Journal of the American Medical Informatics Association最新文献

Background and significance: Physician and advanced practice provider (APP) well-being is a critical focus in healthcare. Emerging technology such as generative artificial intelligence (GAI) scribes reduces physician and APP administrative burden created by electronic health records. Early adopters of this technology have demonstrated promising improvements in clinical documentation, well-being, and cognitive load. However, further exploration across professional roles is warranted.

Objective: The goal of this quality improvement initiative was to explore how GAI scribes impacted well-being, cognitive load, and practice efficiency among physicians and APPs across professional roles.

Methods: A cross-sectional anonymous survey was conducted prior to implementation of GAI scribe technology and 3 months after physicians and APPs were onboarded.

Results: Physicians and APPs showed a reduction in cognitive task load following scribe technology implementation. Physicians reported reduced burnout and intent to leave; however, APPs did not have a significant reduction in burnout or intent to leave.

Conclusion: Artificial intelligence scribe technology shows potential for improving well-being among physicians and APPs by reducing cognitive load and clinical documentation time. Although some differences were found, overall, the technology appears to hold promise across professional roles.

Objectives: Federated Ecosystems for Analytics and Standardized Technologies (FEAST) is a modular, cloud-based platform developed through the ARPA-H Biomedical Data Fabric initiative to enable secure, federated analysis of real-world biomedical data. To guide and iteratively refine its modular design, the FEAST team conducted a cross-institutional survey to systematically identify and prioritize research needs related to authorized-access data across diverse biomedical domains. This study presents a structured synthesis of submitted use cases to uncover infrastructure gaps, data integration challenges, and translational opportunities. The results from the survey inform both front-end user-facing functionality and backend data requirements, shaping how the interface supports user interactions, data types, and compliance with security and interoperability standards.

Materials and methods: A structured survey form was distributed to researchers affiliated with participating institutions, including DNA-HIVE, The George Washington University (GW-FEAST), Weill Cornell Medicine, Vanderbilt University Medical Center, Georgetown University, European Bioinformatics Institute, and Kaiser Permanente. Respondents completed standardized fields describing the data types of interest, project goals, analytic methods, and perceived technical barriers. The collected responses were curated and analyzed to identify common needs related to privacy, interoperability, scalability, and workflow reproducibility.

Results: The survey compiled 61 use cases spanning genomics, imaging, clinical phenotyping, EHR-driven analytics, and precision medicine. Common themes included the need for multi-modal data integration, HL7 FHIR-based secure access, federated model training without PII retention, and containerized microservices for scalable deployment. Convergent needs across institutions emphasized consistent demand for FAIR-compliant infrastructure and readiness for real-world data analytics.

Conclusion: The FEAST Use Cases survey provides a cross-sectional view of biomedical informatics priorities grounded in real-world data needs. The findings offer a strategic blueprint for developing federated, privacy-preserving infrastructure to support secure, collaborative, and scalable biomedical research.

Purpose: To highlight the importance of reporting negative results in large language model (LLM) research, particularly as these systems are increasingly integrated into healthcare.

Potential: LLMs offer transformative capabilities in text generation, summarization, and clinical decision support. Transparent documentation of both successes and failures can accelerate innovation, improve reproducibility, and guide safe deployment.

Caution: Publication bias toward positive findings conceals model limitations, biases, and reproducibility challenges. In healthcare, underreporting failures risks patient safety, ethical lapses, and wasted resources. Structural barriers, including a lack of standards and limited funding for failure analysis, perpetuate this cycle.

Conclusions: Negative results should be recognized as valuable contributions that delineate the boundaries of LLM applicability. Structured reporting, educational initiatives, and stronger incentives for transparency are essential to ensure responsible, equitable, and trustworthy use of LLMs in healthcare.

Objectives: Advancing health through informatics requires attending to justice. Recent policy changes in the United States have introduced significant barriers to promoting justice within informatics due to targeted funding cuts and hostility to science, especially science that prioritizes justice.

Materials and methods: We present five key principles for advancing a justice-oriented informatics agenda, synthesized from our workshop held at the American Medical Informatics Association 2022 Annual Symposium.

Results: These principles are: (1) Recognize knowledge and methodologies across communities; (2) Acknowledge historical and cultural contexts of interactions; (3) Facilitate transparency and accountability through clear measures and metrics; (4) Foster trust and sustainability; and (5) Equitably allocate compensation and resources.

Discussion and conclusion: We discuss barriers to implementing these principles that have arisen since the 2022 workshop and provide recommendations for moving towards justice-oriented informatics. We offer examples of how these principles may be used to frame challenges and adapt to new barriers within BMI.

Background: The use of generative large language models (LLMs) with electronic health record (EHR) data is rapidly expanding to support clinical and research tasks. This systematic review characterizes the clinical fields and use cases that have been studied and evaluated to date.

Methods: We followed the Preferred Reporting Items for Systematic Review and Meta-Analyses guidelines to conduct a systematic review of articles from PubMed and Web of Science published between January 1, 2023, and November 9, 2024. Studies were included if they used generative LLMs to analyze real-world EHR data and reported quantitative performance evaluations. Through data extraction, we identified clinical specialties and tasks for each included article, and summarized evaluation methods.

Results: Of the 18 735 articles retrieved, 196 met our criteria. Most studies focused on radiology (26.0%), oncology (10.7%), and emergency medicine (6.6%). Regarding clinical tasks, clinical decision support made up the largest proportion of studies (62.2%), while summarizations and patient communications made up the smallest, at 5.6% and 5.1%, respectively. In addition, GPT-4 and GPT-3.5 were the most commonly used generative LLMs, appearing in 60.2% and 57.7% of studies, respectively. Across these studies, we identified 22 unique non-NLP metrics and 35 unique NLP metrics. While NLP metrics offer greater scalability, none demonstrated a strong correlation with gold-standard human evaluations.

Conclusion: Our findings highlight the need to evaluate generative LLMs on EHR data across a broader range of clinical specialties and tasks, as well as the urgent need for standardized, scalable, and clinically meaningful evaluation frameworks.

Objective: To characterize the nature and consequence(s) of interdependent physician electronic health record (EHR) work across inpatient shifts.

Materials and methods: Pooled cross-sectional analysis of EHR metadata associated with hospital medicine patients at an academic medical center, January-June 2022. Using patient-day observation data, we use a mixed effects regression model with daytime physician random effects to examine nightshift behavior (handoff time, total EHR time) as a function of behaviors by the preceding daytime team. We also assess whether nighttime patient deterioration is predicted by team coordination behaviors across shifts.

Results: We observed 19 671 patient days (N = 2708 encounters). Physicians used the handoff tool consistently, generally spending 8-12 minutes per shift editing patient information. When the day service team was more activated (highest tercile of handoff time, overall EHR time), nightshift experienced increased levels of EHR work and patient risk of overnight decline was elevated. (ie, Busy predicts busy). However, lower levels of dayshift activation were also associated with nightshift spillovers, including higher overnight EHR work and increased likelihood of patient clinical decline. Patient-days in the lowest and highest terciles of dayshift EHR time had a 1 percentage point increased relative risk of overnight decline (baseline prevalence of 4.4%) compared to the middle tercile (P = .04).

Discussion: We find evidence of spillovers in EHR work from dayshift to nightshift. Additionally, the lowest and highest levels of dayshift EHR activity are associated with increased risk of overnight patient decline. Results are associational and motivate further examination of additional confounding factors.

Conclusion: Analyses reveal opportunities to address task interdependence across shifts, using technology to flexibly shape and support collaborative teaming practices in complex clinical environments.

Objective: Accurate phenotyping is an essential task for researchers utilizing electronic health record (EHR)-linked biobank programs like the All of Us Research Program to study human genetics. However, little guidance is available on how to select an EHR-based phenotyping procedure that maximizes downstream statistical power. This study aims to estimate accuracy of three phenotype definitions of ovarian, female breast, and colorectal cancers in All of Us (v7 release) and determine which is most likely to optimize downstream statistical power for genetic association testing.

Materials and methods: We used empirical carrier frequencies of deleterious variants in known risk genes to estimate the accuracy of each phenotype definition and compute statistical power after accounting for the probability of outcome misclassification.

Results: We found that the choice of phenotype definition can have a substantial impact on statistical power for association testing and that no approach was optimal across all tested diseases. The impact on power was particularly acute for rarer diseases and target risk alleles of moderate penetrance or low frequency. Additionally, our results suggest that the accuracy of higher-complexity phenotyping algorithms is inconsistent across Black and non-Hispanic White participants in All of Us, highlighting the potential for case ascertainment biases to impact downstream association testing.

Discussion: EHR-based phenotyping presents a bottleneck for maximizing power to detect novel risk alleles in All of Us, as well as a potential source of differential outcome misclassification that researchers should be aware of. We discuss the implications of this as well as potential mitigation strategies.

Using 2023-2024 U.S. National Health Interview Survey data, we found that digital health literacy (dHL) mediated nearly half of the difference in telehealth use between Latino adults with non-English and English language preference. These findings identify dHL as a modifiable mechanism linking linguistic and digital access barriers, underscoring the need for multilingual, inclusive, and equitable telehealth design.

Objective: Healthcare decisions are increasingly made with the assistance of machine learning (ML). ML has been known to have unfairness-inconsistent outcomes across subpopulations. Clinicians interacting with these systems can perpetuate such unfairness by overreliance. Recent work exploring ML suppression-silencing predictions based on auditing the ML-shows promise in mitigating performance issues originating from overreliance. This study aims to evaluate the impact of suppression on collaboration fairness and evaluate ML uncertainty as desiderata to audit the ML.

Materials and methods: We used data from the Vanderbilt University Medical Center electronic health record (n = 58 817) and the MIMIC-IV-ED dataset (n = 363 145) to predict likelihood of death or intensive care unit transfer and likelihood of 30-day readmission using gradient-boosted trees and an artificially high-performing oracle model. We derived clinician decisions directly from the dataset and simulated clinician acceptance of ML predictions based on previous empirical work on acceptance of clinical decision support alerts. We measured performance as area under the receiver operating characteristic curve and algorithmic fairness using absolute averaged odds difference.

Results: When the ML outperforms humans, suppression outperforms the human alone (P < 8.2 × 10-6) and at least does not degrade fairness. When the human outperforms the ML, the human is either fairer than suppression (P < 8.2 × 10-4) or there is no statistically significant difference in fairness. Incorporating uncertainty quantification into suppression approaches can improve performance.

Conclusion: Suppression of poor-quality ML predictions through an auditor model shows promise in improving collaborative human-AI performance and fairness.

Background: Despite rapid integration into clinical decision-making, clinical large language models (LLMs) face substantial translational barriers due to insufficient structural characterization and limited external validation.

Objective: We systematically map the clinical LLM research landscape to identify key structural patterns influencing their readiness for real-world clinical deployment.

Methods: We identified 73 clinical LLM studies published between January 2020 and March 2025 using a structured evidence-mapping approach. To ensure transparency and reproducibility in study selection, we followed key principles from the PRISMA 2020 framework. Each study was categorized by clinical task, base architecture, alignment strategy, data type, language, study design, validation methods, and evaluation metrics.

Results: Studies often addressed multiple early stage clinical tasks-question answering (56.2%), knowledge structuring (31.5%), and disease prediction (43.8%)-primarily using text data (52.1%) and English-language resources (80.8%). GPT models favored retrieval-augmented generation (43.8%), and LLaMA models consistently adopted multistage pretraining and fine-tuning strategies. Only 6.9% of studies included external validation, and prospective designs were observed in just 4.1% of cases, reflecting significant gaps in translational reliability. Evaluations were predominantly quantitative only (79.5%), though qualitative and mixed-method approaches are increasingly recognized for assessing clinical usability and trustworthiness.

Conclusion: Clinical LLM research remains exploratory, marked by limited generalizability across languages, data types, and clinical environments. To bridge this gap, future studies must prioritize multilingual and multimodal training, prospective study designs with rigorous external validation, and hybrid evaluation frameworks combining quantitative performance with qualitative clinical usability metrics.