International Journal of Medical Informatics最新文献_第4页

Clinician preferences for explainable AI in critical care: a comparative study of interpretable models and visualizations for intubation decision support 临床医生在重症监护中对可解释人工智能的偏好：可解释模型和插管决策支持可视化的比较研究

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-18 DOI: 10.1016/j.ijmedinf.2026.106287

Tiantian Xian , Nikolay Mehandjiev , Panos Constantinides , Yu-wang Chen , Qudamah Quboa , Gareth Kitchen

Background:

The complexity of many AI models hinders their clinical adoption because the clinicians using them do not regard them as transparent. This study addresses the lack of clinician-centered explainable AI (XAI) interfaces by designing and evaluating intuitive visual explanations for intubation prediction, testing the hypothesis that workflow-compatible designs enhance acceptance.

Objective:

This study compares three, time-aware, visual explanations for XAI-based intubation prediction and evaluate their acceptance, comprehension, and perceived utility among clinicians.

Methods:

We developed machine learning models to estimate the near-term risk of deterioration in the patient’s condition which may lead to mechanical intubation using ICU time-series data. We generated global and local explanations using SHAP and designed three customized visual formats—a temporal force plot, a temporal bar chart, and a dual-encoded SHAP heatmap. Clinicians (n = 206) evaluated comprehension and usability using objective questions and a Likert-based survey.

Results:

Based on 4608 critically ill patients with 10 medical variables over 7 hours of data for each patient, the Random Forest (RF) model achieved the highest area under the curve (AUC): 0.94. Furthermore, the local explanations were customized and evaluated by 206 clinicians through a survey conducted on the Prolific platform. A customized heatmap representation was selected as the visualization with the highest perceived clinical utility and alignment with clinical workflows.

Discussion:

The reported findings support the need for explanation formats to be tailored to clinical reasoning and task context, supporting the concept of cognitive fit. The heatmap’s close alignment with clinicians’ mental models and its graphical integrity enhances interpretability and trust. This study demonstrates that explanation effectiveness depends on contextual relevance, rather than a universal standard, and that the presentation format itself significantly shapes clinicians’ trust in XAI systems.

Conclusion:

This study advances clinical XAI by introducing a time-aware explanation framework for ICU intubation decisions. By integrating temporal trends with model reasoning, our visualizations closely align with clinicians’ cognitive workflows. Rigorous clinician-centered evaluation identified the dual-encoded SHAP heatmap as the most useful and workflow-compatible visualization, highlighting the importance of explanation design alongside predictive accuracy for clinical adoption.

背景：许多人工智能模型的复杂性阻碍了它们的临床应用，因为使用它们的临床医生并不认为它们是透明的。本研究通过设计和评估插管预测的直观视觉解释，解决了缺乏以临床为中心的可解释AI （XAI）界面的问题，验证了工作流兼容设计提高接受度的假设。目的：本研究比较了基于xai的插管预测的三种具有时间意识的视觉解释，并评估了它们在临床医生中的接受程度、理解程度和感知效用。方法：我们开发了机器学习模型来估计患者病情恶化的近期风险，这可能导致使用ICU时间序列数据进行机械插管。我们使用SHAP生成了全局和局部解释，并设计了三种定制的视觉格式——时间力图、时间条形图和双编码SHAP热图。临床医生（n = 206）使用客观问题和李克特调查评估理解和可用性。结果：基于4608例危重患者，10个医学变量，每个患者7小时的数据，随机森林（Random Forest， RF）模型的曲线下面积（AUC）最高，为0.94。此外，通过在多产平台上进行的调查，206名临床医生对当地的解释进行了定制和评估。选择自定义热图表示作为具有最高临床效用和与临床工作流程一致的可视化。讨论：报告的研究结果支持需要根据临床推理和任务背景量身定制解释格式，支持认知契合的概念。热图与临床医生的心理模型密切一致，其图形完整性增强了可解释性和信任度。本研究表明，解释的有效性取决于上下文相关性，而不是通用标准，并且演示格式本身显著地影响了临床医生对XAI系统的信任。结论：本研究通过引入ICU插管决策的时间意识解释框架来推进临床XAI。通过将时间趋势与模型推理相结合，我们的可视化与临床医生的认知工作流程紧密结合。严格的以临床医生为中心的评估确定了双编码的SHAP热图是最有用的和工作流程兼容的可视化，强调了解释设计和临床采用预测准确性的重要性。

{"title":"Clinician preferences for explainable AI in critical care: a comparative study of interpretable models and visualizations for intubation decision support","authors":"Tiantian Xian , Nikolay Mehandjiev , Panos Constantinides , Yu-wang Chen , Qudamah Quboa , Gareth Kitchen","doi":"10.1016/j.ijmedinf.2026.106287","DOIUrl":"10.1016/j.ijmedinf.2026.106287","url":null,"abstract":"<div><h3>Background:</h3><div>The complexity of many AI models hinders their clinical adoption because the clinicians using them do not regard them as transparent. This study addresses the lack of clinician-centered explainable AI (XAI) interfaces by designing and evaluating intuitive visual explanations for intubation prediction, testing the hypothesis that workflow-compatible designs enhance acceptance.</div></div><div><h3>Objective:</h3><div>This study compares three, time-aware, visual explanations for XAI-based intubation prediction and evaluate their acceptance, comprehension, and perceived utility among clinicians.</div></div><div><h3>Methods:</h3><div>We developed machine learning models to estimate the near-term risk of deterioration in the patient’s condition which may lead to mechanical intubation using ICU time-series data. We generated global and local explanations using SHAP and designed three customized visual formats—a temporal force plot, a temporal bar chart, and a dual-encoded SHAP heatmap. Clinicians (<em>n</em> = 206) evaluated comprehension and usability using objective questions and a Likert-based survey.</div></div><div><h3>Results:</h3><div>Based on 4608 critically ill patients with 10 medical variables over 7 hours of data for each patient, the Random Forest (RF) model achieved the highest area under the curve (AUC): 0.94. Furthermore, the local explanations were customized and evaluated by 206 clinicians through a survey conducted on the Prolific platform. A customized heatmap representation was selected as the visualization with the highest perceived clinical utility and alignment with clinical workflows.</div></div><div><h3>Discussion:</h3><div>The reported findings support the need for explanation formats to be tailored to clinical reasoning and task context, supporting the concept of cognitive fit. The heatmap’s close alignment with clinicians’ mental models and its graphical integrity enhances interpretability and trust. This study demonstrates that explanation effectiveness depends on contextual relevance, rather than a universal standard, and that the presentation format itself significantly shapes clinicians’ trust in XAI systems.</div></div><div><h3>Conclusion:</h3><div>This study advances clinical XAI by introducing a time-aware explanation framework for ICU intubation decisions. By integrating temporal trends with model reasoning, our visualizations closely align with clinicians’ cognitive workflows. Rigorous clinician-centered evaluation identified the dual-encoded SHAP heatmap as the most useful and workflow-compatible visualization, highlighting the importance of explanation design alongside predictive accuracy for clinical adoption.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"210 ","pages":"Article 106287"},"PeriodicalIF":4.1,"publicationDate":"2026-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146006657","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Rule-augmented constraint learning for semantic error detection in MIMIC-III knowledge graph 基于规则增强约束学习的MIMIC-III知识图语义错误检测

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-17 DOI: 10.1016/j.ijmedinf.2026.106297

Özge Noben, Ömer Durukan Kılıç, Tjitze Rienstra, Michel Dumontier, Remzi Celebi

High-quality, error-free data is essential for developing reliable data-driven models, particularly in clinical decision support systems where inaccurate predictions can have serious consequences. While KGs offer a structured and semantically rich representation for clinical data, ensuring their consistency and correctness remains a challenge. Existing rule mining techniques provide solutions for the automatic extraction of logical constraints from KGs, but they often produce redundant or clinically irrelevant rules, especially when dealing with numeric or categorical literals such as age or lab values. KG constraints—rules intended to capture implausible or conflicting facts in the KG—can be used to spot semantic errors: facts that might conform to the underlying schema but contradict domain knowledge. In this work, we propose a novel framework for constraint learning in clinical KGs that identifies and transforms high-confidence rules into clinically plausible constraints. We propose two approaches, based on class disjointness and literal clustering combined with rule mining. We validate the clinical relevance of these generated rules using expert-curated constraints and large language models (LLMs). The results on the MIMIC-III clinical dataset show that rule filtering based constraint learning effectively preserves clinically meaningful rules that align with established medical knowledge. For numeric data, we achieve reliable value groupings through our clustering-based method, and the rules derived from these groupings were validated by LLMs. Their outputs confirm the clinical relevance of a portion of those discovered rules. By providing interpretable and scalable solutions to semantic inconsistencies in KGs, this study contributes to increasing the KG trustworthiness and its clinical usability.

高质量、无差错的数据对于开发可靠的数据驱动模型至关重要，特别是在临床决策支持系统中，不准确的预测可能会产生严重后果。虽然KGs为临床数据提供了结构化和语义丰富的表示，但确保它们的一致性和正确性仍然是一个挑战。现有的规则挖掘技术为从KGs中自动提取逻辑约束提供了解决方案，但它们经常产生冗余或临床无关的规则，特别是在处理数字或分类文字（如年龄或实验室值）时。KG约束—旨在捕获KG中不可信或冲突事实的规则—可用于发现语义错误：可能符合底层模式但与领域知识相矛盾的事实。在这项工作中，我们提出了一个新的框架，用于临床KGs的约束学习，该框架识别并将高置信度规则转化为临床合理的约束。我们提出了两种方法，基于类脱节和文字聚类结合规则挖掘。我们使用专家策划的约束和大型语言模型（llm）验证这些生成规则的临床相关性。MIMIC-III临床数据集的结果表明，基于规则过滤的约束学习有效地保留了与已建立的医学知识相一致的临床有意义的规则。对于数值数据，我们通过基于聚类的方法实现了可靠的值分组，并通过llm验证了从这些分组中导出的规则。他们的结果证实了这些发现的规则的一部分的临床相关性。通过提供可解释和可扩展的解决方案来解决KG的语义不一致，本研究有助于提高KG的可信度和临床可用性。

{"title":"Rule-augmented constraint learning for semantic error detection in MIMIC-III knowledge graph","authors":"Özge Noben, Ömer Durukan Kılıç, Tjitze Rienstra, Michel Dumontier, Remzi Celebi","doi":"10.1016/j.ijmedinf.2026.106297","DOIUrl":"10.1016/j.ijmedinf.2026.106297","url":null,"abstract":"<div><div>High-quality, error-free data is essential for developing reliable data-driven models, particularly in clinical decision support systems where inaccurate predictions can have serious consequences. While KGs offer a structured and semantically rich representation for clinical data, ensuring their consistency and correctness remains a challenge. Existing rule mining techniques provide solutions for the automatic extraction of logical constraints from KGs, but they often produce redundant or clinically irrelevant rules, especially when dealing with numeric or categorical literals such as age or lab values. KG constraints—rules intended to capture implausible or conflicting facts in the KG—can be used to spot semantic errors: facts that might conform to the underlying schema but contradict domain knowledge. In this work, we propose a novel framework for constraint learning in clinical KGs that identifies and transforms high-confidence rules into clinically plausible constraints. We propose two approaches, based on class disjointness and literal clustering combined with rule mining. We validate the clinical relevance of these generated rules using expert-curated constraints and large language models (LLMs). The results on the MIMIC-III clinical dataset show that rule filtering based constraint learning effectively preserves clinically meaningful rules that align with established medical knowledge. For numeric data, we achieve reliable value groupings through our clustering-based method, and the rules derived from these groupings were validated by LLMs. Their outputs confirm the clinical relevance of a portion of those discovered rules. By providing interpretable and scalable solutions to semantic inconsistencies in KGs, this study contributes to increasing the KG trustworthiness and its clinical usability.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"210 ","pages":"Article 106297"},"PeriodicalIF":4.1,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146006656","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing diabetes monitoring systems’ reports: A novel integrated diabetes report (IDR) 加强糖尿病监测系统报告：一种新的糖尿病综合报告（IDR）。

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-17 DOI: 10.1016/j.ijmedinf.2026.106288

Tahmineh Aldaghi , Robert Bem , Jan Muzik

Aim

Individuals with diabetes require continuous self-management. Diabetes monitoring systems generate structured reports that help individuals and healthcare providers interpret data and optimize treatment strategies. To design and validate an Integrated Diabetes Report (IDR) that improves the clarity, usability, and clinical relevance of diabetes data visualizations.

Method

A review of 13 diabetes monitoring systems revealed five main report categories: overlay, logbook, device-specific, daily, and overview reports. While the overview report was the most frequently used, it lacked comprehensive visualization and essential clinical metrics. To address these gaps, a multidisciplinary panel of four experts collaborated to design a more integrated reporting framework.

Results

Across systems, glucose statistics were included in all reports, followed by insulin data (in 12 systems), carbohydrate intake (in 6 systems), hypo-hyperglycemic indices (in 2 systems), sleep indices (in 2 systems), and medication details (in 1 system). Key gaps included minimal data on physical activity, limited documentation of carbohydrates, and the absence of consolidated insulin visualization. The IDR introduces a complications section, an integrated graph combining AGP with basal and bolus insulin, and an advanced insulin profile comparing seven calculated indices.

Conclusion

The IDR improves clinical interpretation, supports treatment decisions, and enhances risk assessment for diabetes management.

目的：糖尿病患者需要持续的自我管理。糖尿病监测系统生成结构化报告，帮助个人和医疗保健提供者解释数据并优化治疗策略。设计并验证糖尿病综合报告（IDR），以提高糖尿病数据可视化的清晰度、可用性和临床相关性。方法：对13个糖尿病监测系统的回顾揭示了五种主要报告类别：覆盖报告、日志报告、特定设备报告、每日报告和概述报告。虽然概述报告是最常用的，但它缺乏全面的可视化和必要的临床指标。为了解决这些差距，一个由四名专家组成的多学科小组合作设计了一个更加综合的报告框架。结果：在各个系统中，所有报告均包含葡萄糖统计数据，其次是胰岛素数据（12个系统）、碳水化合物摄入量（6个系统）、低血糖指数（2个系统）、睡眠指数（2个系统）和用药细节（1个系统）。主要的差距包括：关于身体活动的数据很少，关于碳水化合物的记录有限，以及缺乏整合的胰岛素可视化。IDR引入了并发症部分，将AGP与基础胰岛素和大剂量胰岛素结合起来的综合图表，以及比较七个计算指标的高级胰岛素概况。结论：IDR改善了临床解释，支持了治疗决策，并加强了糖尿病管理的风险评估。

{"title":"Enhancing diabetes monitoring systems’ reports: A novel integrated diabetes report (IDR)","authors":"Tahmineh Aldaghi , Robert Bem , Jan Muzik","doi":"10.1016/j.ijmedinf.2026.106288","DOIUrl":"10.1016/j.ijmedinf.2026.106288","url":null,"abstract":"<div><h3>Aim</h3><div>Individuals with diabetes require continuous self-management. Diabetes monitoring systems generate structured reports that help individuals and healthcare providers interpret data and optimize treatment strategies. To design and validate an Integrated Diabetes Report (IDR) that improves the clarity, usability, and clinical relevance of diabetes data visualizations.</div></div><div><h3>Method</h3><div>A review of 13 diabetes monitoring systems revealed five main report categories: overlay, logbook, device-specific, daily, and overview reports. While the overview report was the most frequently used, it lacked comprehensive visualization and essential clinical metrics. To address these gaps, a multidisciplinary panel of four experts collaborated to design a more integrated reporting framework.</div></div><div><h3>Results</h3><div>Across systems, glucose statistics were included in all reports, followed by insulin data (in 12 systems), carbohydrate intake (in 6 systems), hypo-hyperglycemic indices (in 2 systems), sleep indices (in 2 systems), and medication details (in 1 system). Key gaps included minimal data on physical activity, limited documentation of carbohydrates, and the absence of consolidated insulin visualization. The IDR introduces a complications section, an integrated graph combining AGP with basal and bolus insulin, and an advanced insulin profile comparing seven calculated indices.</div></div><div><h3>Conclusion</h3><div>The IDR improves clinical interpretation, supports treatment decisions, and enhances risk assessment for diabetes management.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"209 ","pages":"Article 106288"},"PeriodicalIF":4.1,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146013318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Beyond binary diagnosis: Key questions on AI accuracy, real-world applicability, and safety in clinical decision support 超越二元诊断：人工智能准确性、现实世界适用性和临床决策支持安全性的关键问题。

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-17 DOI: 10.1016/j.ijmedinf.2026.106292

Jin Ye

This comment relates to Kücking et al.’s (2026) study on the bidirectional effects of artificial intelligence recommendations and healthcare provider related factors on the accuracy of wound impregnation diagnosis. While acknowledging the valuable contributions of this research, including distinguishing between correct/incorrect artificial intelligence outputs, rigorous simulation design, and emphasis on clinical safety, we have raised key questions to enhance the interpretation of results and real-world translation. The main focuses include the moderating role of artificial intelligence system accuracy in automation bias, external effectiveness in real clinical environments, potential mechanisms for gender differences in diagnostic performance, the impact of visual cue design on decision-making, and the potential of explainable artificial intelligence (XAI) in risk mitigation. This review aims to promote further research and facilitate the safe and effective integration of artificial intelligence based clinical decision support systems (CDSS) into clinical practice.

这一评论涉及k cking等人（2026）关于人工智能推荐和医疗保健提供者相关因素对伤口浸渍诊断准确性的双向影响的研究。在承认这项研究的宝贵贡献的同时，包括区分正确/不正确的人工智能输出，严格的模拟设计，以及对临床安全性的强调，我们提出了一些关键问题，以加强对结果的解释和现实世界的翻译。主要重点包括人工智能系统准确性在自动化偏差中的调节作用，真实临床环境中的外部有效性，诊断表现性别差异的潜在机制，视觉线索设计对决策的影响，以及可解释人工智能（XAI）在风险缓解中的潜力。本文综述旨在促进进一步的研究，促进基于人工智能的临床决策支持系统（CDSS）安全有效地整合到临床实践中。

引用次数: 0

Less time Coding, more time Caring: Performance evaluation of ChatGPT-5 for ICD-10 coding of radiology reports 少时间编码，多时间关怀：ChatGPT-5对放射学报告ICD-10编码的性能评价

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-17 DOI: 10.1016/j.ijmedinf.2026.106296

Tristan Ruhwedel , Julian M.M. Rogasch , Paul Martin Dahlke , Seyd Shnayien , Christian Furth , Christoph Wetz , Holger Amthauer , Imke Schatka , Nick Lasse Beetz

Introduction

Worldwide radiologists are facing a high administrative workload. ICD-10 coding is mandatory for reimbursement in many health systems and a frequent source of billing errors. Large language models have shown promise in supporting coding related tasks, but previous studies with earlier ChatGPT versions reported mixed results and evidence specific to radiology reports remains scarce. We therefore aimed to investigate whether ChatGPT-5 can be consulted when assigning ICD-10 codes to radiology reports and whether this leads to a measurable time advantage.

Methods

2,738 fictious radiology reports across multiple modalities were derived from the PARROT database. Additionally, 100 fictitious PET/CT reports were created. Each report was assigned a single, most relevant ICD-10 code using ChatGPT-5. For PARROT, ChatGPT-derived codes were compared with predefined database reference labels. For PET/CT, ChatGPT-derived codes were compared with codes assigned by an independent manual coder. Exact and character-level concordance were assessed. In cases of discordance, a blinded adjudicator selected the most accurate ICD-10 code. Coding efficiency was evaluated for PET/CT reports by measuring coding time per report.

Results

For PARROT, exact-code concordance was 1,590/2,738 (58.1 %). In a random subset of 200 mismatches, blinded adjudication preferred the ChatGPT derived code in 123 and the reference label in 77 cases (p = 0.0015). Coding non-English reports resulted in significantly lower concordance (first character: p = 0.002; second/third characters: p < 0.001; last characters: p = 0.012) and longer coding times than English reports (p = 0.002). Regarding PET/CT reports, median coding time was 8 s with ChatGPT and 135 s without. The median time saved was 127 s per report.

Conclusion

Applied to daily clinical care, higher code correctness might reduce billing errors, while saved time could be reallocated to patient care. Radiologists should collaborate with developers to create versions of LLMs that operate within data-secure environments.

世界各地的放射科医生都面临着很高的行政工作量。在许多卫生系统中，ICD-10编码是报销的强制性规定，也是账单错误的常见来源。大型语言模型在支持编码相关任务方面显示出了希望，但是先前对早期ChatGPT版本的研究报告了混合的结果，并且针对放射学报告的证据仍然很少。因此，我们的目的是研究在将ICD-10代码分配给放射学报告时是否可以咨询ChatGPT-5，以及这是否会带来可测量的时间优势。方法从PARROT数据库中提取2,738份不同模式的虚构放射学报告。此外，还创建了100个虚构的PET/CT报告。每个报告使用ChatGPT-5分配一个最相关的ICD-10代码。对于PARROT， chatgpt衍生的代码与预定义的数据库参考标签进行了比较。对于PET/CT， chatgpt衍生代码与独立手动编码器分配的代码进行比较。准确和字符水平的一致性进行了评估。在不一致的情况下，盲法裁判选择最准确的ICD-10代码。通过测量每个报告的编码时间来评估PET/CT报告的编码效率。结果PARROT的准确编码一致性为1590 / 2738（58.1%）。在200个不匹配的随机子集中，盲法判决倾向于123例ChatGPT衍生代码和77例参考标签（p = 0.0015）。编码非英语报告的一致性显著低于英语报告（第一个字符：p = 0.002；第二/第三个字符：p <； 0.001；最后一个字符：p = 0.012），编码时间较长（p = 0.002）。关于PET/CT报告，ChatGPT的中位编码时间为8秒，未ChatGPT的中位编码时间为135秒。每个报告节省的平均时间为127秒。结论应用于临床日常护理中，提高编码正确性可减少计费错误，节省的时间可重新分配给患者护理。放射科医生应该与开发人员合作，创建在数据安全环境中运行的llm版本。

{"title":"Less time Coding, more time Caring: Performance evaluation of ChatGPT-5 for ICD-10 coding of radiology reports","authors":"Tristan Ruhwedel , Julian M.M. Rogasch , Paul Martin Dahlke , Seyd Shnayien , Christian Furth , Christoph Wetz , Holger Amthauer , Imke Schatka , Nick Lasse Beetz","doi":"10.1016/j.ijmedinf.2026.106296","DOIUrl":"10.1016/j.ijmedinf.2026.106296","url":null,"abstract":"<div><h3>Introduction</h3><div>Worldwide radiologists are facing a high administrative workload. ICD-10 coding is mandatory for reimbursement in many health systems and a frequent source of billing errors. Large language models have shown promise in supporting coding related tasks, but previous studies with earlier ChatGPT versions reported mixed results and evidence specific to radiology reports remains scarce. We therefore aimed to investigate whether ChatGPT-5 can be consulted when assigning ICD-10 codes to radiology reports and whether this leads to a measurable time advantage.</div></div><div><h3>Methods</h3><div>2,738 fictious radiology reports across multiple modalities were derived from the PARROT database. Additionally, 100 fictitious PET/CT reports were created. Each report was assigned a single, most relevant ICD-10 code using ChatGPT-5. For PARROT, ChatGPT-derived codes were compared with predefined database reference labels. For PET/CT, ChatGPT-derived codes were compared with codes assigned by an independent manual coder. Exact and character-level concordance were assessed. In cases of discordance, a blinded adjudicator selected the most accurate ICD-10 code. Coding efficiency was evaluated for PET/CT reports by measuring coding time per report.</div></div><div><h3>Results</h3><div>For PARROT, exact-code concordance was 1,590/2,738 (58.1 %). In a random subset of 200 mismatches, blinded adjudication preferred the ChatGPT derived code in 123 and the reference label in 77 cases (p = 0.0015). Coding non-English reports resulted in significantly lower concordance (first character: p = 0.002; second/third characters: p < 0.001; last characters: p = 0.012) and longer coding times than English reports (p = 0.002). Regarding PET/CT reports, median coding time was 8 s with ChatGPT and 135 s without. The median time saved was 127 s per report.</div></div><div><h3>Conclusion</h3><div>Applied to daily clinical care, higher code correctness might reduce billing errors, while saved time could be reallocated to patient care. Radiologists should collaborate with developers to create versions of LLMs that operate within data-secure environments.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"210 ","pages":"Article 106296"},"PeriodicalIF":4.1,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Biometric Data in Post-Traumatic Stress Disorder Detection: A Scoping Review of Digital Health Applications. 创伤后应激障碍检测中的生物特征数据：数字健康应用的范围审查。

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-15 DOI: 10.1016/j.ijmedinf.2026.106289

Phue Thet Khaing, Masaharu Nakayama

Context: Post-traumatic stress disorder (PTSD) is mainly assessed through self-reports and clinician interviews, which can delay recognition and limit reach. Biometric markers captured using digital technologies may enable earlier and more objective detections.

Purpose: To map biometric modalities used for PTSD detection in digital health, identify underused markers, characterise machine learning (ML)/artificial intelligence (AI) approaches, and assess sex-related analyses.

Methods: Guided by PRISMA-ScR, a protocol on the Open Science Framework was pre-registered and searches in PubMed, IEEE Xplore, and Google Scholar (2015-2025) were conducted. The full search string was: ("post-traumatic stress disorder" OR "PTSD") AND ("biometric data" OR "biosensor" OR "wearable technology") AND ("detection" OR "screening" OR "diagnosis" OR "monitoring") AND ("digital health" OR "mobile health" OR "AI-based" OR "machine learning"). Peer-reviewed human studies using biometric data with digital tools and/or ML/AI for PTSD detection were eligible. Of 3,312 records, 89 underwent full-text review, and 18 studies met the inclusion criteria.

Analysis: Data were categorised by biometric modality, digital platform (wearable devices, mobile applications, ML/AI systems), study population, and performance metrics (area under the curve, sensitivity, specificity). Findings were grouped thematically (physiological, neuroimaging, behavioural, genetic, multimodal) and synthesised narratively to identify trends, gaps, and the application of sex-stratified modelling.

Results: Most studies focused on physiological (e.g., heart rate, sleep) and neuroimaging (functional magnetic resonance imaging, electroencephalography) signals; behavioural and genetic modalities were underexplored. Data were frequently captured via wearables and mobile platforms, with ML commonly applied. Performance reporting was uneven, sex-stratified analyses were rare, and several promising modalities (e.g., eye-tracking, electrodermal activity) remain underused.

Conclusion: Digital biometric approaches can detect PTSD; however, progress has been slowed by heterogeneous study designs, inconsistent reporting, and limited attention to sex differences. Establishing common reporting standards, evaluating multimodal models in real-world settings, and developing algorithms incorporating sex for more equitable screening are warranted.

背景：创伤后应激障碍（PTSD）的评估主要通过自我报告和临床医生访谈，这可能会延迟识别和限制到达。使用数字技术捕获的生物特征标记可以实现更早和更客观的检测。目的：绘制用于数字健康中PTSD检测的生物识别模式，识别未充分利用的标记，表征机器学习(ML)/人工智能（AI）方法，并评估与性别相关的分析。方法：在PRISMA-ScR的指导下，预注册开放科学框架协议，并在PubMed、IEEE Xplore和谷歌Scholar（2015-2025）中进行检索。完整的搜索字符串是：（“创伤后应激障碍”或“PTSD”）和（“生物特征数据”或“生物传感器”或“可穿戴技术”）和（“检测”或“筛查”或“诊断”或“监测”）和（“数字健康”或“移动健康”或“基于人工智能”或“机器学习”）。使用生物特征数据与数字工具和/或ML/AI进行创伤后应激障碍检测的同行评审人类研究符合条件。在3312项记录中，89项进行了全文审查，18项研究符合纳入标准。分析：根据生物识别模式、数字平台（可穿戴设备、移动应用程序、ML/AI系统）、研究人群和性能指标（曲线下面积、灵敏度、特异性）对数据进行分类。研究结果按主题分组（生理、神经影像学、行为、遗传、多模态），并以叙事方式综合，以确定趋势、差距和性别分层模型的应用。结果：大多数研究集中在生理（如心率、睡眠）和神经影像学（功能磁共振成像、脑电图）信号；行为和遗传模式尚未得到充分探索。数据经常通过可穿戴设备和移动平台捕获，通常使用ML。绩效报告不平衡，性别分层分析很少，一些有前途的模式（如眼动追踪，皮肤电活动）仍未得到充分利用。结论：数字生物识别方法可以检测创伤后应激障碍；然而，异质性研究设计、不一致的报告以及对性别差异的关注有限，延缓了研究进展。有必要建立共同的报告标准，在现实环境中评估多模式模型，并开发包含性别的算法，以实现更公平的筛查。

{"title":"Biometric Data in Post-Traumatic Stress Disorder Detection: A Scoping Review of Digital Health Applications.","authors":"Phue Thet Khaing, Masaharu Nakayama","doi":"10.1016/j.ijmedinf.2026.106289","DOIUrl":"https://doi.org/10.1016/j.ijmedinf.2026.106289","url":null,"abstract":"<p><strong>Context: </strong>Post-traumatic stress disorder (PTSD) is mainly assessed through self-reports and clinician interviews, which can delay recognition and limit reach. Biometric markers captured using digital technologies may enable earlier and more objective detections.</p><p><strong>Purpose: </strong>To map biometric modalities used for PTSD detection in digital health, identify underused markers, characterise machine learning (ML)/artificial intelligence (AI) approaches, and assess sex-related analyses.</p><p><strong>Methods: </strong>Guided by PRISMA-ScR, a protocol on the Open Science Framework was pre-registered and searches in PubMed, IEEE Xplore, and Google Scholar (2015-2025) were conducted. The full search string was: (\"post-traumatic stress disorder\" OR \"PTSD\") AND (\"biometric data\" OR \"biosensor\" OR \"wearable technology\") AND (\"detection\" OR \"screening\" OR \"diagnosis\" OR \"monitoring\") AND (\"digital health\" OR \"mobile health\" OR \"AI-based\" OR \"machine learning\"). Peer-reviewed human studies using biometric data with digital tools and/or ML/AI for PTSD detection were eligible. Of 3,312 records, 89 underwent full-text review, and 18 studies met the inclusion criteria.</p><p><strong>Analysis: </strong>Data were categorised by biometric modality, digital platform (wearable devices, mobile applications, ML/AI systems), study population, and performance metrics (area under the curve, sensitivity, specificity). Findings were grouped thematically (physiological, neuroimaging, behavioural, genetic, multimodal) and synthesised narratively to identify trends, gaps, and the application of sex-stratified modelling.</p><p><strong>Results: </strong>Most studies focused on physiological (e.g., heart rate, sleep) and neuroimaging (functional magnetic resonance imaging, electroencephalography) signals; behavioural and genetic modalities were underexplored. Data were frequently captured via wearables and mobile platforms, with ML commonly applied. Performance reporting was uneven, sex-stratified analyses were rare, and several promising modalities (e.g., eye-tracking, electrodermal activity) remain underused.</p><p><strong>Conclusion: </strong>Digital biometric approaches can detect PTSD; however, progress has been slowed by heterogeneous study designs, inconsistent reporting, and limited attention to sex differences. Establishing common reporting standards, evaluating multimodal models in real-world settings, and developing algorithms incorporating sex for more equitable screening are warranted.</p>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"211 ","pages":"106289"},"PeriodicalIF":4.1,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146115094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

When artificial intelligence guides and misguides clinicians: A critical appraisal of AI recommendation correctness and diagnostic decision-making 当人工智能引导和误导临床医生：对人工智能推荐正确性和诊断决策的批判性评估。

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-14 DOI: 10.1016/j.ijmedinf.2026.106293

Hasan Nawaz Tahir , Anfal Khan , Muhammad Yousaf , Shahnila Javed , Muhammad Kamran Khan , Yousaf Ali

引用次数: 0

“Calibration or contamination?” Reassessing the evaluation of large language models for clinical mortality prediction “校准还是污染？”重新评估大型语言模型对临床死亡率预测的评价

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-14 DOI: 10.1016/j.ijmedinf.2026.106291

Zhihao Lei

引用次数: 0

Communicable diseases platform (CDP): Real-Time clinical analytics for infections 传染病平台（CDP）：感染的实时临床分析

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-12 DOI: 10.1016/j.ijmedinf.2026.106277

Manuri De Silva , Alice Voskoboynik , Sailavan Ramesh , Janice Campbell , Saravanan Satkumaran , Daryl R. Cheng

Objective

Communicable diseases, especially seasonal respiratory illnesses, contribute significantly to paediatric hospital presentations and admissions. Existing surveillance systems often require retrospective manual data collation and focus on either demographic or clinical data, not both. The Communicable Diseases Platform (CDP) is a dynamic data platform that aggregates both data types for all communicable disease presentations to The Royal Children’s Hospital Melbourne (RCH).

Methods

In the pilot phase, the CDP extracted de-identified aggregated data from hospital electronic medical records for patients with positive respiratory swabs. A dashboard displayed positivity rate and cumulative hospital admissions trends from 2016 to 2025, further filterable by pathogen, age, presentation type and interventions.

Discussion

The CDP improves understanding of clinical profiles, disease burden and seasonal patterns, supporting better outbreak control, patient flow prediction and clinical surveillance. Future developments include immunisation data integration and machine learning algorithm evaluation for real-time vaccine effectiveness estimations and communicable disease predictive modelling.

目的：传染性疾病，特别是季节性呼吸道疾病，是儿科就诊和住院的主要原因。现有的监测系统通常需要回顾性的人工数据整理，并侧重于人口统计或临床数据，而不是两者兼而有之。传染病平台（CDP）是一个动态数据平台，汇集了墨尔本皇家儿童医院（RCH）所有传染病报告的两种数据类型。方法在试点阶段，CDP从医院电子病历中提取呼吸道拭子阳性患者的去识别汇总数据。仪表板显示了2016年至2025年的阳性率和累计住院趋势，并进一步按病原体、年龄、表现类型和干预措施进行过滤。CDP提高了对临床概况、疾病负担和季节性模式的理解，支持更好的疫情控制、患者流量预测和临床监测。未来的发展包括免疫数据集成和机器学习算法评估，用于实时疫苗有效性估计和传染病预测建模。

{"title":"Communicable diseases platform (CDP): Real-Time clinical analytics for infections","authors":"Manuri De Silva , Alice Voskoboynik , Sailavan Ramesh , Janice Campbell , Saravanan Satkumaran , Daryl R. Cheng","doi":"10.1016/j.ijmedinf.2026.106277","DOIUrl":"10.1016/j.ijmedinf.2026.106277","url":null,"abstract":"<div><h3>Objective</h3><div>Communicable diseases, especially seasonal respiratory illnesses, contribute significantly to paediatric hospital presentations and admissions. Existing surveillance systems often require retrospective manual data collation and focus on either demographic or clinical data, not both. The Communicable Diseases Platform (CDP) is a dynamic data platform that aggregates both data types for all communicable disease presentations to The Royal Children’s Hospital Melbourne (RCH).</div></div><div><h3>Methods</h3><div>In the pilot phase, the CDP extracted de-identified aggregated data from hospital electronic medical records for patients with positive respiratory swabs. A dashboard displayed positivity rate and cumulative hospital admissions trends from 2016 to 2025, further filterable by pathogen, age, presentation type and interventions.</div></div><div><h3>Discussion</h3><div>The CDP improves understanding of clinical profiles, disease burden and seasonal patterns, supporting better outbreak control, patient flow prediction and clinical surveillance. Future developments include immunisation data integration and machine learning algorithm evaluation for real-time vaccine effectiveness estimations and communicable disease predictive modelling.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"209 ","pages":"Article 106277"},"PeriodicalIF":4.1,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Clinicians’ perspectives on electronic medical records use in diabetes outpatient Care: A qualitative study 临床医生对糖尿病门诊使用电子病历的看法：一项定性研究。

IF 4.1 2区医学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

International Journal of Medical Informatics

Pub Date : 2026-01-11 DOI: 10.1016/j.ijmedinf.2026.106275

Wenyong Wang , Mahnaz Samadbeik , Gaurav Puri , Donald S.A. McLeod , Elton Lobo , Tuan Duong , Titus Kirwa , Clair Sullivan

Background

Electronic Medical Records (EMRs) aim to improve efficiency, safety, and quality of care. However, the impact of EMR implementation, particularly in outpatient diabetes care, remains underexplored. This study explored clinicians’ perspectives on EMR use in diabetes outpatient care.

Methods

This qualitative study, conducted in line with COREQ guidelines, involved four focus groups with 22 clinicians (doctors, nurses, and allied health) at a metropolitan diabetes service in Queensland, Australia. Data were analysed using deductive content analysis, guided by the Quintuple Aim and Technology Acceptance Model/Unified Theory of Acceptance and Use of Technology frameworks.

Results

Clinicians reported mixed outcomes across the Quintuple Aim domains, shaped by technology adoption constructs. Facilitators such as improved efficiency, access to patient information, and prescribing safety reflected perceived usefulness and positive attitudes, contributing to favourable outcomes across multiple Quintuple Aim. Barriers such as navigation complexity, technical issues, alert fatigue, and overwhelming training led to negative outcomes in EMR use. Tensions around documentation practices and patient expectations of system use, resulted in mixed outcomes. Overall, clinicians viewed EMRs as essential, but sustained adoption required improved usability, tailored training, and better system integration.

Conclusion

This study concludes that while the EMRs improved safety, efficiency, and access to information, their design and implementation also introduced burdens that negatively affected clinician experience. EMRs significantly shape the healthcare workforce, influencing workflow, wellbeing, and professional engagement. In outpatient diabetes care, specific workflow challenges such as glycaemic data integration highlight that existing EMR designs may not fully support the complexity of chronic disease management. To maximise benefits, EMR initiatives should be approached as quality improvement activities, with role-specific training, reliable infrastructure, and clinician involvement in system optimisation. Future research should address usability challenges, enhance integration, and ensure that both clinician and patient perspectives guide digital health transformation.

背景：电子病历（EMRs）旨在提高医疗效率、安全性和质量。然而，EMR实施的影响，特别是在门诊糖尿病护理方面，仍未得到充分探讨。本研究探讨临床医生在糖尿病门诊医疗中使用电子病历的观点。方法：本定性研究按照COREQ指南进行，涉及澳大利亚昆士兰州一家大都市糖尿病服务中心的22名临床医生（医生、护士和专职健康人员）的四个焦点小组。在“五重目标”和“技术接受模型”/“技术接受与使用统一理论”框架的指导下，采用演绎内容分析对数据进行分析。结果：临床医生报告了五项目标领域的混合结果，这些结果受到技术采用结构的影响。提高效率、获取患者信息和处方安全等促进因素反映了感知到的有用性和积极态度，有助于在多个“五大目标”中取得有利结果。导航复杂性、技术问题、警报疲劳和压倒性的培训等障碍导致EMR使用的负面结果。文档实践和患者对系统使用的期望之间的紧张关系导致了不同的结果。总体而言，临床医生认为电子病历是必要的，但持续采用需要改进可用性、量身定制的培训和更好的系统集成。结论：本研究得出结论，虽然电子病历提高了安全性、效率和信息获取，但其设计和实施也带来了负担，对临床医生的体验产生了负面影响。电子病历极大地塑造了医疗保健人力，影响了工作流程、健康和专业参与度。在门诊糖尿病护理中，特定的工作流程挑战，如血糖数据整合，突出表明现有的电子病历设计可能无法完全支持慢性疾病管理的复杂性。为了最大限度地提高效益，电子病历计划应该作为质量改进活动，具有特定角色的培训、可靠的基础设施和临床医生参与系统优化。未来的研究应解决可用性挑战，加强整合，并确保临床医生和患者的观点都能指导数字健康转型。

{"title":"Clinicians’ perspectives on electronic medical records use in diabetes outpatient Care: A qualitative study","authors":"Wenyong Wang , Mahnaz Samadbeik , Gaurav Puri , Donald S.A. McLeod , Elton Lobo , Tuan Duong , Titus Kirwa , Clair Sullivan","doi":"10.1016/j.ijmedinf.2026.106275","DOIUrl":"10.1016/j.ijmedinf.2026.106275","url":null,"abstract":"<div><h3>Background</h3><div>Electronic Medical Records (EMRs) aim to improve efficiency, safety, and quality of care. However, the impact of EMR implementation, particularly in outpatient diabetes care, remains underexplored. This study explored clinicians’ perspectives on EMR use in diabetes outpatient care.</div></div><div><h3>Methods</h3><div>This qualitative study, conducted in line with COREQ guidelines, involved four focus groups with 22 clinicians (doctors, nurses, and allied health) at a metropolitan diabetes service in Queensland, Australia. Data were analysed using deductive content analysis, guided by the Quintuple Aim and Technology Acceptance Model/Unified Theory of Acceptance and Use of Technology frameworks.</div></div><div><h3>Results</h3><div>Clinicians reported mixed outcomes across the Quintuple Aim domains, shaped by technology adoption constructs. Facilitators such as improved efficiency, access to patient information, and prescribing safety reflected perceived usefulness and positive attitudes, contributing to favourable outcomes across multiple Quintuple Aim. Barriers such as navigation complexity, technical issues, alert fatigue, and overwhelming training led to negative outcomes in EMR use. Tensions around documentation practices and patient expectations of system use, resulted in mixed outcomes<strong>.</strong> Overall, clinicians viewed EMRs as essential, but sustained adoption required improved usability, tailored training, and better system integration.</div></div><div><h3>Conclusion</h3><div>This study concludes that while the EMRs improved safety, efficiency, and access to information, their design and implementation also introduced burdens that negatively affected clinician experience. EMRs significantly shape the healthcare workforce, influencing workflow, wellbeing, and professional engagement. In outpatient diabetes care, specific workflow challenges such as glycaemic data integration highlight that existing EMR designs may not fully support the complexity of chronic disease management. To maximise benefits, EMR initiatives should be approached as quality improvement activities, with role-specific training, reliable infrastructure, and clinician involvement in system optimisation. Future research should address usability challenges, enhance integration, and ensure that both clinician and patient perspectives guide digital health transformation.</div></div>","PeriodicalId":54950,"journal":{"name":"International Journal of Medical Informatics","volume":"209 ","pages":"Article 106275"},"PeriodicalIF":4.1,"publicationDate":"2026-01-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145960768","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0