Agree to Agree? Pilot Deployment of Artificial Intelligence Pneumonia Detection in Seven Emergency Departments

Background/Objectives: Recent advances in artificial intelligence (AI) and machine learning (ML) enable targeted optimization of emergency department (ED) operations. We examine how reworking an ED's vertical processing pathway (VPP) using AI- and ML-driven recommendations affected patient throughput. Methods: We trained a non-linear ML model using triage data from 49,350 ED encounters to generate a personalized risk score that predicted whether an incoming patient is suitable for vertical processing. This model was integrated into a stochastic patient flow framework using queueing theory to derive an optimized VPP design. The resulting protocol prioritized a vertical assessment for patients with Emergency Severity Index (ESI) scores of 4 and 5, as well as 3 when the chief complaints involved skin, urinary, or eye issues. In periods of ED saturation, our data-driven protocol suggested that any waiting room patient should become VPP eligible. We implemented this protocol during a 13-week prospective trial and evaluated its effect on ED performance using before-and-after data. Results: Implementation of the optimized VPP protocol reduced the average ED length of stay (LOS) by 10.75 min (4.15%). Adjusted analyses controlling for potential confounders during the study period estimated a LOS reduction between 7.5 and 11.9 min (2.89% and 4.60%, respectively). No adverse effects were observed in the quality metrics, including 72 h ED revisit or hospitalization rates. Conclusions: A personalized, data-driven VPP protocol, enabled by ML predictions, significantly improved the ED throughput while preserving care quality. Unlike standard fast-track systems, this approach adapts to ED saturation and patient acuity. The methodology is customizable to patient populations and ED operational characteristics, supporting personalized patient flow optimization across diverse emergency care settings.

Purpose: Intracranial hemorrhage (ICH) has high morbidity and mortality, but early intervention has been shown to improve clinical outcomes. Several applications have emerged using artificial intelligence (AI) for automated detection of ICH, including RAPID ICH (RICH, iSchemaView, Menlo Park, CA). We present our initial clinical experience with RICH in a busy Level 1 trauma center. Materials/Methods: The study was supervised by the local IRB. Emergency department (ED) patients and inpatients (IP) receiving head CTs on the ED scanner at a level 1 trauma center were included. RICH output ("no ICH" or "suspected ICH") was recorded for each exam. Initial interpreting emergency radiologists or neuroradiologists had access to RICH output. Radiology reports reporting ICH were positive and those without ICH were negative. A board certified neuroradiologist reviewed each case with access to the initial report, RICH output, and prior exams. In cases with disagreement between readers, a third reader adjudicated the result and their decision was considered final. Expert reads were used as the gold standard and sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) were calculated. Results: 1388 patients were included, 1251 from the ED and 137 IP. For the ED patients, 139 had ICH and 1112 did not, where as in the IP cohort, 100 had ICH and 37 did not. RICH demonstrated overall sensitivity of 79% (73% ED, 88% IP), overall specificity of 95% (96% ED, 84% IP), overall PPV of 76% (65% ED, 94% IP) and overall NPV of 96% (97% ED, 72% IP). Conclusion: RICH had relatively high sensitivity and very high specificity for ICH, with lower sensitivity/higher specificity in the ED, and higher sensitivity/lower specificity in IP. Active worklist reprioritization allows faster triage of potentially positive studies and earlier intervention potentially improving clinical outcomes, especially in IPs with longer average turnaround times for routine studies.

Electronic Health Records: Promises and Realities: A 3-Part Series Part I: The Digital Sea Change, Ready or Not

Medication reconciliation (MR) in the emergency department (ED) is essential to ensure medication safety, especially for patients admitted to the hospital. However, performing MR for all ED patients, including those discharged, can be inefficient. To optimize prioritization, an artificial intelligence (AI)-powered hospital admission prediction dashboard was introduced. The primary aim of this study was to evaluate the effect of an artificial intelligence (AI) powered hospital admission prediction dashboard on the proportion of patients admitted to the hospital with an MR performed in the ED. The secondary aim was to assess its effect on the proportion of patients discharged from the ED with an MR. This retrospective before-after study was conducted at Hospital Group Twente and included ED visits between March 15 and December 31 in both 2023 and 2024. In the pre-intervention period, MR was strived for any patient, including patients discharged directly from the ED (i.e. potentially unnecessary MR as the risk for errors is low) and for admitted patients (i.e. correct MR). In 2024, the post-intervention period, a set of Extreme Gradient Boosting (XGBoost) models was trained on historical data (2015-2022) and integrated into a real-time dashboard to prioritize patients with the highest admission probability to perform MR. Primary outcome was the proportion of patients with correct MR. Secondary outcome was the proportion of patients with a potentially unnecessary MR. Chi-square test was used to compare proportions before and after implementation of the dashboard. The study included 25,505 ED visits. Pre-intervention 12,743 ED visits were included with 5,252 MRs performed. Post-intervention 12,762 ED visits were included with 4,882 MRs. After implementing the dashboard, the proportion of patients with correct MR increased from 86.4 to 89.0% (p = 0.0002), and the proportion of patients with potentially unnecessary MR decreased from 17.9 to 12.6% (p < 0.0001). The AI-powered hospital admission prediction dashboard improved the prioritization of MR in the ED. The proportion of patients with potentially unnecessary MR remains substantial and requires further improvement.

Objective: Clinical trials contribute to the development of clinical practice. However, little is known about the current status of trials on artificial intelligence (AI) conducted in emergency department and intensive care unit. The objective of the study was to provide a comprehensive analysis of registered trials in such field based on ClinicalTrials.gov.Methods: Registered trials on AI conducted in emergency department and intensive care unit were searched on ClinicalTrials.gov up to 12th January 2021. The characteristics were analyzed using SPSS21.0 software.Results: A total of 146 registered trials were identified, including 61 in emergency department and 85 in intensive care unit. They were registered from 2004 to 2021. Regarding locations, 58 were conducted in Europe, 58 in America, 9 in Asia, 4 in Australia, and 17 did not report locations. The enrollment of participants was from 0 to 18,000,000, with a median of 233. Universities were the primary sponsors, which accounted for 43.15%, followed by hospitals (35.62%), and industries/companies (9.59%). Regarding study designs, 85 trials were interventional trials, while 61 were observational trials. Of the 85 interventional trials, 15.29% were for diagnosis and 38.82% for treatment; of the 84 observational trials, 42 were prospective, 14 were retrospective, 2 were cross-sectional, 2 did not report clear information and 1 was unknown. Regarding the trials' results, 69 trials had been completed, while only 10 had available results on ClinicalTrials.gov.Conclusions: Our study suggest that more AI trials are needed in emergency department and intensive care unit and sponsors are encouraged to report the results.

Emergency departments (ED) are becoming increasingly overwhelmed, increasing poor outcomes. Triage scores aim to optimize the waiting time and prioritize the resource usage. Artificial intelligence (AI) algorithms offer advantages for creating predictive clinical applications. Evaluate a state-of-the-art machine learning model for predicting mortality at the triage level and, by validating this automatic tool, improve the categorization of patients in the ED. An institutional review board (IRB) approval was granted for this retrospective study. Information of consecutive adult patients (ages 18-100) admitted at the emergency department (ED) of one hospital were retrieved (January 1, 2012-December 31, 2018). Features included the following: demographics, admission date, arrival mode, referral code, chief complaint, previous ED visits, previous hospitalizations, comorbidities, home medications, vital signs, and Emergency Severity Index (ESI). The following outcomes were evaluated: early mortality (up to 2days post ED registration) and short-term mortality (2-30days post ED registration). A gradient boosting model was trained on data from years 2012-2017 and examined on data from the final year (2018). The area under the curve (AUC) for mortality prediction was used as an outcome metric. Single-variable analysis was conducted to develop a nine-point triage score for early mortality. Overall, 799,522 ED visits were available for analysis. The early and short-term mortality rates were 0.6% and 2.5%, respectively. Models trained on the full set of features yielded an AUC of 0.962 for early mortality and 0.923 for short-term mortality. A model that utilized the nine features with the highest single-variable AUC scores (age, arrival mode, chief complaint, five primary vital signs, and ESI) yielded an AUC of 0.962 for early mortality. The gradient boosting model shows high predictive ability for screening patients at risk of early mortality utilizing data available at the time of triage in the ED.

The integration of artificial intelligence (AI) and machine learning (ML) in healthcare has become a major point of interest and raises the question of its impact on the emergency department (ED) triaging process. AI’s capacity to emulate human cognitive processes coupled with advancements in computing has shown positive outcomes in various aspects of healthcare but little is known about the use of AI in triaging patients in ED. AI algorithms may allow for earlier diagnosis and intervention; however, overconfident answers may present dangers to patients. The purpose of this review was to explore comprehensively recently published literature regarding the effect of AI and ML in ED triage and identify research gaps. A systemized search was conducted in September 2023 using the electronic databases EMBASE, Ovid MEDLINE, and Web of Science. To meet inclusion criteria, articles had to be peer-reviewed, written in English, and based on primary data research studies published in US journals 2013-2023. Other criteria included 1) studies with patients needing to be admitted to hospital EDs, 2) AI must have been used when triaging a patient, and 3) patient outcomes must be represented. The search was conducted using controlled descriptors from the Medical Subject Headings (MeSH) that included the terms “artificial intelligence” OR “machine learning” AND “emergency ward” OR “emergency care” OR “emergency department” OR “emergency room” AND “patient triage” OR “triage” OR “triaging.” The search initially identified 1,142 citations. After a rigorous, systemized screening process and critical appraisal of the evidence, 29 studies were selected for the final review. The findings indicated that 1) ML models consistently demonstrated superior discrimination abilities compared to conventional triage systems, 2) the integration of AI into the triage process yielded significant enhancements in predictive accuracy, disease identification, and risk assessment, 3) ML accurately determined the necessity of hospitalization for patients requiring urgent attention, and 4) ML improved resource allocation and quality of patient care, including predicting length of stay. The suggested superiority of ML models in prioritizing patients in the ED holds the potential to redefine triage precision.

Time Is Brain

Background/Objectives: Rapid and accurate electrocardiogram (ECG) interpretation is essential for timely recognition of ST-elevation myocardial infarction (STEMI) and initiation of reperfusion therapy in the emergency department (ED). We evaluated the diagnostic performance of a real-time artificial intelligence (AI) ECG interpretation system and its pragmatic impact when integrated into routine ED workflows. Methods: This prospective, single-center pragmatic observational study was conducted in a regional emergency medical center ED in Busan, Republic of Korea (1 January-31 December 2024). Consecutive adults (≥18 years) undergoing 12-lead ECG for cardiovascular-related symptoms were enrolled (N = 1524). A predefined alternating-day protocol allocated visits to physician-only interpretation days (physician-days, N = 763) or AI output disclosure days (AI-days, N = 761). Diagnostic performance for STEMI was assessed using paired ECG-level comparisons between physician-alone interpretation and AI output against a blinded expert-panel reference standard; clinical impact outcomes included reperfusion-related time metrics, hospital length of stay (LOS), and in-hospital mortality. Results: Against the expert reference standard, AI showed higher STEMI sensitivity than physician-alone interpretation (96.7% vs. 68.3%; McNemar p = 0.027), while specificity was lower (75.9% vs. 84.5%; p = 0.018). In pragmatic day-level comparisons, door-to-balloon time was shorter on AI-days (40.0 ± 19.81 vs. 47.34 ± 21.90 min; p = 0.001), and time to PCI was significantly reduced among patients with atypical presentations (42.3 ± 18.21 vs. 57.1 ± 20.11 min; p = 0.013). Among admitted patients, hospital LOS was shorter on AI-days (13 ± 9.21 vs. 17 ± 10.31 days; p = 0.010), whereas in-hospital mortality did not differ significantly between groups (17.0% vs. 16.77%; p = 0.191). Conclusions: Real-time AI-ECG integration in the ED was associated with improved STEMI detection sensitivity and shorter reperfusion-related time metrics, particularly in atypical presentations, and with reduced hospital LOS among admitted patients. Short-term mortality was comparable between groups. Further multicenter studies are warranted to confirm generalizability and to balance benefits against potential false-positive-related operational impacts.

Artificial intelligence is gaining increasing relevance in the field of radiology. This study retrospectively evaluates how a commercially available deep learning algorithm can detect pneumonia in chest radiographs (CR) in emergency departments. The chest radiographs of 948 patients with dyspnea between 3 February and 8 May 2020, as well as 15 October and 15 December 2020, were used. A deep learning algorithm was used to identify opacifications associated with pneumonia, and the performance was evaluated by using ROC analysis, sensitivity, specificity, PPV and NPV. Two radiologists assessed all enrolled images for pulmonal infection patterns as the reference standard. If consolidations or opacifications were present, the radiologists classified the pulmonal findings regarding a possible COVID-19 infection because of the ongoing pandemic. The AUROC value of the deep learning algorithm reached 0.923 when detecting pneumonia in chest radiographs with a sensitivity of 95.4%, specificity of 66.0%, PPV of 80.2% and NPV of 90.8%. The detection of COVID-19 pneumonia in CR by radiologists was achieved with a sensitivity of 50.6% and a specificity of 73%. The deep learning algorithm proved to be an excellent tool for detecting pneumonia in chest radiographs. Thus, the assessment of suspicious chest radiographs can be purposefully supported, shortening the turnaround time for reporting relevant findings and aiding early triage.

Emergency department (ED) overcrowding occurs when healthcare demand exceeds available resources, significantly impairing the quality of patient care and the efficiency of the healthcare system. This review comprehensively analyzes the definition, causes, measurement methods, consequences, and proposed solutions of ED overcrowding. ED crowding is marked by prolonged waiting times, disrupted patient flow, interrupted medical service delivery, and risks to patient and staff safety. Patient-related factors include the preference for EDs for non-emergent conditions, complicated health issues in the elderly, low health literacy, and frequent visits by individuals with chronic illnesses. Systemic factors involve access block, staff shortages, inefficiencies in discharge processes, and delays in diagnostic tests and consultations. External factors include inadequate primary care services, socioeconomic disparities, seasonal fluctuations, epidemics, and natural disasters. To evaluate ED crowding, measurement tools such as the National Emergency Department Overcrowding Scale and the Emergency Department Work Index are utilized; however, the lack of standardization and variability across different hospital types remains a point of contention. Overcrowding compromises patient safety, leading to treatment delays, increased medical errors, staff burnout, and escalating healthcare costs. Proposed solutions include short-term measures (fast-track units, early discharge planning, patient redirection), long-term strategies (enhancing primary care, establishment of acute care centers, management of inpatient bed capacity, and planning to reduce the burden of chronic illnesses), and technological approaches (artificial intelligence, simulation models, telemedicine, and patient tracking technologies). Gaps in the literature include issues related to the validity and reliability of measurement tools, limited data on the economic and clinical impacts of interventions, insufficient research in low- and middle-income countries, and inadequate examination of patient behaviors and the psychological effects on healthcare personnel. The effective management of ED crowding requires the coordinated implementation of systemic reforms, innovative technologies, and collaborative efforts across clinical, academic, and policy-making domains. This is a critical step toward enhancing patient safety, improving clinical outcomes, and ensuring the sustainability of the healthcare system.

Artificial intelligence versus orthopedic surgeons as an orthopedic consultant in the emergency department.

Background: Pneumonia remains a major contributor to emergency department morbidity, and timely administration of antibiotics is a key determinant of outcomes. Conventional radiograph workflows often delay interpretation during high-volume periods, prolonging clinical decision-making. Recent advances in artificial intelligence have enabled automated prioritization of imaging studies, yet evidence from randomized trials assessing its real-time clinical impact remains limited. Objective: To evaluate whether AI-assisted triage of chest radiographs in the emergency department reduces time-to-antibiotics and improves early clinical outcomes in patients with suspected pneumonia compared with standard workflow. Methods: A randomized controlled trial was conducted over a six-month period, enrolling adult patients presenting with symptoms suggestive of pneumonia who received chest radiographs as part of routine evaluation. Participants were randomized in a 1:1 ratio to AI-assisted triage or standard radiograph workflow. The AI system automatically flagged radiographs with suspected infiltrates, prioritizing them for expedited radiologist review. Data collected included baseline demographics, time-to-antibiotics, radiologist report turnaround time, length of stay in the emergency department, and early clinical response at 48 hours using a standardized ordinal scale. Statistical analyses were performed using independent t-tests and chi-square tests, with significance set at p &lt; 0.05. Results: A total of 140 participants were analyzed. The AI-assisted group demonstrated a significantly shorter time-to-antibiotics compared with the standard workflow group, along with faster radiology reporting times and modest reductions in emergency department length of stay. The proportion of patients demonstrating early clinical improvement at 48 hours was also higher in the AI-assisted arm. No adverse effects related to AI implementation were observed. Conclusion: AI-assisted triage of chest radiographs meaningfully improved critical process measures and early clinical outcomes in suspected pneumonia, supporting its integration into acute-care imaging workflows.

BackgroundAs the COVID-19 epidemic increases in severity, the burden of quarantine stations outside emergency departments (EDs) at hospitals is increasing daily. To address the high screening workload at quarantine stations, all staff members with medical licenses are required to work shifts in these stations. Therefore, it is necessary to simplify the workflow and decision-making process for physicians and surgeons from all subspecialties.ObjectiveThe aim of this paper is to demonstrate how the National Cheng Kung University Hospital artificial intelligence (AI) trilogy of diversion to a smart quarantine station, AI-assisted image interpretation, and a built-in clinical decision-making algorithm improves medical care and reduces quarantine processing times.MethodsThis observational study on the emerging COVID-19 pandemic included 643 patients. An “AI trilogy” of diversion to a smart quarantine station, AI-assisted image interpretation, and a built-in clinical decision-making algorithm on a tablet computer was applied to shorten the quarantine survey process and reduce processing time during the COVID-19 pandemic.ResultsThe use of the AI trilogy facilitated the processing of suspected cases of COVID-19 with or without symptoms; also, travel, occupation, contact, and clustering histories were obtained with the tablet computer device. A separate AI-mode function that could quickly recognize pulmonary infiltrates on chest x-rays was merged into the smart clinical assisting system (SCAS), and this model was subsequently trained with COVID-19 pneumonia cases from the GitHub open source data set. The detection rates for posteroanterior and anteroposterior chest x-rays were 55/59 (93%) and 5/11 (45%), respectively. The SCAS algorithm was continuously adjusted based on updates to the Taiwan Centers for Disease Control public safety guidelines for faster clinical decision making. Our ex vivo study demonstrated the efficiency of disinfecting the tablet computer surface by wiping it twice with 75% alcohol sanitizer. To further analyze the impact of the AI application in the quarantine station, we subdivided the station group into groups with or without AI. Compared with the conventional ED (n=281), the survey time at the quarantine station (n=1520) was significantly shortened; the median survey time at the ED was 153 minutes (95% CI 108.5-205.0), vs 35 minutes at the quarantine station (95% CI 24-56; P<.001). Furthermore, the use of the AI application in the quarantine station reduced the survey time in the quarantine station; the median survey time without AI was 101 minutes (95% CI 40-153), vs 34 minutes (95% CI 24-53) with AI in the quarantine station (P<.001).ConclusionsThe AI trilogy improved our medical care workflow by shortening the quarantine survey process and reducing the processing time, which is especially important during an emerging infectious disease epidemic.

This systematic literature review investigates potentialities to exploit AI-supported technologies in Chinese hospital emergency departments. Modern digital medical tools combined with artificial intelligence technologies enable rapid acquisition, gathering and analysis of extensive medical data, providing timely insights for emergency physicians to make timely treatment decisions that can improve patient outcomes. Correction of motion artefacts can be achieved in medical imaging using artificial intelligence technologies without the need for rescanning, while at the same time preserving patient safety by minimizing the amount of ionizing radiation. The possibilities are to produce a high-quality medical image being corrected from motion artefacts by artificial intelligence models and methods to increase the diagnostic accuracy of medical images. Medical radiation imaging and computer tomography imaging are currently the two categories of imaging modalities responsible for ionizing radiation dose, involved in at least 80% of medical ionizing radiation dose for an average patient. Acquisition of enough CDS is difficult in fast spinning imaging sequence, e.g. 0.5 seconds for Gastrografin X-ray CT imaging scan sequence, a patient motion artefact greatly reduces its image quality and makes its follow-up diagnosis difficult or even impossible. Critical care services will be significantly improved, more accessible, and more cost-effective by using artificial intelligence (AI) (Nguyen &amp; Nguyen, 2020; Patel &amp; Cushing, 2021; Shaik et al., 2023). Remote monitoring could be on-time and in-time monitoring and treatment without the need for admission, dramatically decreasing emergency department (ED) admission times, especially the bed-constrained situations can realize a patient immediately sent to still available bed. Automated triage of non-urgent patients by the AI-based artificial intelligent emergency department platform by symptom extraction and monitoring. Machine-learning models including the support vector machine classifier, random forest classifier, adaptive boost classifier, and decision tree classifier generate models with good performance. A multi-agent system constructed with traditional machine-learning algorithms performs better than a single traditional machine-learning model. Very good scalability in the computation execution framework is created by using the ensemble learning from local agents to overall model levels.