RÉSUMÉ
In this study, we developed a model to predict culture test results for pulmonary tuberculosis (PTB) with a customized multimodal approach and evaluated its performance in different clinical settings. Moreover, we investigated potential performance improvements by combining this approach with deep learning-based automated detection algorithms (DLADs). This retrospective observational study enrolled patients over 18 years of age who consecutively visited the level 1 emergency department and underwent chest radiograph and sputum testing. The primary endpoint was positive sputum culture for PTB. We compared the performance of the diagnostic models by replacing radiologists' interpretations of chest radiographs with screening scores calculated through DLAD. The optimal diagnostic model had an area under the receiver operating characteristic curve of 0.924 (95% CI 0.871-0.976) and an area under precision recall curve of 0.403 (95% CI 0.195-0.580) while maintaining a specificity of 81.4% when sensitivity was fixed at 90%. Multicomponent models showed improved performance for detecting PTB when chest radiography interpretation was replaced by DLAD. Multicomponent diagnostic models with DLAD customized for different clinical settings are more practical than traditional methods for detecting patients with PTB. This novel diagnostic approach may help prevent the spread of PTB and optimize healthcare resource utilization in resource-limited clinical settings.
Sujet(s)
Apprentissage profond , Tuberculose pulmonaire , Adulte , Humains , Algorithmes , Poumon , Radiographie thoracique/méthodes , Études rétrospectives , Sensibilité et spécificité , Tuberculose pulmonaire/imagerie diagnostiqueRÉSUMÉ
This study aimed to develop a machine learning-based clinical decision support system for emergency departments based on the decision-making framework of physicians. We extracted 27 fixed and 93 observation features using data on vital signs, mental status, laboratory results, and electrocardiograms during emergency department stay. Outcomes included intubation, admission to the intensive care unit, inotrope or vasopressor administration, and in-hospital cardiac arrest. eXtreme gradient boosting algorithm was used to learn and predict each outcome. Specificity, sensitivity, precision, F1 score, area under the receiver operating characteristic curve (AUROC), and area under the precision-recall curve were assessed. We analyzed 303,345 patients with 4,787,121 input data, resampled into 24,148,958 1 h-units. The models displayed a discriminative ability to predict outcomes (AUROC > 0.9), and the model with lagging 6 and leading 0 displayed the highest value. The AUROC curve of in-hospital cardiac arrest had the smallest change, with increased lagging for all outcomes. With inotropic use, intubation, and intensive care unit admission, the range of AUROC curve change with the leading 6 was the highest according to different amounts of previous information (lagging). In this study, a human-centered approach to emulate the clinical decision-making process of emergency physicians has been adopted to enhance the use of the system. Machine learning-based clinical decision support systems customized according to clinical situations can help improve the quality of care.
Sujet(s)
Aggravation clinique , Systèmes d'aide à la décision clinique , Arrêt cardiaque , Humains , Apprentissage machine , Arrêt cardiaque/diagnostic , Service hospitalier d'urgences , Études rétrospectivesRÉSUMÉ
In this retrospective observational study, we aimed to develop a machine-learning model using data obtained at the prehospital stage to predict in-hospital cardiac arrest in the emergency department (ED) of patients transferred via emergency medical services. The dataset was constructed by attaching the prehospital information from the National Fire Agency and hospital factors to data from the National Emergency Department Information System. Machine-learning models were developed using patient variables, with and without hospital factors. We validated model performance and used the SHapley Additive exPlanation model interpretation. In-hospital cardiac arrest occurred in 5431 of the 1,350,693 patients (0.4%). The extreme gradient boosting model showed the best performance with area under receiver operating curve of 0.9267 when incorporating the hospital factor. Oxygen supply, age, oxygen saturation, systolic blood pressure, the number of ED beds, ED occupancy, and pulse rate were the most influential variables, in that order. ED occupancy and in-hospital cardiac arrest occurrence were positively correlated, and the impact of ED occupancy appeared greater in small hospitals. The machine-learning predictive model using the integrated information acquired in the prehospital stage effectively predicted in-hospital cardiac arrest in the ED and can contribute to the efficient operation of emergency medical systems.
Sujet(s)
Arrêt cardiaque , Humains , Service hospitalier d'urgences , Apprentissage machine , Études rétrospectives , HôpitauxRÉSUMÉ
Models for predicting acute myocardial infarction (AMI) at the prehospital stage were developed and their efficacy compared, based on variables identified from a nationwide systematic emergency medical service (EMS) registry using conventional statistical methods and machine learning algorithms. Patients in the EMS cardiovascular registry aged >15 years who were transferred from the public EMS to emergency departments in Korea from January 2016 to December 2018 were enrolled. Two datasets were constructed according to the hierarchical structure of the registry. A total of 184,577 patients (Dataset 1) were included in the final analysis. Among them, 72,439 patients (Dataset 2) were suspected to have AMI at prehospital stage. Between the models derived using the conventional logistic regression method, the B-type model incorporated AMI-specific variables from the A-type model and exhibited a superior discriminative ability (p = 0.02). The models that used extreme gradient boosting and a multilayer perceptron yielded a higher predictive performance than the conventional logistic regression-based models for analyses that used both datasets. Each machine learning algorithm yielded different classification lists of the 10 most important features. Therefore, prediction models that use nationwide prehospital data and are developed with appropriate structures can improve the identification of patients who require timely AMI management.
RÉSUMÉ
Intermittent manual measurement of vital signs may not rapidly predict sepsis development in febrile patients admitted to the emergency department (ED). We aimed to evaluate the predictive performance of a wireless monitoring device that continuously measures heart rate (HR) and respiratory rate (RR) and a machine learning analysis in febrile but stable patients in the ED. We analysed 468 patients (age, ≥18 years; training set, n = 277; validation set, n = 93; test set, n = 98) having fever (temperature >38 °C) and admitted to the isolation care unit of the ED. The AUROC of the fragmented model with device data was 0.858 (95% confidence interval [CI], 0.809−0.908), and that with manual data was 0.841 (95% CI, 0.789−0.893). The AUROC of the accumulated model with device data was 0.861 (95% CI, 0.811−0.910), and that with manual data was 0.853 (95% CI, 0.803−0.903). Fragmented and accumulated models with device data detected clinical deterioration in febrile patients at risk of septic shock 9 h and 5 h 30 min earlier, respectively, than those with manual data. Continuous vital sign monitoring using a wearable device could accurately predict clinical deterioration and reduce the time to recognise potential clinical deterioration in stable ED patients with fever.
Sujet(s)
Aggravation clinique , Choc septique , Dispositifs électroniques portables , Adolescent , Service hospitalier d'urgences , Fièvre/diagnostic , Humains , Apprentissage machine , Choc septique/diagnostic , Signes vitaux/physiologieRÉSUMÉ
Upper gastrointestinal bleeding (UGIB) is a major cause of clinical deterioration worldwide. A large number of patients with UGIB cannot be diagnosed through endoscopy, which is normally the diagnostic method of choice. Therefore, this study aimed to investigate the diagnostic value of multi-detector computed tomography (MDCT) for patients with suspected UGIB. In this retrospective observational study of 386 patients, we compared contrast-enhanced abdominopelvic MDCT to endoscopy to analyze the performance of MDCT in identifying the status, location of origin, and etiology of UGIB. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were examined. In the assessment of bleeding status, MDCT was able to accurately identify 32.9% (21.9-43.9, 95% confidence interval [CI]) of patients with active bleeding, 27.4% (18.9-35.9, 95% CI) of patients with recent bleeding, and 94.8% (91.8-97.8, 95% CI) of patients without bleeding evidence (P<0.001). MDCT showed an accuracy of 60.9%, 60.6%, and 50.9% in identifying bleeding in the esophagus, stomach, and duodenum, respectively (P = 0.4028). The accuracy in differentiating ulcerative, cancerous, and variceal bleeding was 58.3%, 65.9%, and 56.6%, respectively (P = 0.6193). MDCT has limited use as a supportive screening method to identify the presence of gastrointestinal bleeding.
Sujet(s)
Varices oesophagiennes et gastriques , Service hospitalier d'urgences , Varices oesophagiennes et gastriques/complications , Hémorragie gastro-intestinale/étiologie , Humains , Tomodensitométrie multidétecteurs , Études rétrospectivesRÉSUMÉ
INTRODUCTION: Febrile neutropenia (FN) is one of the major complications with high mortality rates in cancer patients undergoing chemotherapy. The Multinational Association for Supportive Care in Cancer (MASCC) risk-index score has limited applicability for routine use in the emergency department (ED). This study aimed to develop simplified new nomograms that can predict 28-day mortality and the development of serious medical complications in patients with FN by using a combination of complete blood count (CBC) parameters with quick Sequential Organ Failure Assessment (qSOFA). METHODS: In this retrospective observational study, various models comprising qSOFA score and individual CBC parameters (red cell distribution width, delta neutrophil index, mean platelet volume (MPV)) were evaluated for association with outcomes by a multivariate logistic analysis. Subsequently, nomograms were developed for outcome prediction. The primary outcome was mortality at 28 days from ED presentation; the secondary outcome was the development of serious medical complications. RESULTS: A total of 378 patients were included. Among the CBC parameters, only MPV was significantly associated with 28-day mortality and serious medical complications in patients with FN. The nomogram developed to predict 28-day mortality and serious medical complications showed good discrimination with area under the receiver-operating characteristic curve (AUC) values of 0.729 and 0.862 (95% CI, 0.780-0.943), respectively, which were not different from those of the MASCC score (0.814, 95% CI, 0.705-0.922; p = .07 and 0.921, 95% CI, 0.863-0.979; p = .11, respectively) in the validation set. The calibration of both nomograms demonstrated good agreement in the validation set. CONCLUSION: In this study, a novel prognostic nomogram using qSOFA score and MPV to identify cancer patients with FN with high risk of 28-day mortality and serious medical complications was verified and validated. Prompt management of fatal complications of FN can be possible through early prediction of poor outcomes with these new nomograms.KEY MESSAGESAmong the evaluated CBC parameters, only mean platelet volume was associated with 28-day mortality and serious medical complications in cancer patients with febrile neutropenia.A novel and rapid prognostic nomogram was developed using quick Sequential Organ Failure Assessment score and mean platelet volume to identify cancer patients with febrile neutropenia having high risk of 28-day mortality and serious medical complications.The nomogram developed to predict 28-day mortality and serious medical complications in patients with febrile neutropenia showed good discrimination and provides rapid patient evaluation that is especially applicable in the emergency department.
Sujet(s)
Neutropénie fébrile , Tumeurs , Hémogramme , Service hospitalier d'urgences , Neutropénie fébrile/complications , Neutropénie fébrile/diagnostic , Humains , Tumeurs/complications , Tumeurs/traitement médicamenteux , Scores de dysfonction d'organes , Pronostic , Études rétrospectivesRÉSUMÉ
PURPOSE: Considering the risk of coronavirus disease (COVID-19) transmission through infected droplets, emergency department (ED) operations in response to febrile patients should be planned. We investigated the general and clinical characteristics of febrile patients visiting the ED and changes in admission rates via the ED during the COVID-19 outbreak. MATERIALS AND METHODS: We performed a retrospective analysis of prospectively collected patients who visited 402 EDs in the Republic of Korea with febrile symptoms between January 27 and May 31, 2020 and compared them to those enrolled before the COVID-19 outbreak. The primary outcome was admission rate; the secondary outcome was length of stay (LOS) in the ED. RESULTS: In total, 266519 patients had febrile symptoms at ED presentation after the COVID-19 outbreak. In 2019, before the outbreak, there were 437762 patients. The rate of ED visits among pediatric patients (aged <15 years) decreased to 21.4% after the COVID-19 outbreak, compared with 41.8% in 2019. The proportion of patients admitted after ED management was higher after the outbreak (31.3%) than before (25.2%). The adjusted odds ratio for admission was 1.04 (95% confidence interval: 1.02-1.05) after the outbreak. Compared to before the COVID-19 outbreak, the median ED LOS increased by 16 min after the outbreak. CONCLUSION: This study confirmed that admission rates and ED LOS increased for febrile patients visiting the ED after the COVID-19 outbreak. This could provide evidence for developing ED-related strategies in response to the ongoing COVID-19 outbreak and other infectious disease pandemics.
Sujet(s)
COVID-19 , Enfant , Épidémies de maladies , Service hospitalier d'urgences , Humains , Études rétrospectives , SARS-CoV-2RÉSUMÉ
INTRODUCTION: The coronavirus disease (COVID-19) pandemic has delayed the management of other serious medical conditions. This study presents an efficient method to prevent the degradation of the quality of diagnosis and treatment of other critical diseases during the pandemic. METHODS: We performed a retrospective observational study. The primary outcome was ED length of stay (ED LOS). The secondary outcomes were the door-to-balloon time in patients with suspected ST-segment elevation myocardial infarction and door-to-brain computed tomography time for patients with suspected stroke. The outcome measures were compared between patients who were treated in the red and orange zones designated as the changeable isolation unit and those who were treated in the non-isolation care unit. To control confounding factors, we performed propensity score matching, following which, outcomes were analyzed for non-inferiority. RESULTS: The mean ED LOS for hospitalized patients in the isolation and non-isolation care units were 406.5 min (standard deviation [SD], 237.9) and 360.2 min (SD, 226.4), respectively. The mean difference between the groups indicated non-inferiority of the isolation care unit (p = 0.037) but not in the patients discharged from the ED (p>0.999). The mean difference in the ED LOS for patients admitted to the ICU between the isolation and non-isolation care units was -22.0 min (p = 0.009). The mean difference in the door-to-brain computed tomography time between patients with suspected stroke in the isolation and non-isolation care units was 7.4 min for those with confirmed stroke (p = 0.013), and -20.1 min for those who were discharged (p = 0.012). The mean difference in the door-to-balloon time between patients who underwent coronary angiography in the isolation and non-isolation care units was -2.1 min (p<0.001). CONCLUSIONS: Appropriate and efficient handling of a properly planned ED plays a key role in improving the quality of medical care for other critical diseases during the COVID-19 outbreak.
Sujet(s)
COVID-19 , Service hospitalier d'urgences/organisation et administration , Durée du séjour , Infarctus du myocarde/diagnostic , Accident vasculaire cérébral/imagerie diagnostique , Adulte , Sujet âgé , Sujet âgé de 80 ans ou plus , Épidémies de maladies , Femelle , Humains , Mâle , Adulte d'âge moyen , Infarctus du myocarde/thérapie , Études rétrospectives , Accident vasculaire cérébral/thérapieRÉSUMÉ
Post contrast-acute kidney injury (PC-AKI) is defined as the deterioration of renal function after administration of iodinated contrast media. HMGB1 is known to play an important role in the development of acute kidney injury. The purpose of this study was to investigate the association between HMGB1 and PC-AKI and the protective effect of glycyrrhizin, a direct inhibitor of HMGB1, in rats. Rats were divided into three groups: control, PC-AKI and PC-AKI with glycyrrhizin. Oxidative stress was measured with MDA levels and H2DCFDA fluorescence intensity. The mRNA expressions of pro-inflammatory cytokines (IL-1α, IL-1ß, IL-6 and TNF-α) and kidney injury markers (KIM-1, NGAL and IL-18) were assessed using RT-PCR and ELISA in kidney tissue. In addition, the serum and intracellular protein levels of HMGB1were analyzed with the enzyme-linked immunosorbent assay (ELISA) and western blotting. Histologic changes were assessed with H&E staining using the transmission electron microscope (TEM). Moreover, serum creatinine (SCr), blood urea nitrogen (BUN) and lactate dehydrogenase (LDH) levels were assessed. Oxidative stress, pro-inflammatory cytokines, kidney injury markers and LDH were significantly higher in PC-AKI compared to the controls, but were lower in PC-AKI with glycyrrhizin. Intracellular and serum HMGB1 levels significantly increased after contrast media exposure, whereas they markedly decreased after glycyrrhizin pretreatment. SCr and BUN also decreased in PC-AKI with glycyrrhizin compared to PC-AKI. In PC-AKI, we could frequently observe tubular dilatation with H&E staining and cytoplasmic vacuoles on TEM, whereas these findings were attenuated in PC-AKI with glycyrrhizin. Our findings indicate that HMGB1 plays an important role in the development of PC-AKI and that glycyrrhizin has a protective effect against renal injury and dysfunction by inhibiting HMGB1 and reducing oxidative stress.
Sujet(s)
Atteinte rénale aigüe , Acide glycyrrhizique , Protéine HMGB1 , Animaux , Rein/métabolisme , Mâle , Stress oxydatif , RatsRÉSUMÉ
BACKGROUND: The American Heart Association guidelines recommend switching chest compression providers at least every 2 min depending on their fatigue during cardiopulmonary resuscitation (CPR). Although the provider's heart rate is widely used as an objective indicator for detecting fatigue, the accuracy of this measure is debatable. OBJECTIVES: This study was designed to determine whether real-time heart rate is a measure of fatigue in compression providers. STUDY DESIGN: A simulation-based prospective interventional study including 110 participants. METHODS: Participants performed chest compressions in pairs for four cycles using advanced cardiovascular life support simulation. Each participant's heart rate was measured using wearable healthcare devices, and qualitative variables regarding individual compressions were obtained from computerized devices. The primary outcome was correct depth of chest compressions. The main exposure was the change in heart rate, defined as the difference between the participant's heart rate during individual compressions and that before the simulation was initiated. RESULTS: With a constant compression duration for one cycle, the overall accuracy of compression depth significantly decreased with increasing heart rate. Female participants displayed significantly decreased accuracy of compression depth with increasing heart rate (odds ratio [OR]: 0.97; 95% confidence interval [CI]: 0.95-0.98; Pâ<â.001). Conversely, male participants displayed significantly improved accuracy with increasing heart rate (OR: 1.03; 95% CI: 1.02-1.04; Pâ<â.001). CONCLUSION: Increasing heart rate could reflect fatigue in providers performing chest compressions with a constant duration for one cycle. Thus, provider rotation should be considered according to objectively measured fatigue during CPR.
Sujet(s)
Réanimation cardiopulmonaire/méthodes , Techniciens médicaux des services d'urgence/statistiques et données numériques , Fatigue/physiopathologie , Rythme cardiaque/physiologie , Maladies professionnelles/physiopathologie , Adulte , Réanimation cardiopulmonaire/enseignement et éducation , Techniciens médicaux des services d'urgence/enseignement et éducation , Fatigue/diagnostic , Femelle , Humains , Mâle , Mannequins , Maladies professionnelles/diagnostic , Études prospectives , Formation par simulation , Travail/physiologie , Jeune adulteRÉSUMÉ
OBJECTIVES: This study aimed to validate the effectiveness of an emergency short-stay ward (ESSW) and its impact on clinical outcomes. METHODS: This retrospective observational study was performed at an urban tertiary hospital. An ESSW has been operating in this hospital since September 2017 to reduce emergency department (ED) boarding time and only targets patients indicated for admission to the general ward from the ED. Propensity-score matching was performed for comparison with the control group. The primary outcome was ED boarding time, and the secondary outcomes were subsequent intensive care unit (ICU) admission and 30-day in-hospital mortality. RESULTS: A total of 7461 patients were enrolled in the study; of them, 1523 patients (20.4%) were admitted to the ESSW. After propensity-score matching, there was no significant difference in the ED boarding time between the ESSW group and the control group (P = 0.237). Subsequent ICU admission was significantly less common in the ESSW group than in the control group (P < 0.001). However, the 30-day in-hospital mortality rate did not differ significantly between the two groups (P = 0.292). When the overall hospital bed occupancy ranged from 90% to 95%, the proportion of hospitalization was the highest in the ESSW group (29%). An interaction effect test using a general linear model confirmed that the ESSW served as an effect modifier with respect to bed occupancy and boarding time (P < 0.001). CONCLUSION: An ESSW can alleviate prolonged boarding time observed with hospital bed saturation. Moreover, the ESSW is associated with a low rate of subsequent ICU admission.
Sujet(s)
Taux d'occupation des lits/statistiques et données numériques , Service hospitalier d'urgences/organisation et administration , Mortalité hospitalière , Unités hospitalières/organisation et administration , Hospitalisation , Durée du séjour , Chambre de patient/ressources et distribution , Adulte , Sujet âgé , Surpeuplement , Femelle , Humains , Unités de soins intensifs/statistiques et données numériques , Modèles linéaires , Mâle , Adulte d'âge moyen , Transfert de patient , Score de propension , République de Corée , Études rétrospectives , Facteurs tempsRÉSUMÉ
BACKGROUND: An accurate disease severity score that can quickly predict the prognosis of patients with sepsis in the emergency department (ED) can aid clinicians in distributing resources appropriately or making decisions for active resuscitation measures. This study aimed to compare the prognostic performance of quick sequential organ failure assessment (qSOFA) with that of other disease severity scores in patients with septic shock presenting to an ED. METHODS: We performed a prospective, observational, registry-based study. The discriminative ability of each disease severity score to predict 28-day mortality was evaluated in the overall cohort (which included patients who fulfilled previously defined criteria for septic shock), the newly defined sepsis subgroup, and the newly defined septic shock subgroup. RESULTS: A total of 991 patients were included. All disease severity scores had poor discriminative ability for 28-day mortality. The sequential organ failure assessment and acute physiology and chronic health evaluation II scores had the highest area under the receiver-operating characteristic curve (AUC) values, which were significantly higher than the AUC values of other disease severity scores in the overall cohort and the sepsis and septic shock subgroups. The discriminative ability of each disease severity score decreased as the mortality rate of each subgroup increased. CONCLUSIONS: All disease severity scores, including qSOFA, did not display good discrimination for 28-day mortality in patients with serious infection and refractory hypotension or hypoperfusion; additionally, none of the included scoring tools in this study could consistently predict 28-day mortality in the newly defined sepsis and septic shock subgroups.