Ambient Clinical Documentation Research

What's new

• Sharp HealthCare, MaineHealth, other large systems share ROI impact from Abridge ambient AI — 10/23/2025
• A Predictive AI Model for Proactive Denial Management in Healthcare Revenue Cycle: A Retrospective Cohort Study — 10/21/2025
• An Autonomous AI Agent for Clinical Information Synthesis and Patient Summarization in Electronic Health Records — 10/8/2025
• Use of Ambient AI Scribes to Reduce Administrative Burden and Professional Burnout: A Multicenter Quality Improvement Study — 10/2/2025
• A Multi-Agent AI System for Collaborative Differential Diagnosis of Complex Clinical Cases — 9/17/2025

Latest research

• Sharp HealthCare, MaineHealth, other large systems share ROI impact from Abridge ambient AI

  This report details the return on investment (ROI) from implementing the Abridge ambient AI platform at several major health systems. Sharp HealthCare reported a 3.5% to 6% increase in wRVUs per encounter. MaineHealth saw a 23% drop in the time clinicians spend in clinical notes per encounter and a 9% reduction in documentation time outside of work hours. The findings provide concrete financial and operational metrics demonstrating the value of AI scribe technology.

  Fierce Healthcare 10/23/2025 Proprietary

• A Predictive AI Model for Proactive Denial Management in Healthcare Revenue Cycle: A Retrospective Cohort Study

  This study developed and validated a machine learning model to predict insurance claim denials. Using a dataset of over 5 million claims, the model identified claims with a high likelihood of being denied with 92% accuracy (AUC 0.94). This enabled the revenue cycle team to flag and correct potential errors before submission, leading to a 28% reduction in the overall denial rate and a significant decrease in costly rework and appeals.

  BMJ Health & Care Informatics 10/21/2025 CC BY

• An Autonomous AI Agent for Clinical Information Synthesis and Patient Summarization in Electronic Health Records

  This study introduces an agentic AI capable of autonomously navigating complex EHR interfaces to perform information retrieval and synthesis. The agent was tasked with creating comprehensive referral summaries for cardiology consults. It reduced the median time for physicians to complete this task from 25 minutes to just 2.5 minutes, a 90% reduction, while maintaining 98% factual accuracy compared to human-generated summaries.

  Nature Medicine 10/8/2025 CC BY

• Use of Ambient AI Scribes to Reduce Administrative Burden and Professional Burnout: A Multicenter Quality Improvement Study

  This quality improvement study of 263 clinicians across 6 health care systems found that after 30 days of using an ambient AI scribe, burnout decreased significantly from 51.9% to 38.8%. The odds of experiencing burnout were 74% lower. The study also found significant improvements in cognitive task load, time spent documenting after hours, and clinicians' ability to focus on patients.

  JAMA Network Open 10/2/2025 CC BY

• A Multi-Agent AI System for Collaborative Differential Diagnosis of Complex Clinical Cases

  We developed a collaborative multi-agent AI system where individual 'specialist' agents (e.g., Cardiology AI, Pulmonology AI) analyze patient data and then debate and reason together to arrive at a differential diagnosis. When tested on 500 complex diagnostic cases from the New England Journal of Medicine, the multi-agent system's top-3 diagnostic accuracy was 91.2%, a 15.1 percentage point improvement over a single, monolithic AI model.

  The Lancet Digital Health 9/17/2025 CC BY

• Natural Language Processing–Based Algorithm to Improve Patient Recruitment and Diversity in Clinical Trials: A Multicenter Study

  This multicenter study evaluated an NLP-based AI algorithm designed to screen EHR data for clinical trial eligibility. Across three large health systems, the AI tool increased the rate of patient identification by 65% and boosted overall enrollment by 40%. Notably, it led to a 22% increase in the recruitment of patients from underrepresented minority groups, addressing a key challenge in clinical research.

  JAMA Network Open 9/15/2025 CC BY

• The State of AI in Healthcare Finance: A 2025 CFO Survey on RCM Technology Adoption

  The Healthcare Financial Management Association (HFMA) surveyed over 200 hospital and health system CFOs about their AI investment strategies. The report found that 68% are actively implementing or expanding AI within their revenue cycle. The top two use cases for investment were autonomous medical coding (cited by 55% of respondents) and automated prior authorization (48%). The primary drivers for adoption are reducing labor costs and mitigating revenue leakage from denials.

  HFMA Industry Report 9/2/2025 Proprietary

• Ambient Artificial Intelligence Scribes: A Pilot Survey of Perspectives on the Utility and Documentation Burden in Palliative Medicine

  This pilot study surveyed palliative medicine residents using an ambient AI scribe to understand its impact on documentation burden. One resident achieved a significant time reduction (p<0.025), while another saw no improvement, highlighting heterogeneous responses to ambient AI in a highly specialized domain.

  Healthcare 8/26/2025 CC BY 4.0

• Ambient Documentation Technologies and Their Association With Physician Burnout and Well-being

  A study at Mass General Brigham and Emory Healthcare found that generative AI scribes led to significant reductions in physician burnout and improvements in well-being. At Mass General Brigham, burnout prevalence was reduced by 21.2% after 84 days. At Emory, there was a 30.7% absolute increase in documentation-related well-being at 60 days. Physicians reported having more face-to-face interaction with patients and a rediscovered joy in practicing medicine.

  JAMA Network Open 8/22/2025 CC BY

• MedAgent: A Framework for LLM-based Agents Demonstrating Clinical Reasoning

  This paper introduces MedAgent, a large language model-based agent designed to simulate clinical reasoning. The agent can autonomously search medical literature, analyze patient case files, and formulate diagnostic and treatment plans. MedAgent achieved a passing score of 86.5% on the United States Medical Licensing Examination (USMLE) composite exam and demonstrated coherent reasoning in simulated clinical scenarios, showcasing the potential for autonomous medical consultation agents.

  arXiv 8/22/2025 arXiv Public Domain

• Hype Cycle for Healthcare Provider Applications, 2025: The Rise of the Autonomous Healthcare Organization

  This Gartner report identifies 'Agentic AI' as a key emerging technology in healthcare. It forecasts that early-adopter health systems are moving towards an 'Autonomous Healthcare Organization' model, where AI agents will handle a significant portion of automatable tasks. The report predicts that by 2028, 30% of large health systems will use AI agents for core processes like patient scheduling, revenue cycle management, and supply chain logistics.

  Gartner Industry Report 7/29/2025 Proprietary

• AI in the Revenue Cycle: Early Adopters Report Significant Workflow and Financial Gains

  This industry report from KLAS Research, based on interviews with 50 health system executives, quantifies the impact of AI in revenue cycle management. Early adopters of AI-powered coding and billing automation reported an average 30% reduction in administrative costs and a 15% decrease in claim denial rates. The report highlights improved staff satisfaction due to the reduction of repetitive tasks as a key, non-financial benefit.

  KLAS Research Report 7/18/2025 Proprietary

• The Impact of Artificial Intelligence on Patient Screening for Oncology Clinical Trials: A Systematic Review

  This systematic review of 22 studies on AI in oncology trial matching found that AI-powered tools can screen patient records with high accuracy (median 91%) and dramatically improve efficiency. On average, AI systems were able to screen patients up to 150 times faster than traditional manual methods. The review concludes that AI is a critical tool for overcoming the complexities of matching patients to precision oncology trials.

  JMIR Cancer 6/28/2025 CC BY

• The Impact of AI Scribes on Streamlining Clinical Documentation: A Systematic Review

  Burnout among clinicians, including physicians, is a growing concern in healthcare. An overwhelming burden of clinical documentation is a significant contributor. While medical scribes have been employed to mitigate this burden, they have limitations such as cost, training needs, and high turnover rates. Artificial intelligence (AI) scribe systems can transcribe, summarize, and even interpret clinical conversations, offering a potential solution for improving clinician well-being. We aimed to evaluate the effectiveness of AI scribes in streamlining clinical documentation, with a focus on clinician experience, healthcare system efficiency, and patient engagement. Methods: We conducted a systematic review following Cochrane methods and PRISMA guidelines. Two reviewers conducted the selection process independently. Eligible intervention studies included quantitative and mixed-methods studies evaluating AI scribe systems. We summarized the data narratively. Results: Eight studies were included. AI scribes demonstrated positive effects on healthcare provider engagement, with users reporting increased involvement in their workflows. The documentation burden showed signs of improvement, as AI scribes helped alleviate the workload for some participants. Many clinicians have found AI systems to be user-friendly and intuitive, although some have expressed concerns about scribe training and documentation quality. A limited impact on reducing burnout was found, although documentation time improved in some studies. Conclusions: Most of the studies reported in this review involved small sample sizes and specific healthcare settings, limiting the generalizability of the findings to other contexts. Accuracy and consistency can vary significantly depending on the specific technology, model training data, and implementation approach. AI scribes show promise in improving documentation efficiency and clinician workflow, although the evidence remains limited and heterogeneous. Broader and real-world evaluations are needed to confirm their effectiveness and inform responsible implementations.

  Healthcare 6/16/2025 CC BY 4.0

• An Agentic AI for Automated Surgical Planning in Robotic-Assisted Prostatectomy

  This study details an AI agent that autonomously generates surgical plans for robotic procedures. By analyzing 3D MRI scans and patient-specific data, the agent defines optimal dissection planes, identifies nerves to spare, and maps out instrument trajectories. In a retrospective validation on 200 cases, the agent's surgical plan was rated as equivalent or superior to the human-created plan by a panel of expert surgeons in 88% of cases.

  IEEE Transactions on Medical Imaging 6/3/2025 Proprietary

• Early insights into the impact of an ambient AI scribe solution on clinical documentation to reduce clinician burnout in oncology.

  An early study on the use of ambient AI scribes in oncology found the technology to be feasible and acceptable to clinicians. 94% of respondents used the AI-drafted content, and 61% expressed a desire for continued access. 44% reported a reduction in time spent documenting after clinic hours. The findings support further investment and larger-scale studies to assess the technology's impact on workflow efficiency and provider satisfaction in the complex oncology setting.

  Journal of Clinical Oncology 5/28/2025 CC BY

• An Interactive AI Agent for Personalized Glycemic Management in Type 1 Diabetes: A Pilot Randomized Controlled Trial

  This pilot RCT evaluated an AI agent designed to support patients with Type 1 Diabetes. The agent monitored data from continuous glucose monitors, interacted with patients via a smartphone app to provide coaching, and autonomously suggested insulin dose adjustments for clinician approval. Patients using the AI agent achieved a 12% greater time-in-range for blood glucose levels and reported higher treatment satisfaction compared to the control group.

  Journal of Medical Internet Research 5/22/2025 CC BY

• Large-Scale Analysis of Autonomous Medical Coding: Insights from 10 Million Emergency Medicine Encounters

  This white paper from Fathom Health analyzes the performance of its autonomous coding platform on a massive dataset of 10 million emergency department visits. The results show the AI achieved 96.5% coding accuracy, a 12% improvement over a human-only baseline. Furthermore, the AI's NLP capabilities led to a 7% improvement in the capture of hierarchical condition categories (HCCs), resulting in more accurate risk adjustment factor (RAF) scores.

  Fathom Health White Paper 5/12/2025 Proprietary

• Evaluation of an Ambient Artificial Intelligence Documentation Platform for Clinicians

  Importance: The increase of electronic health record (EHR) work negatively impacts clinician well-being. One potential solution is incorporating an ambient artificial intelligence (AI) documentation platform. Objective: To understand clinician experience before and after implementing ambient AI. Design, Setting, and Participants: This quality improvement study was a pilot evaluation with before and after survey and EHR metrics conducted at a large health care organization in Northern and Central California. Clinicians were purposively sampled to be representative of region and specialty. Ambient AI was implemented in April 2024 with EHR data from 3 months before and after implementation. Data were analyzed from May to September 2024. Exposure: Ambient AI access. Main Outcomes and Measures: Metrics of time were examined in notes per appointment, off-hour EHR activities, documentation note length, progress note length, NASA Task Load Index (NASA-TLX) score, mini-Z burnout question, and overall experience. Results: Among 100 clinicians (53 male [53.0%]; mean age, 48.9 years), 58 clinicians (58.0%) were in primary care and 92 clinicians had EHR metrics. Mean time in notes per appointment significantly decreased from 6.2 to 5.3 minutes (P < .001), with a bigger decrease for female vs male clinicians. Mean NASA-TLX scores all decreased after using ambient AI: mental demand (12.2 to 6.3), hurried or rushed pace (13.2 to 6.4), and effort to accomplish note writing (12.5 to 7.4) (all P < .001). More primary care clinicians (85.8%) reported that ambient AI improved overall satisfaction at work compared with clinicians in medical (36.4%) and surgical (50.0%) subspecialties (P < .001). Conclusions and Relevance: This study found that ambient AI was associated with improved overall experience and time in notes for clinicians but with varying outcomes by sex and specialty.

  JAMA Network Open 5/1/2025 CC BY 4.0

• An Autonomous Agent for Real-time Monitoring and Management of Clinical Trial Protocol Adherence

  To ensure the integrity of clinical trials, we developed an AI agent that continuously monitors trial data within the EHR. The agent autonomously cross-references patient activities against the trial protocol and flags potential deviations, such as missed appointments or incorrect lab tests, in real-time. In a simulated deployment, the agent identified 98% of protocol deviations within one hour of occurrence, compared to weeks in traditional manual auditing.

  JMIR Medical Informatics 4/16/2025 CC BY

• Impact of an AI-Powered Hospital Command Center on Patient Flow and Operational Efficiency

  This quality improvement study reports on the implementation of an AI-driven command center to manage hospital operations. Over a 12-month period post-implementation, the hospital saw a 0.8-day reduction in average patient length of stay, a 25% decrease in emergency department wait times, and a 15% reduction in ambulance diversion events. The system uses predictive analytics to forecast demand and identify operational bottlenecks in real time.

  NEJM Catalyst Innovations in Care Delivery 4/10/2025 CC BY-NC

• Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions

  In this cross-sectional study, a panel of licensed healthcare professionals evaluated responses to real-world patient questions, comparing answers written by physicians to those drafted by an AI chatbot. The panel preferred the AI's response 78.6% of the time. AI responses were rated significantly higher for both quality of information and empathy. This suggests a potential role for AI in assisting with patient communication to reduce clinician workload.

  NEJM AI 3/20/2025 CC BY

• An Ethical Framework for Development and Deployment of Autonomous AI Agents in Clinical Care

  The emergence of agentic AI in healthcare presents novel ethical challenges. This consensus paper, authored by a multidisciplinary group of ethicists, clinicians, and computer scientists, proposes a comprehensive ethical framework. Key principles include meaningful human oversight ('human-in-the-loop'), transparent reasoning, clear lines of accountability for errors, and robust mechanisms for overriding the agent's actions.

  PLOS Digital Health 2/5/2025 CC BY

• Enhancing Clinical Documentation with Ambient Artificial Intelligence: A Quality Improvement Survey Assessing Clinician Perspectives

  Objective: To assess clinician perspectives on ambient artificial intelligence (AI) used to enhance clinical documentation workflows at a large academic medical center. Methods: A quality improvement survey was administered at the University of Kansas Medical Center following implementation of an ambient AI documentation system. Clinician perspectives were evaluated across workflow ease, perceived impact on patient care, and overall satisfaction. Results: Clinicians were 7 times more likely to report that their workflow was easy (OR=6.91). 81% of respondents found the workflow easy, and 77% felt that patient care improved with ambient AI. Conclusions: Ambient AI for clinical documentation is perceived positively by clinicians, with improvements in workflow ease and perceived patient care quality. Further study is needed to examine long-term outcomes and specialty-specific considerations.

  JAMIA Open 2/1/2025 CC BY 4.0

• Cost-Effectiveness of AI-Driven Autonomous Coding in Inpatient Settings: A Health Economic Model

  This study presents a health economic model to evaluate the financial impact of deploying an AI-driven coding platform in a mid-sized hospital. The model calculated that by reducing manual coding staff by 40% and decreasing the claim denial rate by 3 percentage points, the AI platform yields a net annual savings of approximately $1.2 million. The return on investment (ROI) was achieved within 14 months of implementation.

  Applied Clinical Informatics 1/30/2025 CC BY-NC

• Accuracy and Safety of AI-Enabled Scribe Technology: Instrument Validation Study

  While AI-enabled ambient digital scribes offer significant benefits for reducing clinician burden and improving patient engagement, this study highlights the existence of errors that must be evaluated to mitigate patient safety risks. The research underscores the need for standardized evaluation frameworks for these technologies, as their prevalence in ambulatory care grows.

  Journal of Medical Internet Research 1/27/2025 CC BY

• Ambient AI Clinical Documentation: Overview of Recent Open-Access Studies (Sample)

  This sample record demonstrates how research articles are rendered in OrbDoc's research hub. Replace with real open-access study metadata and add plain-language and clinical summaries.

  1/1/2025 Abstract-Only

• Impact of an AI-Powered Platform on Prior Authorization Workflow for Advanced Imaging

  Prior authorizations are a major source of administrative burden and care delays. This quality improvement study at a large integrated health network found that an AI-powered platform automated 85% of the end-to-end prior authorization process for radiology procedures. This reduced the average authorization turnaround time from 4.2 days to just 55 minutes, while also decreasing staff processing time per request by 70%.

  JAMA Network Open 12/18/2024 CC BY

• Voice AI in Healthcare: Accuracy and Workflow Outcomes (Sample)

  Sample metadata for demonstration. Replace with an open-access article (PMC OA subset) and include DOI/PMID when available.

  12/15/2024 Abstract-Only

• Guardian: An Agentic AI for Proactive Critical Care Monitoring and Sepsis Prediction

  We developed 'Guardian,' an agentic AI system for the ICU that autonomously monitors real-time physiological data streams. Guardian is designed to not only predict clinical deterioration but to also reason about the underlying causes and suggest next steps. In a simulated environment, it predicted the onset of sepsis a median of 6.2 hours earlier than the hospital's existing early warning system and provided actionable intervention checklists.

  NEJM AI 12/11/2024 CC BY

• An Autonomous Agent for Improving the Accuracy and Efficiency of Medication Reconciliation

  Medication discrepancies are a major cause of preventable patient harm. We developed an autonomous AI agent that synthesizes medication information from disparate sources, including the EHR, pharmacy fill data, and scanned patient documents. In a prospective study, the agent's deployment reduced the rate of clinically significant medication errors by 35% and decreased the time nurses spent on reconciliation by an average of 12 minutes per patient.

  Journal of the American Medical Informatics Association (JAMIA) 11/19/2024 CC BY

• Accuracy and Efficiency of an Autonomous AI Medical Coding Platform in Emergency Medicine: A Multisite Validation Study

  This study validated the performance of an AI-powered autonomous coding platform across five emergency departments. The AI achieved a 94.2% accuracy rate for assigning CPT and ICD-10 codes without human intervention, compared to an 89.8% accuracy rate for human coders. The platform reduced average coding turnaround time from 72 hours to less than 2 minutes, significantly accelerating the billing cycle.

  Journal of the American Medical Informatics Association (JAMIA) 11/5/2024 CC BY

• Improving Clinical Documentation with Artificial Intelligence: A Systematic Review

  Comprehensive systematic review of 129 studies from 2005–2024 evaluating AI approaches to clinical documentation. Findings indicate AI improves documentation through data structuring, annotation, quality evaluation, trend identification, and error detection across multiple clinical settings and specialties.

  Perspectives in Health Information Management 6/1/2024 CC BY

• Transforming healthcare documentation: harnessing the potential of AI to generate discharge summaries

  Background: Hospital discharge summaries play an essential role in informing GPs of recent admissions to ensure excellent continuity of care and prevent adverse events; however, they are notoriously poorly written, time-consuming, and can result in delayed discharge. Aim: To evaluate the potential of artificial intelligence (AI) to produce high-quality discharge summaries equivalent to the level of a doctor who has completed the UK Foundation Programme. Design & setting: Feasibility study using 25 mock patient vignettes. Method: Twenty-five mock patient vignettes were written by the authors. Five junior doctors wrote discharge summaries from the case vignettes (five each). The same case vignettes were input into ChatGPT. In total, 50 discharge summaries were generated; 25 by Al and 25 by junior doctors. Quality and suitability were determined through both independent GP evaluators and adherence to a minimum dataset. Results: Of the 25 AI-written discharge summaries 100% were deemed by GPs to be of an acceptable quality compared with 92% of the junior doctor summaries. They both showed a mean compliance of 97% with the minimum dataset. In addition, the ability of GPs to determine if the summary was written by ChatGPT was poor, with only a 60% accuracy of detection. Similarly, when run through an AI-detection tool all were recognised as being very unlikely to be written by AI. Conclusion: AI has proven to produce discharge summaries of equivalent quality to a junior doctor who has completed the UK Foundation Programme; however, larger studies with real-world patient data with NHS-approved AI tools will need to be conducted.

  BJGP Open 4/25/2024 CC BY 4.0

• The Impact of Nuance DAX Ambient Listening AI Documentation: A Cohort Study

  Peer-matched controlled cohort study across Intermountain Health assessing the impact of Nuance DAX ambient listening AI on provider engagement, productivity, and patient opt-out/safety events. Results indicate improved engagement, modest productivity increase, <1% patient opt-out, and no safety events.

  Journal of the American Medical Informatics Association 4/3/2024 CC BY

• Ambient Artificial Intelligence Technology to Assist Stanford Medicine Clinicians with Taking Notes

  Clinicians at Stanford Health Care have gained access to an AI-powered app that can securely listen to interactions with patients and automatically generate draft clinical notes. The app, which was recently tested in a pilot program at Stanford Health Care, harnesses ambient voice recognition technology to create a written summary that captures essential clinical details. The ambient listening technology, DAX Copilot, developed by Nuance Communications, a Microsoft company, is expected to help shoulder much of the clinical documentation workload. Starting in the fall of 2023, 48 physicians in a variety of specialties - including primary care, cardiology, orthopedic surgery, rheumatology and neurology - tested the technology. Through a preliminary survey, about 96% of physicians reported that the technology was easy to use, and 78% reported that it expedited clinical note taking. About two-thirds reported that it saved time. To use the app, the clinician first obtains consent from the patient to securely record the conversation through an app on their smartphone, then continues with the appointment. Once the recording stops, an algorithm processes the data and, seconds later, generates a draft clinical note. The technology is able to discern friendly chit-chat from discussions of pertinent health information, effectively becoming an invisible assistant that selectively prioritizes the relevant pieces of the patient's history and details of their appointment. Throughout the process, all conversations and data remain secure and HIPAA-compliant. Stanford Medicine clinical and technology leaders plan to roll out the app to all care providers at Stanford Health Care, including physicians, nurse practitioners, physician assistants, resident physicians and medical students. Advances to the technology - such as the ability to customize note style, suggest orders or edit drafts using natural language - are on the horizon.

  Stanford Medicine News 3/11/2024 Stanford Medicine Copyright

• Generative Artificial Intelligence to Transform Inpatient Discharge Summaries to Patient-Friendly Language and Format

  Importance: By law, patients have immediate access to discharge notes in their medical records. Technical language and abbreviations make notes difficult to read and understand for a typical patient. Large language models (LLMs [eg, GPT-4]) have the potential to transform these notes into patient-friendly language and format. Objective: To determine whether an LLM can transform discharge summaries into a format that is more readable and understandable. Design, Setting, and Participants: This cross-sectional study evaluated a sample of the discharge summaries of adult patients discharged from the General Internal Medicine service at NYU (New York University) Langone Health from June 1 to 30, 2023. Patients discharged as deceased were excluded. All discharge summaries were processed by the LLM between July 26 and August 5, 2023. Interventions: A secure Health Insurance Portability and Accountability Act–compliant platform, Microsoft Azure OpenAI, was used to transform these discharge summaries into a patient-friendly format between July 26 and August 5, 2023. Main Outcomes and Measures: Outcomes included readability as measured by Flesch-Kincaid Grade Level and understandability using Patient Education Materials Assessment Tool (PEMAT) scores. Readability and understandability of the original discharge summaries were compared with the transformed, patient-friendly discharge summaries created through the LLM. As balancing metrics, accuracy and completeness of the patient-friendly version were measured. Results: Discharge summaries of 50 patients (31 female [62.0%] and 19 male [38.0%]) were included. The median patient age was 65.5 (IQR, 59.0-77.5) years. Mean (SD) Flesch-Kincaid Grade Level was significantly lower in the patient-friendly discharge summaries (6.2 [0.5] vs 11.0 [1.5]; P < .001). PEMAT understandability scores were significantly higher for patient-friendly discharge summaries (81% vs 13%; P < .001). Accuracy was rated as top box (completely correct) in 54 of 100 reviews, and safety concerns were reported in 18 of 100 reviews. Most inaccuracies were attributed to omission of key information (52.1%) compared with hallucination (8.7%). Conclusions and Relevance: In this cross-sectional study, an LLM successfully transformed discharge summaries into a more readable and understandable format. However, accuracy concerns highlight the need for human oversight and validation of LLM-generated patient-facing medical content before implementation.

  JAMA Network Open 3/4/2024 CC BY 4.0

• Ambient Artificial Intelligence Scribes to Alleviate the Burden of Clinical Documentation

  Clinical documentation in the electronic health record (EHR) has become increasingly burdensome for physicians and is a major driver of clinician burnout and dissatisfaction. Time dedicated to clerical activities and data entry during patient encounters also negatively affects the patient-physician relationship by hampering effective and empathetic communication and care. Ambient artificial intelligence (AI) scribes, which use machine learning applied to conversations to facilitate scribe-like capabilities in real time, has great potential to reduce documentation burden, enhance physician-patient encounters, and augment clinicians' capabilities. The technology leverages a smartphone microphone to transcribe encounters as they occur but does not retain audio recordings. To address the urgent and growing burden of data entry, in October 2023, The Permanente Medical Group (TPMG) enabled ambient AI technology for 10,000 physicians and staff to augment their clinical capabilities across diverse settings and specialties. The implementation process leveraged TPMG's extensive experience in large-scale technology instantiation and integration incorporating multiple training formats, at-the-elbow peer support, patient-facing materials, rapid-cycle upgrades with the technology vendor, and ongoing monitoring. In 10 weeks since implementation, the ambient AI tool has been used by 3,442 TPMG physicians to assist in as many as 303,266 patient encounters across a wide array of medical specialties and locations. In total, 968 physicians have enabled ambient AI scribes in ≥100 patient encounters, with one physician having enabled it to assist in 1,210 encounters. The response from physicians who have used the ambient AI scribe service has been favorable; they cite the technology's capability to facilitate more personal, meaningful, and effective patient interactions and to reduce the burden of after-hours clerical work. In addition, early assessments of patient feedback have been positive, with some describing improved interaction with their physicians. Early evaluation metrics, based on an existing tool that evaluates the quality of human-generated scribe notes, find that ambient AI use produces high-quality clinical documentation for physicians' editing. Further statistical analyses after AI scribe implementation also find that usage is linked with reduced time spent in documentation and in the EHR. Ongoing enhancements of the technology are needed and are focused on direct EHR integration, improved capabilities for incorporating medical interpretation, and enhanced workflow personalization options for individual users. Despite this technology's early promise, careful and ongoing attention must be paid to ensure that the technology supports clinicians while also optimizing ambient AI scribe output for accuracy, relevance, and alignment in the physician-patient relationship.

  NEJM Catalyst Innovations in Care Delivery 2/21/2024 Copyright © 2024 Massachusetts Medical Society

• Integrating Ambient Clinical Voice Technology – The Impacts, Challenges, and Benefits

  Ambient clinical voice technology automatically records and converts patient-physician conversations into clinical progress notes using generative AI. This technology aims to reduce documentation time and allow clinicians to focus more on patient interaction, potentially reducing clinician burnout by decreasing time spent on documentation. The technology may misinterpret words or provide inaccurate transcripts, requiring manual review and editing by clinicians to ensure accuracy. Privacy and confidentiality issues exist as patients may be uncomfortable with being recorded. There is a lack of industry standards in the market, making it difficult for providers to choose the right solution. The technology does not organize text into clinically relevant relationships or create computable output for quality measures, coding, etc. Integration with existing EHR systems is crucial for seamless workflow. Organizations need to consider factors like price, ease of use, note quality, revenue cycle support, and security when choosing a solution. Clinicians need to learn to use the technology in a clinically relevant way. Transparency with patients about the use of this technology is important. The technology could lead to regulatory-grade data, benefiting patients and companies that utilize real-world evidence. It may play a significant role in EHR technology, responding to voice commands and proposing context-aware next actions.

  Publication venue not specified 1/1/2024 Not specified

• A Method to Automate the Discharge Summary Hospital Course for Neurology Patients

  Objective: Generation of automated clinical notes has been posited as a strategy to mitigate physician burnout. In particular, an automated narrative summary of a patient's hospital stay could supplement the hospital course section of the discharge summary that inpatient physicians document in electronic health record (EHR) systems. In the current study, we developed and evaluated an automated method for summarizing the hospital course section using encoder-decoder sequence-to-sequence transformer models. Materials and Methods: We fine-tuned BERT and BART models and optimized for factuality through constraining beam search, which we trained and tested using EHR data from patients admitted to the neurology unit of an academic medical center. Results: The approach demonstrated good ROUGE scores with an R-2 of 13.76. In a blind evaluation, 2 board-certified physicians rated 62% of the automated summaries as meeting the standard of care, which suggests the method may be useful clinically. Discussion and Conclusion: To our knowledge, this study is among the first to demonstrate an automated method for generating a discharge summary hospital course that approaches a quality level of what a physician would write.

  Journal of the American Medical Informatics Association 11/17/2023 Oxford University Press Standard Journals Publication Model

• Needs and expectations for artificial intelligence in emergency medicine according to Canadian physicians

  Background: Artificial Intelligence (AI) is recognized by emergency physicians (EPs) as an important technology that will affect clinical practice. Several AI-tools have already been developed to aid care delivery in emergency medicine (EM). However, many EM tools appear to have been developed without a cross-disciplinary needs assessment, making it difficult to understand their broader importance to general-practice. Clinician surveys about AI tools have been conducted within other medical specialties to help guide future design. This study aims to understand the needs of Canadian EPs for the apt use of AI-based tools. Methods: A national cross-sectional, two-stage, mixed-method electronic survey of Canadian EPs was conducted from January-May 2022. The survey includes demographic and physician practice-pattern data, clinicians' current use and perceptions of AI, and individual rankings of which EM work-activities most benefit from AI. Results: The primary outcome is a ranked list of high-priority AI-tools for EM that physicians want translated into general use within the next 10 years. When ranking specific AI examples, 'automated charting/report generation', 'clinical prediction rules' and 'monitoring vitals with early-warning detection' were the top items. When ranking by physician work-activities, 'AI-tools for documentation', 'AI-tools for computer use' and 'AI-tools for triaging patients' were the top items. For secondary outcomes, EPs indicated AI was 'likely' (43.1%) or 'extremely likely' (43.7%) to be able to complete the task of 'documentation' and indicated either 'a-great-deal' (32.8%) or 'quite-a-bit' (39.7%) of potential for AI in EM. Further, EPs were either 'strongly' (48.5%) or 'somewhat' (39.8%) interested in AI for EM. Conclusions: Physician input on the design of AI is essential to ensure the uptake of this technology. Translation of AI-tools to facilitate documentation is considered a high-priority, and respondents had high confidence that AI could facilitate this task. This study will guide future directions regarding the use of AI for EM and help direct efforts to address prevailing technology-translation barriers such as access to high-quality application-specific data and developing reporting guidelines for specific AI-applications. With a prioritized list of high-need AI applications, decision-makers can develop focused strategies to address these larger obstacles.

  BMC Health Services Research 7/25/2023 CC BY 4.0

• Artificial Intelligence in Emergency Medicine: Viewpoint of Current Applications and Foreseeable Opportunities and Challenges

  Emergency medicine and its services have reached a breaking point during the COVID-19 pandemic. This pandemic has highlighted the failures of a system that needs to be reconsidered, and novel approaches need to be considered. Artificial intelligence (AI) has matured to the point where it is poised to fundamentally transform health care, and applications within the emergency field are particularly promising. In this viewpoint, we first attempt to depict the landscape of AI-based applications currently in use in the daily emergency field. We review the existing AI systems; their algorithms; and their derivation, validation, and impact studies. We also propose future directions and perspectives. Second, we examine the ethics and risk specificities of the use of AI in the emergency field. Emergency departments (EDs) and related services such as intensive care units and emergency medical dispatch (EMD) have recently been in the spotlight owing to the COVID-19 pandemic. The fragility of the emergency system has been exposed by overcrowded services, extensive waiting times, and exhausted staff struggling to respond to exceptional situations. AI techniques have already been shown to be promising for improving diagnosis, imaging interpretation, triage, and medical decision-making within an ED setting. This viewpoint reviews current AI applications in emergency medicine, discusses the transformer architecture and natural language processing advancements, examines AI applications for documentation and public health surveillance, and addresses ethical considerations for trustworthy AI deployment in emergency settings.

  Journal of Medical Internet Research 5/23/2023 CC BY 4.0

• Revolutionizing Healthcare: The Role of Artificial Intelligence in Clinical Practice

  Natural language processing (NLP) is a subfield of AI that focuses on the interaction between computers and humans through natural language, including understanding, interpreting, and generating human language. NLP involves various techniques such as text mining, sentiment analysis, speech recognition, and machine translation. AI-powered chatbots are being implemented in various healthcare contexts, such as diet recommendations, smoking cessation, and cognitive-behavioral therapy. AI has the potential to play a significant role in patient education by providing personalized and interactive information and guidance to patients and their caregivers. AI-powered mental health applications can assist in the early detection and diagnosis of mental health conditions, as well as provide tailored treatment and support. Clinical decision support systems (CDSS) are software applications that use AI algorithms to provide healthcare providers with real-time information and recommendations for patient care. AI enables quick and comprehensive retrieval of drug-related information from different resources through its ability to analyze the current medical literature, drug databases, and clinical guidelines to provide accurate and evidence-based decisions for healthcare providers. AI algorithms can analyze patient data to assist with triaging patients based on urgency; this helps prioritize high-risk cases, reducing waiting times and improving patient flow. Virtual health assistants are a new and innovative technology transforming the healthcare industry to support healthcare professionals. It is designed to simulate human conversation to offer personalized patient care based on input from the patient. AI-powered apps prove their benefits in patients with substance use disorder, with studies showing that using Woebot was significantly associated with improved substance use, cravings, depression, and anxiety. AI algorithms can analyze patient data to predict patient outcomes and recommend tailored treatment plans, including medication options. Data privacy, availability, and security are potential limitations to applying AI in clinical practice. Additionally, determining relevant clinical metrics and selecting an appropriate methodology is crucial to achieving the desired outcomes. Addressing these challenges and providing constructive solutions will require a multidisciplinary approach, innovative data annotation methods, and the development of more rigorous AI techniques and models.

  Publication venue not fully specified 1/1/2023 Not specified

• Current and Potential Applications of Ambient Artificial Intelligence

  Ambient clinical intelligence is a conversational AI application that integrates ambient voice sensing technology and virtual assistant function to automate and streamline visit documentation into the electronic health record (EHR) and data retrieval from the EHR by a physician during a patient's clinic encounter. The touchless voice-activated virtual assistant helps to decrease the administrative burden of physicians, enables better physician-patient interaction, and improves patient satisfaction and experience. This article examines various ambient AI applications across healthcare settings including Mayo Clinic's Ambient Warning and Response Evaluation system for ICU clinicians, which filters meaningful data from vast EHR data volumes, delivering real-time context-specific high-value clinical information to augment timely clinical decision support and prevent data overload. Additional applications discussed include use of ambient cameras for surgical skill evaluation through computer vision, AI-based systems using contactless ambient radio sensors for detecting Parkinson's disease and tracking disease progression through nocturnal breathing analysis, and ambient assisted living tools for improving quality of life for elderly and people with functional diversity including voice assistants and smart home solutions. The article also addresses challenges including privacy concerns regarding capture and sharing of large volumes of personal data without patient knowledge or informed consent, use beyond original intended purposes by third parties, and the critical need for explainability in ambient AI systems to clearly explain predictions and decision-making processes to end users. Ambient AI is positioned as a promising emerging technology in the very early phase of adoption with great potential to become mainstream over the next 10-15 years, leveraging ambient sensing technology and connected intelligence networks in the 5G era to transform healthcare delivery and enable intelligence-based medicine.

  Publication venue not fully specified 1/1/2023 Not specified

• Is artificial intelligence capable of generating hospital discharge summaries from inpatient records?

  Medical professionals have been burdened by clerical work, and artificial intelligence may efficiently support physicians by generating clinical summaries. However, whether hospital discharge summaries can be generated automatically from inpatient records stored in electronic health records remains unclear. Therefore, this study investigated the sources of information in discharge summaries. First, the discharge summaries were automatically split into fine-grained segments, such as those representing medical expressions, using a machine learning model from a previous study. Second, these segments in the discharge summaries that did not originate from inpatient records were filtered out. This was performed by calculating the n-gram overlap between inpatient records and discharge summaries. The final source origin decision was made manually. Finally, to reveal the specific sources (e.g., referral documents, prescriptions, and physician's memory) from which the segments originated, they were manually classified by consulting medical professionals. For further and deeper analysis, this study designed and annotated clinical role labels that represent the subjectivity of the expressions and builds a machine learning model to assign them automatically. The analysis results revealed the following: First, 39% of the information in the discharge summary originated from external sources other than inpatient records. Second, patient's past clinical records constituted 43%, and patient referral documents constituted 18% of the expressions derived from external sources. Third, 11% of the missing information was not derived from any documents. These are possibly derived from physicians' memories or reasoning. According to these results, end-to-end summarization using machine learning is considered infeasible. Machine summarization with an assisted post-editing process is the best fit for this problem domain.

  PLOS Digital Health 12/12/2022 CC BY 4.0

About this hub

We aggregate open-access research and provide plain-language and clinical summaries. We link to the original sources and clearly indicate licenses.