Artificial Intelligence in Emergency Trauma Care: A Preliminary Scoping Review

Figures & data

Table 1 Selected Characteristics of Studies Identified for Inclusion

Reference	Author(s)	Title	Design	Country / Region	Summary of Key Findings	Domain
[9]	Cheng, Lin, Hsu, Chen, Huang, Hsieh, Fu, Chung, Liao	Deep Learning for Automated Detection and Localization of Traumatic Abdominal Solid Organ Injuries on CT Scans.	Retrospective Cohort	Taiwan	A DLM was specifically developed to detect solid organ injuries in patients who have undergone CT scans for blunt abdominal trauma, aiming to assist in the rapid identification of life-threatening injuries. The model was trained on 1302 CT scans (87% of the total) and tested on 194 scans (13%), taken from patients at a single trauma center between 2008 and 2017. The DLM showed high accuracy and specificity in detecting injuries to the spleen, liver, and kidneys, indicating its potential as a tool to support clinicians in making quicker decisions regarding trauma care.	Diagnostics
[10]	Michel, Manns, Boudersa, Jaubert, Dupic, Vivien, Burgun, Campeotto, Tsopra	Clinical decision support system in emergency telephone triage: A scoping review of technical design, implementation and evaluation.	Scoping Review	US	Identified 19 CDSS for emergency telephone triage, highlighting a mix of knowledge-based[9] and data-driven[7] systems, primarily aimed at assisting nurses or non-medical staff with patient orientation and severity assessment. Eleven CDSS were implemented in real-world settings, but only three were integrated with Electronic Health Records (EHR), indicating a gap in leveraging existing health data. The review underscores the necessity for a hybrid, user-tailored CDSS that can interface with oral, video, and digital data, emphasizing iterative evaluation of CDSS’s intrinsic characteristics and their clinical impact throughout the IT lifecycle.	Triage
[11]	Russe, Rebmann, Tran, Kellner, Reisert, Bamberg, Kotter, Kim	AI-based X-ray fracture analysis of the distal radius: accuracy between representative classification, detection and segmentation deep learning models for clinical practice.	Comparative Observational	Germany	Evaluated AI models for detecting distal radius fractures in radiographs, comparing custom-trained classification, detection, segmentation models, and a commercial solution, showing high accuracies (up to 0.97) in fracture detection. A total of 2856 radiographs were analyzed, with AI models achieving high-performance metrics, including Cohen’s and Fleiss’ kappa values indicating strong agreement in fracture detection across models. The findings suggest that the choice of an AI tool for fracture analysis depends on the specific requirements for automation, ranging from automated classification to AI-assisted reading or minimizing false negatives.	Diagnostics
[12]	Piliuk, Tomforde	Artificial intelligence in emergency medicine. A systematic literature review.	Systematic Review	Germany	This study systematically categorizes and examines existing contributions in AI applications for emergency medicine, identifying obstacles stemming from healthcare regulations and the fragmented nature of published research. Findings reveal a predominance of specialized studies in diagnostics and triage, often utilizing disparate data sources and methodologies, highlighting the necessity for standardized approaches and end-to-end solutions integrating human-machine interaction for improved generalization.	Diagnostics
[13]	Choi, Vendrow, Moor	Development and Validation of a Model to Quantify Injury Severity in Real Time.	Retrospective Cohort	US	Developed and validated the Length of Stay, Disposition, Mortality (LDM) Injury Index, a deep learning model that quantifies injury severity in real-time, using three outcomes: predicted hospital length of stay, probability of discharge to a facility, and probability of inpatient mortality. The model demonstrated comparable or better performance metrics (such as AUROC, recall, and specificity) in external validation compared to the traditional Injury Severity Score (ISS), particularly in predicting facility discharge and mortality, albeit with some overestimation in mortality predictions. The model, which uses 176 potential injuries for its predictions, showed excellent calibration for predicting facility discharge but requires further study to evaluate its effectiveness at scale.	Diagnostics
[14]	Gao, Soh, Liu, Lim, Ting, Cheng, Wong, Liew, Oh, Tan, Venkataraman, Goh, Yan	Application of a deep learning algorithm in the detection of hip fractures.	Observational Diagnostic Accuracy/Retrospective Cohort	Singapore	A deep convolutional neural network (DCNN) demonstrated high accuracy (91%) and sensitivity (98%) in detecting hip fractures on plain frontal pelvic radiographs (PXRs), with a low false-negative rate (2%) and an area under the receiver operating characteristic curve (AUC) of 0.98. The visualization algorithm, gradient-weighted class activation mapping (Grad-CAM), achieved a 95.9% accuracy for identifying fracture lesions, confirming the validity of the model. The study concludes that DCNNs offer an efficient and economical means for hip fracture detection and localization on PXRs, potentially aiding primary physicians in emergent screening and evaluation efforts without disrupting current clinical pathways.	Diagnostics
[15]	Sax, Warton, Sofrygin, Mark, Ballard, Kene, Vinson, Reed	Automated analysis of unstructured clinical assessments improves emergency department triage performance: A retrospective deep learning analysis.	Retrospective Cohort	US	Triage models utilizing both triage variables and clinical assessments significantly outperformed models based solely on triage variables, with AUC values of 0.87 (95% CI 0.87–0.87) for both hospitalization and fast-track eligibility prediction. Hospitalization rate among ED patients was 12.7% (n = 673,659), while 37.0% (n = 1,966,615) were deemed fast-track eligible based on discharge criteria. Models relying solely on triage variables showed lower predictive accuracy, with AUC values of 0.77 (95% CI 0.77–0.78) for hospitalization and 0.70 (95% CI 0.70–0.71) for fast-track eligibility.	Triage
[16]	Ouyang, Chen, Tee, Lin, Kuo, Liao, Cheng, Liao	The Application of Design Thinking in Developing a Deep Learning Algorithm for Hip Fracture Detection.	Intervention Development	Taiwan	Design thinking applied to DL algorithm development improved diagnostic accuracy, sensitivity, and specificity for detecting hip fractures from pelvic plain films in trauma care. The study identified a specific clinical need regarding femoral fracture diagnosis, leading to the development of DL models with enhanced performance. Integration of design thinking enhanced DL algorithm performance, ensuring user-centered solutions for trauma care in healthcare settings.	Diagnostics
[17]	Hosseini, Hosseini, Qayumi, Ahmady, Koohestani	The Aspects of Running Artificial Intelligence in Emergency Care; a Scoping Review.	Scoping Review	Iran	The study reveals a growing trend of AI applications in emergency medicine, covering various areas such as machine learning algorithms, prehospital emergency management, triage, patient disposition, disease prediction, and emergency department management. Despite the potential benefits demonstrated by AI in improving patient outcomes through predictive modeling, ethical concerns regarding AI-based decision-making transparency are highlighted. The scoping review underscores the need for an ethical framework to address the lack of transparency in AI decision-making processes within emergency medicine contexts.	Diagnostics
[18]	He, Dash, Duanmu, Tan, Ouyang, Zou	AI-ENABLED ASSESSMENT OF CARDIAC FUNCTION AND VIDEO QUALITY IN EMERGENCY DEPARTMENT POINT-OF-CARE ECHOCARDIOGRAMS.	Prospective Diagnostic	US	EchoNet-POCUS, a deep learning system, was developed to assist emergency physicians (EPs) in interpreting point-of-care ultrasound (POCUS) echocardiograms and reduce operator-to-operator variability. The system achieved high accuracy in predicting abnormal cardiac function with an area under the receiver operating characteristic curve (AUROC) of 0.92 and moderate accuracy in predicting video quality with an AUROC of 0.81. EchoNet-POCUS demonstrated feasibility for real-time application on bedside echocardiogram videos using standard hardware, as evidenced by a prospective pilot study.	Diagnostics
[19]	Abrigo, Ko, Chen, Lai, Cheung, Chu, Yu	Artificial intelligence for detection of intracranial haemorrhage on head computed tomography scans: diagnostic accuracy in Hong Kong.	Prospective Diagnostic	Hong Kong	An AI algorithm developed using a large, open-access dataset of CT slices demonstrated promising utility in identifying acute intracranial hemorrhage (ICH) on non-contrast head CT scans. The model achieved an area under the curve (AUC) of 0.842 for scan-based detection of ICH, with a pre-specified probability threshold of ≥50% yielding 78.6% accuracy, 73% sensitivity, 79% specificity, 18.6% positive predictive value, and 97.8% negative predictive value. Manual review of CT slices nominated by the model could reduce false-negative scans, indicating the potential for further refinement to enhance the model’s localization capabilities for improved clinical application.	Diagnostics
[20]	Sundrani, Chen, Jin, Abad, Rajpurkar, Kim	Predicting patient decompensation from continuous physiologic monitoring in the emergency department.	Observational Cohort	US	A multimodal machine learning approach combining standard triage data with features from continuous physiologic monitoring accurately predicts the onset of new vital sign abnormalities in ED patients with initially normal vital signs. The best-performing models, utilizing both engineered and transformer-derived features, achieved high predictive performance, with AUROC values of 0.836 for new tachycardia, 0.802 for new hypotension, and 0.713 for new hypoxia in a 90-minute window. Salient features contributing to prediction include vital sign trends, PPG perfusion index, and ECG waveforms, highlighting the potential for continuous application of this approach to improve triage and predict clinical deterioration in apparently stable patients.	Treatment
[21]	Takaki, Inoue, Maki, Furuya, Mikami, Mizutani, Takada, Okimatsu, Yunde, Miura, Shiratani, Nagashima, Maruyama, Shiga, Inage, Orita, Eguchi, Ohtori	Automated fracture screening using an object detection algorithm on whole-body trauma computed tomography.	Retrospective Observational	Japan	Convolutional neural network (CNN) deep learning methods were investigated for automatic localization and classification of pelvic, rib, and spine fractures in trauma care. The CNN model demonstrated promising performance with sensitivity, precision, and F1-score values of 0.786, 0.648, and 0.711, respectively, for grouped mean values of fractures. Surgeons showed improved sensitivity in fracture detection and reduced reading and interpretation time for CT scans, particularly benefiting less experienced orthopedic surgeons, suggesting potential for improved patient care and workflow efficiency in polytrauma cases.	Diagnostics
[22]	Rashid, Zia, Rehman, Meraj, Rauf, Kadry	A Minority Class Balanced Approach Using the DCNN-LSTM Method to Detect Human Wrist Fracture.	Method Development	Pakistan	A fused model of deep learning, combining convolutional neural network (CNN) and long short-term memory (LSTM), was proposed to detect wrist fractures from X-ray images, providing an automated diagnosing tool as a second option for doctors. The dataset comprised 192 wrist X-ray images, and the proposed model utilized image pre-processing and data augmentation techniques to address class imbalance and enhance feature extraction. Comparative analysis with other deep learning models demonstrated that the DCNN-LSTM fusion achieved higher accuracy and exhibited potential for medical applications as a reliable second option in wrist fracture diagnosis, emphasizing its utility in reducing missed diagnoses.	Diagnostics
[23]	Zech, Santomartino, Yi	Artificial Intelligence (AI) for Fracture Diagnosis: An Overview of Current Products and Considerations for Clinical Adoption, From the AJR Special Series on AI Applications.	Review	US	Artificial intelligence (AI) and deep learning have demonstrated strong potential in accurately detecting fractures, enhancing radiologists’ performance in research settings. Despite the increasing availability of AI products for clinical use, guidance for radiologists on adopting this technology is limited. This review outlines how AI and deep learning algorithms can assist radiologists in diagnosing fractures and provides an overview of commercially available FDA-cleared AI tools for fracture detection, along with considerations for their clinical adoption.	Diagnostics
[24]	Wei, Li, Sing, Yang, Beeram, Puvanesarajah, Valle, Tornetta, Fritz, Yi	Detecting total hip arthroplasty dislocations using deep learning: clinical and Internet validation.	Observational Diagnostic Accuracy	US	Convolutional neural networks (CNNs) were trained and evaluated for the automated detection of periprosthetic dislocations of total hip arthroplasty (THA) on radiographs, aiming to expedite diagnosis and treatment. Multiple CNNs achieved excellent diagnostic performance, with area under the receiver operating characteristic curve (AUROC) values of 1 for both internal and external test sets, indicating high generalizability. Class activation mapping (CAM) revealed consistent emphasis on the THA region by CNNs for both dislocated and non-dislocated cases, supporting their potential use for efficient triage in the emergency department.	Diagnostics
[25]	Yao, Leung, Tsai, Huang, Fu	Novel Deep Learning-Based System for Triage in the Emergency Department Using Electronic Medical Records: Retrospective Cohort Study.	Retrospective Cohort	Taiwan	A deep learning-based triage system was developed using electronic medical records from emergency department (ED) patients to predict clinical outcomes after ED treatments. The system, utilizing convolutional neural networks combined with recurrent neural networks and attention mechanisms, achieved high accuracy and area under the receiver operating characteristic curve (AUROC) values for predicting hospitalization, mortality, and admission to the intensive care unit. Results from both the National Hospital Ambulatory Medical Care Survey and an external dataset from the National Taiwan University Hospital demonstrated superior performance compared to traditional methods, indicating its potential for implementation in real-world clinical settings to improve patient triage and resource allocation in busy EDs.	Triage
[26]	Sanchez-Salmerón, Gómez-Urquiza, Albendín-García, Correa-Rodríguez, Martos-Cabrera, Velando-Soriano, Suleiman-Martos	Machine learning methods applied to triage in emergency services: A systematic review.	Systematic Review	Spain	Machine learning (ML) methods consistently outperformed traditional triage scales/scores, such as the Emergency Severity Index, in predicting important outcomes like mortality, critical care outcomes, admission, and the need for hospitalization in the emergency department. Among the ML models considered, XGBoost and Deep Neural Networks demonstrated the highest levels of prediction accuracy, while Logistic Regression performed comparatively worse. The findings suggest that ML systems have the potential to significantly improve the triage process in emergency departments by accurately predicting important variables, thus aiding in more effective patient management and resource allocation.	Triage
[27]	Dipnall, Page, Du, Costa, Lyons, Cameron, Steiger, Hau, Bucknill, Oppy, Edwards, Varma, Jung, Gabbe	Predicting fracture outcomes from clinical registry data using artificial intelligence supplemented models for evidence-informed treatment (PRAISE) study protocol.	Prospective Observational	Australia	The ”PRAISE” study aims to utilize artificial intelligence (AI) methods on unstructured data to describe fracture characteristics and assess if this information improves the identification of key fracture characteristics and prediction of patient-reported outcome measures and clinical outcomes following wrist fractures. The study will include adult patients presenting with wrist fractures in four Victorian hospitals, using routine registry data from the Victorian Orthopaedic Trauma Outcomes Registry (VOTOR) and electronic medical record (EMR) information. A multimodal deep learning fracture reasoning system (DLFRS) will be developed to reason on EMR information, and machine learning prediction models will test performance with or without output from the DLFRS, aiming to provide better prediction of clinical and patient-reported outcomes following distal radius fractures.	Diagnostics
[28]	Kim, Jung, Park, Park, Yi, Yang, Kim, Cho, Ha	Application of convolutional neural networks for distal radio-ulnar fracture detection on plain radiographs in the emergency room.	Observational Diagnostic Accuracy	Korea	Two convolutional neural network models, DenseNet-161 and ResNet-152, were evaluated for wrist fracture detection using image data collected from patients presenting with wrist trauma at the emergency department. Performance evaluation on a test dataset showed that both models achieved high sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for wrist fracture detection. The area under the receiver operating characteristic curves (AUC) indicated excellent performance of DenseNet-161 and ResNet-152 in detecting wrist fractures, suggesting their potential utility as diagnostic aids in the emergency room setting.	Diagnostics
[29]	Joseph, Leventhal, Grossestreuer, Wong, Joseph, Nathanson, Donnino, Elhadad, Sanchez	Deep-learning approaches to identify critically Ill patients at emergency department triage using limited information.	Cross- Sectional	US	Deep-learning models using only immediately available triage data demonstrated promising accuracy in identifying critically ill patients, surpassing the performance of traditional methods like vital sign thresholds and the emergency severity index (ESI). Successively complex deep-learning models, including logistic regression, neural networks with structured data, gradient tree boosting, and neural network models with textual data, showed statistically significant improvements in accuracy, as measured by the area under the receiver-operator curve (AUC). These findings suggest that deep-learning approaches could enhance triage processes by accurately identifying critically ill patients using readily available information, potentially leading to improved clinical and operational outcomes in the emergency department.	Triage
[30]	Miles, Turner, Jacques, Williams, Mason	Using machine-learning risk prediction models to triage the acuity of undifferentiated patients entering the emergency care system: a systematic review.	Systematic Review	United Kingdom	The review assessed the accuracy of machine learning methods in triaging patients presenting in the Emergency Care System, with a focus on calibration, discrimination, and classification statistics. A total of 92 models from 25 studies were included, with two main triage outcomes: hospitalization (56 models) and critical care need (25 models). Machine-learning methods, including neural networks, tree-based methods, and logistic regression, demonstrated accuracy in triaging patients, with neural networks showing the highest median C-statistic for critical care need. However, logistic regression models were noted for their transparency in reporting model performance.	Triage
[31]	Ozkaya, Topal, Bulut, Gursoy, Ozuysal, Karakaya	Evaluation of an artificial intelligence system for diagnosing scaphoid fracture on direct radiography.	Observational Diagnostic Accuracy/Retrospective Cohort	Turkey	The study aimed to evaluate the diagnostic performance of a convolutional neural network (CNN) in detecting scaphoid fractures on anteroposterior wrist radiographs. Results showed that the CNN had a sensitivity of 76% and specificity of 92%, with an AUC of 0.840 and an F-score of 0.826 in identifying scaphoid fractures. The performance of the CNN was comparable to a less experienced orthopaedic specialist but better than that of the emergency department physician. However, the experienced orthopaedic specialist demonstrated the best diagnostic performance according to AUC.	Diagnostics
[32]	Weikert, Noordtzij, Bremerich, Stieltjes, Parmar, Cyriac, Sommer, Sauter	Assessment of a Deep Learning Algorithm for the Detection of Rib Fractures on Whole-Body Trauma Computed Tomography.	Observational Diagnostic Accuracy	Switzerland	A deep learning-based algorithm showed good diagnostic performance in detecting both acute and chronic rib fractures on whole-body trauma CT scans, achieving a sensitivity of 87.4% and specificity of 91.5%. The algorithm detected 587 true-positive findings with a sensitivity of 65.7% on a per-finding level, while also identifying 97 true rib fractures not mentioned in the written CT reports. Correct detection was particularly associated with displacement, indicating the algorithm’s potential as a screening tool to prevent false-negative radiology reports in clinical settings.	Diagnostics
[33]	Jalal, Parker, Ferguson, Nicolaou	Exploring the Role of Artificial Intelligence in an Emergency and Trauma Radiology Department.	Review	Canada	The increasing demand for emergency radiology services, coupled with the need for accurate and timely reporting, has led to radiologists being overburdened with high imaging volumes and workload. Artificial intelligence (AI) holds promise in assisting emergency and trauma radiologists by potentially alleviating their workload through automated image analysis and interpretation. This article aims to provide an evidence-based discussion on the evolving role of AI in emergency and trauma radiology departments, addressing technical processes, challenges in algorithm training and validation, ethical considerations, and the pivotal role of emergency radiologists in implementing AI-guided systems to improve patient care and reduce radiologist burnout.	Diagnostics
[34]	Hwang, Nam, Lim, Park, Jeong, Kang, Hong, Kim, Goo, Park, Kim, Park	Deep Learning for Chest Radiograph Diagnosis in the Emergency Department.	Retrospective Diagnostic Accuracy	Korea	The deep learning (DL) algorithm exhibited a high diagnostic performance in identifying abnormal chest radiographs in the emergency department (ED), with an area under the receiver operating characteristic curve (AUC) of 0.95, a sensitivity of 88.7% at a high-sensitivity cutoff, and a specificity of 69.6% at the same cutoff. When compared to on-call radiology residents, the DL algorithm demonstrated higher sensitivity (88.7% vs 65.6%, p < 0.001) and lower specificity (69.6% vs 98.1%, p < 0.001) at the high-sensitivity cutoff. Reinterpretation of chest radiographs using the algorithm’s outputs led to an improvement in sensitivity among residents (73.4%, p = 0.003) but a reduction in specificity (94.3%, p < 0.001), suggesting that the algorithm aided in improving the sensitivity of resident evaluations.	Diagnostics
[35]	Kim, Chase, Chang, Kim, Park	Predicting Cardiac Arrest and Respiratory Failure Using Feasible Artificial Intelligence with Simple Trajectories of Patient Data.	Retrospective Cohort	Korea	Developed to predict cardiac arrest or acute respiratory failure 1 to 6 hours in advance, FAST-PACE utilizes a concise set of patient features from ICU data of 29,181 patients, excluding lab results. The AI model achieved an AUC of 0.886 for predicting cardiac arrest and 0.869 for respiratory failure 6 hours before occurrence, surpassing existing warning scores like MEWS and NEWS. The study demonstrates that FAST-PACE, relying solely on simple vital signs and patient history, outperforms traditional warning scores, offering a feasible and accurate tool for early prediction of adverse events in emergency situations.	Diagnostics
[36]	Landry, Ting, Zador, Sadeghian, Cusimano	Using artificial neural networks to identify patients with concussion and postconcussion syndrome based on antisaccades.	Prospective Cohort	Canada	Significant differences observed in prosaccade error rate and median antisaccade latency between control and concussion/PCS groups. Artificial neural networks (ANNs) achieved accuracies of 67% for concussion and 72% for PCS compared to controls. ANNs were unable to differentiate between concussion and PCS, indicating persistence of eye movement abnormalities in patients with PCS.	Diagnostics

Figure 1 Study selection flow chart and overview of exclusion schema.

Cheng CT, Lin HH, Hsu CP, et al. Deep Learning for automated detection and localization of traumatic abdominal solid organ injuries on CT scans. J Imaging Inform Med. 2024. doi:10.1007/s10278-024-01038-5

 Google Scholar

Michel J, Manns A, Boudersa S, et al. Clinical decision support system in emergency telephone triage: a scoping review of technical design, implementation and evaluation. Int J Med Inform. 2024;184:105347. doi:10.1016/j.ijmedinf.2024.105347

 PubMed Web of Science ®Google Scholar

Med P, Harris JD, Quatman CE, Manring MM, Siston RA, Flanigan DC. How to write a systematic review. Am J Sports Med. 2014;42(11):2761–2768. doi:10.1177/0363546513497567

 PubMed Web of Science ®Google Scholar

Russe MF, Rebmann P, Tran PH, et al. AI-based X-ray fracture analysis of the distal radius: accuracy between representative classification, detection and segmentation deep learning models for clinical practice. BMJ Open. 2024;14(1):e076954. doi:10.1136/bmjopen-2023-076954

 PubMed Web of Science ®Google Scholar

Piliuk K, Tomforde S. Artificial intelligence in emergency medicine. A systematic literature review. Int J Med Inform. 2023;180:105274. doi:10.1016/j.ijmedinf.2023.105274

 PubMed Web of Science ®Google Scholar

Choi J, Vendrow EB, Moor M, Spain DA. Development and validation of a model to quantify injury severity in real time. JAMA network open. 2023;6(10):e2336196. doi:10.1001/jamanetworkopen.2023.36196

 PubMed Web of Science ®Google Scholar

Gao Y, Soh NYT, Liu N, et al. Application of a deep learning algorithm in the detection of Hip fractures. iScience. 2023;26(8):107350. doi:10.1016/j.isci.2023.107350

 PubMed Web of Science ®Google Scholar

Sax DR, Warton EM, Sofrygin O, et al. Automated analysis of unstructured clinical assessments improves emergency department triage performance: a retrospective deep learning analysis. J Am Coll Emerg Physicians Open. 2023;4(4):e13003. doi:10.1002/emp2.13003

 PubMedGoogle Scholar

Ouyang CH, Chen CC, Tee YS, et al. The application of design thinking in developing a deep learning algorithm for hip fracture detection. Bioengineering. 2023;10(6):735. doi:10.3390/bioengineering10060735

 PubMedGoogle Scholar

Masoumian Hosseini M, Masoumian Hosseini ST, Qayumi K, Ahmady S, Koohestani HR. The aspects of running artificial intelligence in emergency care; a scoping review. Arch Acad Emerg Med. 2023;11(1):e38. doi:10.22037/aaem.v11i1.1974

 PubMedGoogle Scholar

He B, Dash D, Duanmu Y, Tan TX, Ouyang D, Zou J. Ai-Enabled Assessment Of Cardiac Function And Video Quality In Emergency Department Point-Of-Care Echocardiograms. J Emerg Med. 17:2023. doi:10.1016/j.jemermed.2023.02.005

 Google Scholar

Abrigo JM, Ko KL, Chen Q, et al. Artificial intelligence for detection of intracranial haemorrhage on head computed tomography scans: diagnostic accuracy in Hong Kong. Hong Kong Med J. 2023;29(2):112–120. doi:10.12809/hkmj209053

 PubMed Web of Science ®Google Scholar

Sundrani S, Chen J, Jin BT, Abad ZSH, Rajpurkar P, Kim D. Predicting patient decompensation from continuous physiologic monitoring in the emergency department. NPJ Digit Med. 2023;6(1):60. doi:10.1038/s41746-023-00803-0

 PubMed Web of Science ®Google Scholar

Inoue T, Maki S, Furuya T, et al. Automated fracture screening using an object detection algorithm on whole-body trauma computed tomography. Sci Rep. 2022;12(1):16549. doi:10.1038/s41598-022-20996-w

 PubMed Web of Science ®Google Scholar

Rashid T, Zia MS, Najam-Ur-Rehman M, Rauf T, Kadry S HT, Kadry S. A minority class balanced approach using the DCNN-LSTM method to detect human wrist fracture. Life. 2023;13(1):133. doi:10.3390/life13010133

 PubMedGoogle Scholar

Zech JR, Santomartino SM, Yi PH. Artificial Intelligence (AI) for fracture diagnosis: an overview of current products and considerations for clinical adoption, from the ajr special series on ai applications. AJR Am J Roentgenol. 2022;219(6):869–878. doi:10.2214/AJR.22.27873

 PubMed Web of Science ®Google Scholar

Wei J, Li D, Sing DC, et al. Detecting total Hip arthroplasty dislocations using deep learning: clinical and Internet validation. Emerg Radiol. 2022;29(5):801–808. doi:10.1007/s10140-022-02060-2

 PubMed Web of Science ®Google Scholar

Yao LH, Leung KC, Tsai CL, Huang CH, Fu LC. A novel deep learning-based system for triage in the emergency department using electronic medical records: retrospective cohort study. J Med Internet Res. 2021;23(12):e27008. doi:10.2196/27008

 PubMed Web of Science ®Google Scholar

Sánchez-Salmerón R, Gómez-Urquiza JL, Albendín-García L, et al. Machine learning methods applied to triage in emergency services: a systematic review. Int Emerg Nurs. 2022;60:101109. doi:10.1016/j.ienj.2021.101109

 PubMed Web of Science ®Google Scholar

Dipnall JF, Page R, Du L, et al. Predicting fracture outcomes from clinical registry data using artificial intelligence supplemented models for evidence-informed treatment (PRAISE) study protocol. PLoS One. 2021;16(9):e0257361. doi:10.1371/journal.pone.0257361

 PubMed Web of Science ®Google Scholar

Kim MW, Jung J, Park SJ, et al. Application of convolutional neural networks for distal radio-ulnar fracture detection on plain radiographs in the emergency room. Clin Exp Emerg Med. 2021;8(2):120–127. doi:10.15441/ceem.20.091

 PubMed Web of Science ®Google Scholar

Joseph JW, Leventhal EL, Grossestreuer AV, et al. Deep-learning approaches to identify critically Ill patients at emergency department triage using limited information. J Am Coll Emerg Physicians Open. 2020;1(5):773–781. doi:10.1002/emp2.12218

 PubMedGoogle Scholar

Miles J, Turner J, Jacques R, Williams J, Mason S. Using machine-learning risk prediction models to triage the acuity of undifferentiated patients entering the emergency care system: a systematic review. Diagn Progn Res. 2020;4:16. doi:10.1186/s41512-020-00084-1

 PubMedGoogle Scholar

Ozkaya E, Topal FE, Bulut T, Gursoy M, Ozuysal M, Karakaya Z. Evaluation of an artificial intelligence system for diagnosing scaphoid fracture on direct radiography. Eur J Trauma Emerg Surg. 2022;48(1):585–592. doi:10.1007/s00068-020-01468-0

 PubMed Web of Science ®Google Scholar

Weikert T, Noordtzij LA, Bremerich J, et al. Assessment of a deep learning algorithm for the detection of rib fractures on whole-body trauma computed tomography. Korean J Radiol. 2020;21(7):891–899. doi:10.3348/kjr.2019.0653

 PubMed Web of Science ®Google Scholar

Jalal S, Parker W, Ferguson D, Nicolaou S. Exploring the role of artificial intelligence in an emergency and trauma radiology department. Can Assoc Radiol J. 2021;72(1):167–174. doi:10.1177/0846537120918338

 PubMed Web of Science ®Google Scholar

Hwang EJ, Nam JG, Lim WH, et al. Deep Learning for Chest Radiograph Diagnosis in the Emergency Department. Radiology. 2019;293(3):573–580. doi:10.1148/radiol.2019191225

 PubMed Web of Science ®Google Scholar

Kim J, Chae M, Chang HJ, Kim YA, Park E. Predicting CARDIAC ARREST AND RESPIRATORY FAILURE USING FEASIBLE ARTIFICIAL INTELLIGENCE WITH SIMPLE TRAJECTORIES OF PATIENT DATa. J Clin Med. 2019;8(9):1336. doi:10.3390/jcm8091336

 PubMed Web of Science ®Google Scholar

Landry AP, Ting WKC, Zador Z, Sadeghian A, Cusimano MD. Using artificial neural networks to identify patients with concussion and postconcussion syndrome based on antisaccades. J Neurosurg. 2018;1–8. doi:10.3171/2018.6.JNS18607

 Web of Science ®Google Scholar