Scholarlink

Background <p>Electronic health records (EHRs) are central to clinical documentation and care coordination, yet long-standing studies report persistent socio-technical problems such as poor usability, slow response, alert burden, interoperability gaps, and clinician cognitive burden. As EHRs increasingly incorporate decision support and other AI-related functionalities, the evidence base now spans both traditional EHR design/use and EHR-linked advanced functions. However, findings are often reported in ways that make it difficult to trace how interface and information conditions relate to clinicians’ interaction experiences and, where reported, downstream outcomes.</p> Methods <p>We conducted an integrative literature review of medical informatics research published between 2005 and 2025 on EHR design and use, including EHR-linked decision support and AI-related functionalities when explicitly described. Seventy eligible studies were synthesized using a four-layer socio-technical architecture, Interface, Information, Interaction, and Outcome (4I). For readability, “4I” can be read equivalently as “3I + O” (3I plus an explicit Outcome layer). We developed study-level labels and cross-layer evidence links to derive an evidence-traceable set of 29 variables and to summarize how studies characterized each variable as facilitating, constraining, or mixed for adoption and ongoing use.</p> Results <p>The 29 variables clustered across the 4I layers, with 5 primarily interface-related variables, 5 information-related variables, 10 interaction-related variables, and 9 outcome-related variables. Interface and Information conditions (e.g., access/usability, interoperability, trust, governance, and task–function fit) were most often connected to Interaction experiences, including workflow fit, training and support, trust calibration, and cognitive workload. Across the included studies, adoption was described as more favorable when reliable system performance, interpretable outputs, workflow-aligned training, and organizational support co-occurred. Where studies reported downstream effects, these cross-layer patterns were linked to documentation burden and burnout, patient-safety risks, workflow standardization, and perceived augmentation of clinical decision-making.</p> Conclusions <p>The 4I framework provides a structured synthesis that connects established EHR evidence to EHR-linked decision support and AI-related functionalities where documented, while keeping outcome implications explicit. The resulting 29-variable dictionary supports clearer reporting and cross-layer interpretation, and it provides a practical basis for subsequent expert elicitation and empirical assessment as evidence on AI-linked EHR functions continues to develop.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

From interface to outcome: a 4I framework for AI-linked functionality in electronic health records

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月22日

查看全文
  • Ting Zhang,
  • Nina Shin

摘要

Background

Electronic health records (EHRs) are central to clinical documentation and care coordination, yet long-standing studies report persistent socio-technical problems such as poor usability, slow response, alert burden, interoperability gaps, and clinician cognitive burden. As EHRs increasingly incorporate decision support and other AI-related functionalities, the evidence base now spans both traditional EHR design/use and EHR-linked advanced functions. However, findings are often reported in ways that make it difficult to trace how interface and information conditions relate to clinicians’ interaction experiences and, where reported, downstream outcomes.

Methods

We conducted an integrative literature review of medical informatics research published between 2005 and 2025 on EHR design and use, including EHR-linked decision support and AI-related functionalities when explicitly described. Seventy eligible studies were synthesized using a four-layer socio-technical architecture, Interface, Information, Interaction, and Outcome (4I). For readability, “4I” can be read equivalently as “3I + O” (3I plus an explicit Outcome layer). We developed study-level labels and cross-layer evidence links to derive an evidence-traceable set of 29 variables and to summarize how studies characterized each variable as facilitating, constraining, or mixed for adoption and ongoing use.

Results

The 29 variables clustered across the 4I layers, with 5 primarily interface-related variables, 5 information-related variables, 10 interaction-related variables, and 9 outcome-related variables. Interface and Information conditions (e.g., access/usability, interoperability, trust, governance, and task–function fit) were most often connected to Interaction experiences, including workflow fit, training and support, trust calibration, and cognitive workload. Across the included studies, adoption was described as more favorable when reliable system performance, interpretable outputs, workflow-aligned training, and organizational support co-occurred. Where studies reported downstream effects, these cross-layer patterns were linked to documentation burden and burnout, patient-safety risks, workflow standardization, and perceived augmentation of clinical decision-making.

Conclusions

The 4I framework provides a structured synthesis that connects established EHR evidence to EHR-linked decision support and AI-related functionalities where documented, while keeping outcome implications explicit. The resulting 29-variable dictionary supports clearer reporting and cross-layer interpretation, and it provides a practical basis for subsequent expert elicitation and empirical assessment as evidence on AI-linked EHR functions continues to develop.