions, or pure hardware performance, with no consideration of human factors;
• Duplicate publications and documents with unavailable full text.

To ensure the relevance and rigor of the included literature, explicit eligibility criteria were defined prior to screening. These criteria were applied consistently across all retrieved records to align with the core research objectives of this review. Detailed inclusion and exclusion criteria are summarized in Table 1.

TABLE 1. Eligibility criteria for literature retrieval and screening in this scoping review.

2.4 Screening process and quantitative records

A two-stage literature screening procedure was performed by two reviewers independently, with discrepancies resolved through group consensus discussions.

Initial retrieval: A total of 876 records were retrieved from targeted databases and standard institutional platforms.

First-round screening (title and abstract): After eliminating 142 duplicate entries, 597 irrelevant records were excluded based on preliminary thematic judgment, leaving 137 records eligible for full-text assessment.

Second-round screening (full-text): Guided by the predefined inclusion and exclusion criteria, 122 records were further excluded due to mismatched research themes and insufficient content completeness.

Final included literature: Ultimately, 15 core documents, comprising both peer-reviewed articles and authoritative standard documents, were incorporated for subsequent thematic landscape analysis.

This two-stage independent screening workflow strictly follows the PRISMA-ScR 2018 guidelines for scoping reviews. The complete quantitative screening process is visualized in Figure 1, which presents a transparent and reproducible literature selection pipeline. Detailed numerical statistics at each screening stage are summarized in Table 2.

TABLE 2. Quantitative statistics of literature screening for scoping review.

The PRISMA-ScR 2018 flow diagram of study selection is presented as Figure 1.

2.5 Literature quality assessment

This study applied structured quality assessment and bias mitigation procedures during literature selection. Journal articles were evaluated against five criteria: research design, internal validity, data completeness, analytical logic, and citation consistency. Industry and regulatory documents from ANSI, AAMI, and the FDA were assessed based on their authority, timeliness, and applicability. Publications with inadequate evidence, flawed reasoning, or limited methodological rigor were excluded during full-text review.

To reduce publication and selection bias, this review included both peer-reviewed articles and official regulatory and industry documents, reflecting a range of practical evidence. Two reviewers independently conducted screening in two phases; disagreements were resolved through discussion to ensure consistent application of eligibility criteria.

2.6 Literature analysis and synthesis

The included literature was categorized into five core thematic domains: HFE evolutionary paradigms, prevalent human–AI interaction challenges, risks of remote and immersive clinical applications, theoretical and methodological bottlenecks, and industrial and regulatory constraints. This study performed cross-study comparison and synthesized inconsistent academic perspectives in the existing evidence, summarizing key research concerns, common methodological limitations, and understudied research directions for future investigation.

To clarify how the two literature streams are used in this review, the 15 core documents identified through the formal PRISMA-ScR screening process (Table 2; Figure 1) constitute the sole evidence base for the thematic coding and synthesis reported here. Additional sources cited elsewhere in the manuscript, including further regulatory instruments, technical standards and background publications, were not processed through this formal screening workflow. They are cited selectively to provide conceptual definitions, regulatory context and supporting argumentation, and they do not contribute to the thematic synthesis or to the counts reported in Table 2 and Figure 1.

3 Paradigm evolution of HFE in medical devices

The development of human factors engineering for medical devices encompasses linear technical accumulation and iterative paradigm adjustments, which are driven by evolving clinical requirements, technological progress, regulatory updates, and industrial advancement. Over the past 4 decades, this field has undergone gradual shifts in research priorities, analytical perspectives, evaluation criteria, and application scopes. On this basis, this review categorizes the developmental trajectory of medical device HFE into four distinct yet inherently interconnected stages.

3.1 Safety-driven initiation stage (1980–2000): from accident response to regulatory foundation

Before the 1980s, medical device development was purely hardware-oriented. Manufacturers prioritized mechanical reliability, electrical stability and core functionality during prototype validation, while human factors such as clinicians’ operational behaviors, cognitive traits and environmental adaptability were excluded from standardized testing (Burlington, 1996). Early device regulations focused only on mechanical and electrical safety, lacking unified ergonomic assessment for human–device interaction. Interfaces and operational logic were designed from an engineering perspective, with little regard for user cognitive limits in high-stress clinical scenarios.

From the 1980s, the widespread use of high-risk devices including infusion pumps, anesthetic workstations and ventilators contributed to growing preventable clinical adverse events. Contemporary studies confirmed that most patient safety incidents arose from flawed interfaces and problematic workflows, rather than device hardware failures (Leape et al., 1993). In particular, poorly designed dose-entry interfaces on early infusion pumps frequently caused medication errors and avoidable safety risks.

Growing safety requirements prompted the release of ANSI/AAMI HE48:1993, the first standard specially developed for medical device HFE. This standard built a core framework that mandates full-lifecycle consideration of user traits, task analysis and environmental limits during device development (Association for the Advancement of Medical Instrumentation, 1993). Retrospective review documents issued by regulatory authorities further documented numerous safety incidents triggered by poor usability design of medical devices (U.S. Food and Drug Administration, 2011). At the same time, socio-technical systems theory enabled structured examination of user–device–clinic interactions (Appelbaum, 1997), while the Swiss Cheese model framed safety breakdowns as systemic weaknesses instead of independent human mistakes (Reason, 1990).

Despite these regulatory and theoretical advances, HFE practices from 1980 to 2000 remained largely reactive. Most usability improvements were implemented post-incident instead of being integrated into early design stages. Even so, this period established a key principle: human factors should be embedded in initial device design rather than added as a retrospective modification.

Digital upgrading of medical device interfaces is well-established in academic discourse, yet the phased developmental logic of early medical human factors engineering remains unstructured and fragmented. This review categorizes the inaugural developmental phase as safety-driven initiation, consolidating interdisciplinary regulatory, theoretical and practical markers to supplement the systematic evolutionary framework of medical HFE.

3.2 Professional establishment stage (2000–2010): cognitive turn and standardization

From 2000 to 2010, medical HFE entered a phase of professionalization, standardization, and cognitive paradigm transition. Rapid medical informatization increased the complexity of device interfaces and software systems, making excessive cognitive load a prominent patient safety concern. Traditional physical ergonomics was insufficient to explain information misinterpretation, operational delays, and memory-related errors in complex clinical environments.

Cognitive theories, including cognitive load theory, multiple-resource theory, situational awareness models, and standardized human error taxonomies, were gradually integrated into clinical human factors research. Accordingly, analytical scope expanded from individual operator performance to broader healthcare socio-technical systems. The Systems Engineering Initiative for Patient Safety (SEIPS) provided a structured framework to characterize dynamic interactions among human operators, technical devices, and organizational contexts (Carayon et al., 2006; Holden et al., 2013; Kudina and van de Poel, 2024).

Meanwhile, updated regulatory benchmarks, such as IEC 60601–1:2005, facilitated the transition of medical HFE from experience-based practice to a standardized, process-oriented, and quantifiable discipline (International Electrotechnical Commission, 2006). Harmonized human factors design specifications across European Union and United States medical device regulatory frameworks lay a solid foundation for unified usability assessment workflows (Privitera et al., 2017). The subsequent U.S. Food and Drug Administration (FDA) guidance for artificial intelligence/machine learning (AI/ML)-based medical devices further refined this framework by specifying usability validation requirements for adaptive algorithm systems (U.S. Food and Drug Administration, 2024).

This cognitive paradigm shift supported deeper analysis of error-inducing mechanisms, rather than only focusing on post-incident mitigation, representing a key step in the professional development of medical HFE. Usability evaluation was increasingly formalized as a necessary design checkpoint, shifting the dominant HFE approach from retroactive remediation to proactive design optimization.

Though prior research has recognized the integration of HFE principles into clinical workflow design, few studies have unpacked the cognitive engineering mechanisms behind such practical implementations. This developmental stage highlights workflow-centered optimization as a pivotal transition toward improved system-scale human factors governance in clinical practice.

3.3 System integration stage (2010–2020): team collaboration, workflow alignment, and system safety

Between 2010 and 2020, the broad uptake of surgical robots, integrated operating suites and medical Internet of Things systems steered medical HFE toward broader system integration and team-oriented design. Traditional single-user, single-device interaction patterns were progressively replaced by multi-agent, multi-device and multi-workflow clinical architectures. Correspondingly, HFE challenges expanded beyond individual usability concerns to cover team coordination efficiency, cross-platform information consistency and overall compatibility with clinical workflows.

The popularization of surgical robotics brought unique human factors risks, such as inadequate force feedback, prolonged physical strain, impaired depth perception, and altered intraoperative team interactions (Hung et al., 2021). These emerging concerns promoted a shift in medical HFE research, expanding standalone device usability assessment toward system-level safety management and workflow integration.

Research priorities in this period covered explainable AI, decision-making transparency, dynamic human–AI trust calibration, adaptive automation, clinical telepresence, personalized human–machine interaction, and ethical specifications for collaborative healthcare (Chanda et al., 2024; Hellmeier et al., 2024; Sanderson et al., 2023). Unlike conventional one-off pre-market evaluation, full-lifecycle continuous validation has become a key requirement for intelligent medical devices (Hellmeier et al., 2024). Long-term real-time clinical monitoring and iterative algorithm optimization are also essential to maintain stable operational performance in complex clinical environments (Feng et al., 2022). The COVID-19 pandemic has further spurred the growth of home-based care, remote monitoring, and mobile diagnostics. As user groups now include patients, family caregivers, and older adults, devices face heightened requirements for accessibility, ease of use, and cross-cultural adaptability.

Technological advancements in telemedicine and surgical robotics have been well documented in prior research, yet the distinct human factors risks arising from distributed collaborative clinical workflows remain understudied. By unpacking the inherent conflicts between remote usability optimization and real-time clinical safety governance, this analysis enriches the contextual and scenario-based interpretation of medical HFE evolutionary development.

3.4 Intelligent transformation stage (2020–present): human-machine symbiosis, adaptive interaction, and ethical governance

Since 2020, the rapid development of AI, telemedicine, MR, and wearable medical devices has further promoted the intelligent transformation of medical HFE, moving the field toward a human–machine symbiosis paradigm. Modern intelligent medical devices act as collaborative partners with environmental perception, autonomous decision-making, and intelligent recommendation functions, gradually reshaping human–machine relationships in role allocation, interaction, real-time monitoring, and trust dynamics. Among these, intelligent surgical robots have amplified the perceptual and interaction risks observed in earlier stages.

Research focus in this period has become more targeted, with growing attention to context-aware adaptive interaction and standardized ethical governance for intelligent medical systems (Chanda et al., 2024). As noted previously, the COVID-19 pandemic has diversified the user base of intelligent medical devices. In contrast to traditional HFE work centered on interface and workflow optimization, current research is moving toward intelligent, personalized, and ethically bounded frameworks, with the goal of building safe, stable, human-centered human–AI collaboration ecosystems for clinical use.

Representative intelligent medical devices, including the da Vinci and TUMA surgical robots, have demonstrated common human factors challenges in clinical practice. Surgeons often experience limited tactile feedback and visual-perceptual inconsistencies during long procedures, which may increase operational uncertainty and safety risks. Meanwhile, the expanded use of portable remote diagnostic tools and MR surgical navigation in primary and remote care settings has highlighted the need for tailored human factors optimization for intelligent medical devices.

While the turn toward human–AI collaborative care has been recognized in the literature, its developmental continuity with traditional HFE frameworks remains underexplored. This review situates intelligent human–machine symbiosis as the contemporary phase of medical HFE evolution, and identifies key research priorities to support HFE practice within increasingly autonomous clinical systems.

4 Core scientific issues in human factors engineering for intelligent medical devices

The four core research issues outlined in this section are drawn from three evidence-based sources: analysis of the included literature, interpretation of major international medical device regulatory standards, and practical observations of deployed intelligent medical systems. These themes capture prominent constraints and challenges within current medical HFE research, which are elaborated in the following sections.

The widespread use of intelligent medical systems has significantly changed clinical human–machine interaction patterns, such that traditional operation- and cognition-focused HFE frameworks may be insufficient to address emerging interaction and safety risks. On this basis, four interrelated core scientific challenges are proposed as key directions for contemporary medical HFE research.

4.1 Reconstruction of human–machine responsibility, authority, and trust

AI diagnostic models, clinical decision support systems, and intelligent monitoring platforms have reshaped conventional human–machine interaction patterns in healthcare, bringing new cognitive and ethical challenges. AI algorithm reasoning often diverges from standardized clinical logic, and common post hoc explanation methods may increase clinicians’ cognitive load without improving diagnostic safety or operational efficiency (Dlugatch et al., 2024). Dynamic human–machine trust remains difficult to quantify and calibrate, as both overtrust and undertrust can impair system performance and clinical safety (Choudhury et al., 2022). In addition, clinicians’ perceptions of AI tool utility vary considerably across dynamic clinical settings (Payne et al., 2024).

Existing frameworks for assigning automation authority lack scenario-based theoretical backing, and researchers have not reached unified agreement on the optimal automation level to balance situational awareness and workload control (Parasuraman et al., 2008). In addition, algorithmic bias may widen healthcare inequities, as skewed training datasets frequently produce inconsistent diagnostic performance among distinct population groups (Obermeyer et al., 2019). Such intrinsic bias compromises medical equity and restricts standardized HFE design, calling for fairness-focused principles to be embedded across the full device lifecycle. International consensus documents have laid out uniform criteria for dataset transparency and bias reduction to mitigate algorithmic disparities among varied patient cohorts (Alderman et al., 2025).

Real-world applications of AI electrocardiogram analysis and intelligent medical imaging tools reflect prevalent human–machine frictions in clinical practice. Clinicians often undertake extra verification work to validate outputs from low-reliability AI systems, increasing routine operational burden. This practical issue underscores the need for robust human–AI trust calibration mechanisms and clear responsibility allocation in intelligent clinical collaboration (Ankolekar et al., 2022; Henry et al., 2022). Clinicians adopt diverse coping strategies when responding to AI-generated recommendations, including active acceptance, selective neglect, and conditional compromise (Sivaraman et al., 2023). Similarly, radiologists exhibit inconsistent attitudes toward the reliability and practical value of AI tools in routine workflows (York et al., 2023).

The mismatch between technical reliability and user-perceived trustworthiness represents a persistent challenge for clinical practitioners, especially frontline nursing staff (Higgins O. et al., 2024). Existing studies have identified the risks of clinical AI overtrust and undertrust but rarely explore the inherent tension between algorithmic explainability and dynamic trust fluctuation. This section elaborates the core trade-offs among algorithmic transparency, clinical accountability, and operational efficiency in human–AI collaborative diagnostic workflows.

4.2 Individualized adaptation vs. global system consistency

Intelligent medical devices face diverse user groups, creating inherent trade-offs between personalized interactive adaptation and standardized system consistency. End users vary greatly in age, educational background, cognitive ability, motor function, visual performance, and health literacy, leading to differentiated interactive requirements in both clinical and home-based scenarios. A one-size-fits-all design is insufficient to meet hierarchical accessibility needs, while excessive personalized customization may cause operational confusion and inconsistent system performance across devices.

Research has explored adaptive interface tools built on real-time user monitoring within medical HFE, yet their broad clinical rollout is constrained by inconsistent industrial standards and incomplete safety validation. A central practical challenge lies in delivering secure, structured personalized interaction while preserving the overall stability, uniform specifications and replicability of intelligent medical systems.

Wearable physiological monitors and home-based intelligent insulin pumps well illustrate this design dilemma. For elderly users and patients with chronic diseases, tailored parameter settings can improve operational comfort and long-term user adherence. However, excessive individual customization may result in inconsistent operational logic across device terminals, potentially increasing misuse risks for clinicians who operate multiple types of intelligent medical devices in daily workflows.

A further unresolved tension concerns how personalized adaptation can be reconciled with standardized usability validation. Regulatory usability frameworks such as IEC 62366–1:2021 assess safety against predefined use scenarios and representative user profiles, whereas adaptive interfaces modify their behavior dynamically for individual users. A configuration verified as safe in one adapted state therefore cannot be assumed to hold across all possible personalization states. Although ISO 9241–210:2019 endorses iterative human-centered design refinement, neither document specifies how to bound and validate an adaptive interaction space, leaving manufacturers without an accepted method for demonstrating that personalization remains within verified safety limits.

Current research has highlighted the value of personalized interface design in intelligent healthcare, yet pays limited attention to its inherent conflicts with system standardization and regulatory compliance. This analysis clarifies unresolved research issues regarding the dynamic balance between adaptive usability optimization and standardized clinical workflow consistency.

4.3 Human factors constraints in remote medical interaction

Telemedicine, mobile healthcare, and home monitoring systems have removed geographical barriers to clinical services while introducing unique human factors risks associated with media-mediated human–patient interaction. Remote scenarios often involve incomplete situational information and limited telepresence, lacking the nonverbal cues and tactile feedback that support comprehensive clinical assessment and robust decision-making. Network instability, transmission latency, and image distortion further interrupt continuous care delivery and may induce critical operational errors in high-risk settings, such as remote robotic surgery and emergency telemedicine (Fraser et al., 2023).

In addition, uneven digital health literacy exacerbates usage disparities among vulnerable populations, contributing to low device adherence and latent clinical safety risks in home-based long-term monitoring scenarios (Mattison et al., 2022). Self-powered wearable sensing hardware further simplifies long-term remote physiological data collection and reduces charging-related usage barriers for chronic disease patients (Gong et al., 2024). Accessibility-centered design is a core prerequisite to narrow health literacy gaps and make home medical devices usable for aging and low-literacy groups (Perlow, 2010). Multiple systematic reviews have highlighted persistent usability defects of medical devices deployed in home environments, including complex operation logic and poor interface adaptability for elderly users (Tase et al., 2022). Trust formation in remote healthcare has also shifted from traditional interpersonal doctor–patient trust toward technology-mediated trust. Algorithm opacity and widespread data security concerns can reduce user acceptance and constrain the scalable deployment of remote intelligent medical devices (Price and Cohen, 2019).

Existing telemedicine HFE studies mainly focus on technical limitations such as network latency and immediate operational errors (Fraser et al., 2023), with limited discussion on how insufficient telepresence affects long-term doctor–patient relationship stability. Current research also exhibits regional heterogeneity: studies in developed regions prioritize interactive efficiency and system optimization, whereas those in rural and underdeveloped areas focus more on digital divides and accessibility barriers. Such context-specific disparities remain underdiscussed in existing reviews, and this study provides further elaboration on these neglected dimensions.

While previous work has documented basic usability issues in telemedicine systems, few studies have traced these risks to cross-regional cognitive heterogeneity and uneven technical infrastructure. This section identifies bandwidth constraints, cultural ergonomic mismatches, and asynchronous interactive friction as key limiting factors for standardized and safe remote clinical care.

Practical observations from China’s long-distance robotic surgical applications support the above risk analysis. Cross-regional and transcontinental robotic surgeries assisted by 5G and satellite networks, using TORSS and TUMA systems, illustrate safety vulnerabilities caused by network latency and insufficient telepresence. In addition, remote monitoring devices used in rural and underserved regions are often hindered by low user digital literacy, resulting in poor device uptake and ongoing safety concerns in home-based clinical care.

4.4 Safety boundaries and risk control of immersive medical technologies

Virtual reality (VR), augmented reality (AR), holographic navigation, and MR offer unique advantages in surgical guidance, functional rehabilitation, and clinical skill training. However, the human factors safety boundaries of these immersive clinical tools remain insufficiently defined. Excessive immersive engagement may induce dizziness, physical fatigue, and detachment from real clinical surroundings, potentially increasing operational errors in high-acuity care settings (Kadirvelu et al., 2023; Ricotti et al., 2023). Standardized operational guidelines and ethical constraints are therefore essential to support the safe and scalable deployment of immersive intelligent medical systems (Wiens et al., 2019).

Immersive hardware encounters practical obstacles to wide clinical translation, including steep learning curves, high training costs, and spatial collision hazards. Furthermore, unified evaluation frameworks and quantitative safety thresholds have not yet been established (Shajari et al., 2023). Current research adopts disparate metrics to measure perceptual deficits in immersive scenarios, which hinders cross-study comparison and integrated evidence collation. Most empirical results stem from short-duration laboratory tests, with scarce long-term clinical data capturing cumulative operational risks among clinical practitioners. Standardized evaluation architectures are likewise required to enable consistent safety verification and performance testing for surgical robotic platforms (Marcus et al., 2023).

AR surgical navigation devices and VR training systems have been increasingly adopted in tertiary hospitals, with real-world clinical feedback indicating persistent perceptual side effects after prolonged immersive operation. Fragmented, brand-specific evaluation benchmarks further hinder the formation of universal safety certification standards for immersive medical devices.

While previous literature has demonstrated the clinical benefits of VR and MR applications, it pays limited attention to latent safety issues such as cognitive overload and spatial disorientation. This section elaborates key unresolved challenges regarding unified risk assessment, safety threshold calibration, and standardized performance evaluation for immersive clinical technologies.

5 Theoretical controversies, methodological limitations, and practical obstacles

Despite rapid progress in human factors engineering for intelligent medical devices, the field still faces inherent theoretical constraints, methodological shortcomings, and translational bottlenecks that limit its large-scale industrial adoption and standardized clinical implementation.

5.1 Theoretical controversies and unresolved basic problems

The optimal automation level for intelligent medical systems remains debated, with inconsistent findings regarding efficiency gains and safety performance, and no unified analytical framework has resolved such discrepancies (Frederiksen and Sørensen, 2021). General usability design principles often conflict with the specialized functional demands of high-precision surgery and emergency care. In addition, system-level workflow management and individual cognitive ergonomic analysis are frequently decoupled, creating trade-offs between overall operational efficiency and localized cognitive load regulation. Ethical and liability boundaries in human–AI collaborative workflows also remain unclear, with ambiguous accountability distribution among clinicians, algorithm developers, and device manufacturers (Higgins D. C. and Johner, 2023; Jones et al., 2023).

Studies on hierarchical automation intensity have yielded inconsistent results. Laboratory-based experiments suggest that moderate automation balances operational efficiency and situational awareness, whereas field investigations of surgical robotic systems show a clinical preference for higher automation to reduce physical workload and operational fatigue (Hung et al., 2021). Such mismatches between laboratory evidence and real-world clinical practice represent a key unresolved challenge in current medical HFE research.

These inconsistent research findings stem from structural mismatches within existing theoretical frameworks. Traditional static HFE paradigms, originally designed for isolated single-user and single-device scenarios, cannot fully adapt to the dynamic human–AI collaborative workflows of modern intelligent healthcare systems. Previous reviews mostly enumerate scattered research controversies without grouping or interpreting these conflicts under a unified contextual constraint, leading to fragmented analytical perspectives on disciplinary developmental challenges.

Although existing literature has well documented competing views on automation configuration and human–AI accountability allocation, few studies attribute these inconsistencies to paradigm friction: the poor adaptability of conventional static HFE theories to the dynamic and fluid interaction patterns of intelligent clinical environments.

5.2 Methodological limitations and deficits in ecological validity

Mainstream HFE evaluation methods exhibit limited ecological validity, as controlled laboratory trials cannot fully replicate real-world clinical features, such as time pressure, unexpected workflow interruptions, and dynamic team collaboration. Longitudinal in-situ monitoring remains uncommon due to rapid device iteration and high research costs. While qualitative methods can capture clinicians’ practical experiences with AI-assisted tools (Stacy et al., 2022), conventional evaluation frameworks prioritize operational efficiency and explicit error metrics, often overlooking latent safety risks arising from fluctuating cognitive load, reduced situational awareness, and cross-team coordination friction (Reverberi et al., 2022). In addition, data-driven and model-based assessment approaches are insufficiently integrated into routine HFE validation, restricted by limited clinical data availability and immature multimodal fusion techniques (McDuff et al., 2025). Although participatory design frameworks help align intelligent medical systems with clinical practical needs (Rambach et al., 2024), methodological improvements have lagged behind the rapid iteration of intelligent medical devices.

The mismatch between laboratory testing and authentic clinical workflows reduces the translational value of existing HFE research. These methodological limitations are not isolated but form interconnected constraints that hinder clinical implementation. Testing biases may distort core human factors indicators, limited clinical data restricts multimodal model integration, and the lack of long-term monitoring fails to capture cumulative operational risks for clinical users. Existing studies mostly discuss these limitations separately and rarely explore their interactive mechanisms, leading to fragmented findings that cannot effectively support evaluation framework optimization. Such methodological constraints impede the translation of HFE theories into safe and applicable intelligent medical device solutions.

Usability assessments of surgical robotic systems, as mentioned earlier, exemplify such laboratory testing biases. Against this background, the quantitative evaluation of clinician–AI interaction quality has become a growing research direction (Perivolaris et al., 2024). In particular, full-range clinical demand assessment in the early design stage is essential for standardized user-centered development of next-generation intelligent medical devices (Zheng et al., 2024).

5.3 Practical obstacles to industrial translation and clinical application

HFE research faces low industrial translation efficiency, partly due to differing priorities between academia and manufacturers. Industry tends to focus on rapid launch, cost control, and regulatory compliance, rather than iterative user-centered design (Federico and Trotsyuk, 2024). Intelligent medical technologies face additional implementation challenges in high-acuity care settings (Pinsky et al., 2024). Unclear approval pathways for AI and immersive devices slow product iteration, and even well-designed clinical decision-support tools encounter practical deployment barriers (Nair et al., 2023). A shortage of professionals with combined expertise in clinical practice, HFE, AI, and regulation also limits global and cross-cultural implementation, with similar challenges reported for AI-based prescription and administrative tools (Yip et al., 2023; Yoon et al., 2024).

These barriers are interconnected and form a self-reinforcing cycle that limits industrial translation of HFE research. Cost and time pressures reduce industry investment in iterative user-centered optimization, while unclear regulatory requirements for AI and immersive systems delay consistent design standards. In turn, ambiguous implementation guidelines hinder the development of interdisciplinary talent, leaving mature HFE outcomes largely confined to academic research rather than applied in clinical practice. Most existing reviews discuss these barriers separately, limiting their practical guidance for industrial translation.

Although individual adoption challenges have been noted in previous studies, limited work has addressed translational bottlenecks in an integrated manner. This section outlines a multi-level framework of underlying barriers, including economic constraints, regulatory uncertainty, and interdisciplinary talent gaps.

6 Future trends and development paths

6.1 Research focus: from interface usability to human–AI collaborative reliability

Future medical HFE research will evolve toward human–AI collaborative systemic reliability, extending beyond interface optimization to integrated system-level safety governance. As highlighted in the section on core scientific issues, unregulated human–AI trust dynamics present key safety challenges, which can be addressed through role-specific explainable AI to support transparent, standardized clinical decision-making (Alderman et al., 2025). Multimodal interaction designs reduce cognitive burden in high-intensity settings, while affective computing and real-time physiological monitoring enable early detection of operator fatigue and stress (Wiens et al., 2019). These approaches establish intelligent systems as reliable clinical aids that enhance rather than replace professional judgment, improving safety and efficiency across intelligent healthcare workflows.

From a translational perspective, interpretable algorithm outputs require hierarchical, role-adaptive presentation frameworks tailored to physicians, nurses, and patients. Optimized multimodal interaction design further alleviates cognitive overload during high-risk clinical and surgical procedures, balancing operational efficiency and procedural safety.

This research direction shifts the conventional interface-centric HFE evaluation paradigm and integrates classic human factors principles with digital medical technologies. Traditional usability assessments, which rely on static interface indicators and basic cognitive load metrics, fail to capture AI-induced latent operational risks. By contrast, integrated frameworks combining interpretable models and real-time physiological tracking enable quantitative evaluation of core HFE indicators, including sustained situational awareness and stable workload regulation. This digital evaluation paradigm remedies the limitations of static laboratory testing and broadens the interdisciplinary boundary of modern medical HFE research.

Conventional research has centered on isolated usability and interface issues, with less emphasis on holistic challenges in clinical role distribution. This subsection presents key considerations for role-adapted explainable AI systems to foster consistent clinical translation of intelligent medical devices.

6.2 Methodological innovation: data-driven and virtual validation

Data-driven analytical approaches, digital twin simulation, and privacy-preserving computing have emerged as effective methodological solutions to remedy the inherent limitations of conventional laboratory HFE evaluation (Feng et al., 2022; de Hond et al., 2022; Hellmeier et al., 2024). In-silico clinical simulation platforms can reproduce diverse clinical interaction scenarios to test physician–AI matching effects without real patient participation (Hofmann, 2026). Combined with ecological momentary assessment and MR technology, these tools enable in-situ capture of authentic human–machine interaction data, break institutional data silos, and support closed-loop validation of high-risk clinical workflows, thereby enhancing the overall objectivity and efficiency of HFE testing.

Nevertheless, current digital simulation methodologies face practical barriers, including high development costs and insufficient scene restoration fidelity, which hinder their large-scale industrial and clinical deployment. Future methodological research should prioritize lightweight simulation architectures, standardized test datasets, and unified quantitative evaluation metrics. Such optimizations can lower technical and financial thresholds for medical device manufacturers, while improving the cross-institutional comparability and traceability of data-driven HFE assessment.

The integration of real-world big data analytics and digital twin virtual testing also innovates traditional HFE evaluation logic. This combined paradigm captures long-term cumulative operational risks and reconstructs controllable high-risk clinical scenarios that are unattainable in conventional experiments, shifting HFE assessment from passive post hoc error summary to proactive predictive risk governance. Existing literature merely presents these digital methods in isolation, lacking systematic discussion of their synergistic value in optimizing medical HFE evaluation systems.

Against this gap, this subsection proposes targeted methodological optimization priorities, advocating the integrated application of twin simulation and real-world data analytics to advance standardized, high-validity HFE evaluation in clinical practice.

6.3 Application transformation: full lifecycle and ecosystemic HFE design

Practical medical HFE development is shifting from independent single-device optimization toward integrated medical Internet of Things ecosystem development (Sanderson et al., 2023). Industrial design is moving beyond rigid universal standards toward context-sensitive personalization and region-specific compliance, supporting globally applicable cross-cultural HFE frameworks (Kudina and van de Poel, 2024) and more equitable, accessible deployment of intelligent medical devices.

From an industrial perspective, full-lifecycle human factors governance should be embedded across R&D, verification, market launch, and post-market surveillance. Cross-cultural ergonomic research can further refine region-adaptive interface design and operational protocols, while strengthened industry–academia collaboration supports the translation of HFE theories into practical industrial specifications and regulatory standards.

Digital healthcare infrastructure provides a technical foundation for ecosystem-oriented, full-lifecycle HFE implementation. In contrast to conventional fragmented pre-market-focused evaluation, medical IoT and cloud-based supervision enable continuous human factors monitoring throughout the device lifecycle. As identified in the section on core scientific issues, such ecosystem-level approaches help address regional disparities in digital health literacy and reconcile personalized interaction with unified industrial standards, supporting coordinated HFE optimization across the entire industrial chain. Existing reviews often identify full-lifecycle design as a trend but provide limited insight into how digital infrastructure enables equitable and patient-centered HFE delivery.

Rather than prioritizing customization or standardization in isolation, future design should pursue balanced, adaptive strategies that align context-aware interfaces with standardized regulatory and workflow requirements. This subsection outlines practical industrial pathways to achieve such integration, providing actionable guidance for ecosystem-level HFE development.

6.4 Targeted future research recommendations

This review puts forward targeted research agendas to resolve the theoretical disputes, methodological constraints and translational barriers discussed in this review, while integrating core HFE principles with modern digital medical technologies. The analysis delivers structured, actionable research paths for follow-up academic exploration and standardized clinical HFE practice.

First, theoretical research should develop a unified dynamic human–AI co-adaptation framework tailored to intelligent clinical scenarios. Traditional static HFE paradigms, which are based on single-user and single-device interaction, require revision to accommodate real-time algorithmic outputs, dynamic human–AI trust changes, and flexible authority allocation between clinicians and automated systems. This integrated framework can balance automation efficiency, independent clinical judgment, and operational safety, while grounding classical HFE theories in the cognitive mechanisms of human–AI collaborative workflows.

Second, methodological innovation should prioritize hybrid validation pipelines that combine real-world clinical data analytics with digital twin simulation. Conventional laboratory usability testing has limited ecological validity and cannot fully capture long-term cumulative human factors risks in routine clinical operations. Future evaluation systems may integrate large-scale clinical datasets, continuous physiological monitoring, and lightweight virtual simulation to support full-lifecycle HFE assessment. Such hybrid methods support quantitative analysis of dynamic cognitive load, variable situational awareness, and cumulative operational risk, providing empirical support for the safety evaluation and iterative optimization of intelligent medical devices.

Third, translational HFE design should focus on context-aware and region-adaptive development strategies. Heterogeneous digital health literacy, uneven clinical infrastructure, and variable local workflow protocols limit the applicability of universal interface design solutions. Future translational research should develop flexible interactive mechanisms that balance personalized user-centered adaptation with unified industrial specifications and cross-border regulatory compliance. Special attention should be paid to insufficient telepresence in remote healthcare and its long-term influence on clinician–patient interaction stability, two dimensions that receive limited discussion in current studies.

Fourth, cross-domain knowledge transfer from general human factors and industrial engineering merits further exploration. Ergonomic optimization approaches applied in manufacturing, supply chain management, and industrial operational control offer valuable theoretical and methodological references for clinical system design, proactive error mitigation, and multi-stakeholder collaborative governance. The adoption of such interdisciplinary insights can advance medical HFE from single-device usability evaluation toward integrated workflow-centered human-centered design, improving its theoretical completeness and industrial applicability.

Overall, the four research directions promote the integration of traditional HFE theories and emerging digital technologies, providing practical guidance for the safe, equitable, and efficient clinical deployment of next-generation intelligent medical devices.

7 Discussion

This section addresses the four core research issues raised in the introduction and summarizes the key findings of this scoping review.

First, this study establishes a four-stage evolutionary framework for intelligent medical HFE based on progressive disciplinary development. Beyond chronological technical description, this review reorganizes scattered historical advances into four coherent phases and illustrates how digital innovation reshapes clinical human–machine interaction patterns. This framework complements existing descriptive reviews by providing a structured evolutionary perspective on medical HFE progress.

Second, this review integrates four interrelated human factors challenges that have been separately addressed in previous literature. As elaborated in the section on core scientific issues, key unresolved tensions include unstable human–AI trust dynamics, the trade-off between personalized adaptation and standardized clinical protocols, and cross-regional mismatches between telemedicine infrastructure and heterogeneous user digital literacy. This integrated perspective consolidates isolated findings under a unified thematic structure.

Third, this review presents a multi-tiered diagnostic framework covering theoretical conflicts, methodological limitations, and translational constraints. Core theoretical dilemmas stem from structural incompatibility between traditional static, single-device HFE models and the dynamic human–AI interaction patterns of modern intelligent healthcare. Methodological weaknesses reduce evaluation ecological validity and reliability, while industrial costs, regulatory uncertainty, and interdisciplinary talent gaps jointly restrict large-scale clinical translation. This layered analysis provides a structured account of cumulative disciplinary bottlenecks.

Fourth, this analysis clarifies inherent links between emerging digital technologies and classical HFE principles. Rather than listing technologies in isolation, this review explains how AI, digital twin, and medical IoT tools address unmet clinical HFE needs, including dynamic cognitive load regulation, proactive risk warning, and equitable cross-regional interface design. This study integrates technological advances, clinical demands, and regulatory requirements into a unified analytical framework.

Overall, this review synthesizes existing literature to construct an integrated framework covering disciplinary evolution, core challenges, and multi-dimensional constraints. By connecting intelligent digital technologies with human-centered HFE theories, this work provides structured insights into the current state and future development of human factors engineering for intelligent medical devices.

8 Conclusion

With the ongoing intelligent transformation of global healthcare, modern medical human factors engineering has expanded beyond traditional interface optimization toward integrated systemic safety governance for human–AI collaborative clinical workflows. Current HFE research is constrained by four intertwined unresolved challenges: dynamic human–AI trust calibration difficulties, inherent trade-offs between personalized interactive adaptation and standardized clinical workflows, persistent human factors barriers restricting high-quality remote care, and the lack of unified risk assessment protocols for immersive clinical devices. Addressing these multi-dimensional conflicts requires interdisciplinary integration across cognitive science, engineering design, regulatory studies, and frontline clinical practice.

This scoping review presents three core analytical outcomes for medical HFE research. First, it proposes a logic-based four-stage evolutionary framework for medical HFE, clarifying the developmental trajectory of intelligent medical HFE beyond simple chronological descriptions of technological progress. Second, it integrates four independently discussed human factors challenges in existing literature, with focused discussion on the tension between personalized user adaptation and clinical standardization, as well as cross-regional mismatches between telemedicine infrastructure and user digital capacity. Third, it provides a multi-layered diagnostic framework to interpret prevalent theoretical controversies, methodological limitations, and industrial translational bottlenecks, offering structured interdisciplinary insights for future digital HFE exploration.

This study has certain limitations. The analysis only includes English peer-reviewed publications and official industrial and regulatory documents, excluding grey literature, which may introduce selection bias. Future research can expand the literature coverage by incorporating multilingual studies and conducting real-world clinical validation to further verify the applicability of the HFE optimization frameworks summarized in this review.

As global regulatory systems and intelligent medical technologies continue to mature, future HFE research will focus on four key directions: quantitative evaluation of human–AI collaborative reliability, data-driven ecologically valid usability testing, digital twin-based simulated validation, and full-lifecycle user-centered design optimization. Continuous progress in these areas will support the safe, equitable, and patient-centric clinical deployment of next-generation intelligent medical devices, further improving the practical value of medical HFE in intelligent healthcare systems

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Qian Zhang, College of Charleston, United States

Reviewed by: Mohamed Adlan Ait Ameur, University of Strathclyde Department of Design Manufacture and Engineering Management, United Kingdom

Akinola Ogbeyemi, University of Saskatchewan, Canada