Artificial Intelligence-Based Electrocardiographic Biomarker for Outcome Prediction in Patients With Acute Heart Failure: Prospective Cohort Study.

BACKGROUND: Although several biomarkers exist for patients with heart failure (HF), their use in routine clinical practice is often constrained by high costs and limited availability. OBJECTIVE: We examined the utility of an artificial intelligence (AI) algorithm that analyzes printed electrocardiograms (ECGs) for outcome prediction in patients with acute HF. METHODS: We retrospectively analyzed prospectively collected data of patients with acute HF at two tertiary centers in Korea. Baseline ECGs were analyzed using a deep-learning system called Quantitative ECG (QCG), which was trained to detect several urgent clinical conditions, including shock, cardiac arrest, and reduced left ventricular ejection fraction (LVEF). RESULTS: Among the 1254 patients enrolled, in-hospital cardiac death occurred in 53 (4.2%) patients, and the QCG score for critical events (QCG-Critical) was significantly higher in these patients than in survivors (mean 0.57, SD 0.23 vs mean 0.29, SD 0.20; P<.001). The QCG-Critical score was an independent predictor of in-hospital cardiac death after adjustment for age, sex, comorbidities, HF etiology/type, atrial fibrillation, and QRS widening (adjusted odds ratio [OR] 1.68, 95% CI 1.47-1.92 per 0.1 increase; P<.001), and remained a significant predictor after additional adjustments for echocardiographic LVEF and N-terminal prohormone of brain natriuretic peptide level (adjusted OR 1.59, 95% CI 1.36-1.87 per 0.1 increase; P<.001). During long-term follow-up, patients with higher QCG-Critical scores (>0.5) had higher mortality rates than those with low QCG-Critical scores (<0.25) (adjusted hazard ratio 2.69, 95% CI 2.14-3.38; P<.001). CONCLUSIONS: Predicting outcomes in patients with acute HF using the QCG-Critical score is feasible, indicating that this AI-based ECG score may be a novel biomarker for these patients. TRIAL REGISTRATION: ClinicalTrials.gov NCT01389843; https://clinicaltrials.gov/study/NCT01389843.

Examining lack of referrals to heart valve specialists as mechanisms of potential underutilization of aortic valve replacement.

BACKGROUND: Recent studies suggest that aortic valve replacement (AVR) remains underutilized. AIMS: Investigate the potential role of non-referral to heart valve specialists (HVS) on AVR utilization. METHODS: Patients with severe aortic stenosis (AS) between 2015 and 2018, who met class I indication for intervention, were identified. Baseline data and process-related parameters were collected to analyze referral predictors and evaluate outcomes. RESULTS: Among 981 patients meeting criteria AVR, 790 patients (80.5%) were assessed by HVS within six months of index TTE. Factors linked to reduced referral included increasing age (OR: 0.95; 95% CI: 0.94-0.97; P < .001), unmarried status (OR: 0.59; 95% CI: 0.43-0.83; P = .002) and inpatient TTE (OR: 0.27; 95% CI: 0.19-0.38; P < .001). Conversely, higher hematocrit (OR: 1.13; 95% CI: 1.09-1.16; P < .001) and eGFR (OR: 1.01; 95% CI: 1.00-1.02; P = .003), mean aortic valve gradient (OR: 1.03; 95% CI: 1.01-1.04; P < .001) and preserved LVEF (OR: 1.59; 95% CI: 1.02-2.48; P = .04), were associated with increased referral likelihood. Moreover, patients assessed by HVS referral as a time-dependent covariate had a significantly lower two-year mortality risk than those who were not (aHR: 0.30; 95% CI: 0.23-0.39; P < .001). CONCLUSION: A substantial proportion of severe AS patients meeting indications for AVR are not evaluated by HVS and experience markedly increased mortality. Further research is warranted to assess the efficacy of care delivery mechanisms, such as e-consults, and telemedicine, to improve access to HVS expertise.

Performance of ECG-Derived Digital Biomarker for Screening Coronary Occlusion in Resuscitated Out-of-Hospital Cardiac Arrest Patients: A Comparative Study between Artificial Intelligence and a Group of Experts.

Acute coronary syndrome is a significant part of cardiac etiology contributing to out-of-hospital cardiac arrest (OHCA), and immediate coronary angiography has been proposed to improve survival. This study evaluated the effectiveness of an AI algorithm in diagnosing near-total or total occlusion of coronary arteries in OHCA patients who regained spontaneous circulation. Conducted from 1 July 2019 to 30 June 2022 at a tertiary university hospital emergency department, it involved 82 OHCA patients, with 58 qualifying after exclusions. The AI used was the Quantitative ECG (QCG™) system, which provides a STEMI diagnostic score ranging from 0 to 100. The QCG score's diagnostic performance was compared to assessments by two emergency physicians and three cardiologists. Among the patients, coronary occlusion was identified in 24. The QCG score showed a significant difference between occlusion and non-occlusion groups, with the former scoring higher. The QCG biomarker had an area under the curve (AUC) of 0.770, outperforming the expert group's AUC of 0.676. It demonstrated 70.8% sensitivity and 79.4% specificity. These findings suggest that the AI-based ECG biomarker could predict coronary occlusion in resuscitated OHCA patients, and it was non-inferior to the consensus of the expert group.

Artificial intelligence predicts undiagnosed atrial fibrillation in patients with embolic stroke of undetermined source using sinus rhythm electrocardiograms.

BACKGROUND: Artificial intelligence (AI)-enabled sinus rhythm (SR) electrocardiogram (ECG) interpretation can aid in identifying undiagnosed paroxysmal atrial fibrillation (AF) in patients with embolic stroke of undetermined source (ESUS). OBJECTIVE: The purpose of this study was to assess the efficacy of an AI model in identifying AF based on SR ECGs in patients with ESUS. METHODS: A transformer-based vision AI model was developed using 737,815 SR ECGs from patients with and without AF to detect current paroxysmal AF or predict the future development of AF within a 2-year period. Probability of AF was calculated from baseline SR ECGs using this algorithm. Its diagnostic performance was further tested in a cohort of 352 ESUS patients from 4 tertiary hospitals, all of whom were monitored using an insertable cardiac monitor (ICM) for AF surveillance. RESULTS: Over 25.1-month follow-up, AF episodes lasting ≥1 hour were identified in 58 patients (14.4%) using ICMs. In the receiver operating curve (ROC) analysis, the area under the curve for the AI algorithm to identify AF ≥1 hour was 0.806, which improved to 0.880 after integrating the clinical parameters into the model. The AI algorithm exhibited greater accuracy in identifying longer AF episodes (ROC for AF ≥12 hours: 0.837, for AF ≥24 hours: 0.879) and a temporal trend indicating that the AI-based AF risk score increased as the ECG recording approached the AF onset (P for trend <.0001). CONCLUSIONS: Our AI model demonstrated excellent diagnostic performance in predicting AF in patients with ESUS, potentially enhancing patient prognosis through timely intervention and secondary prevention of ischemic stroke in ESUS cohorts.

Smartphone AI vs. Medical Experts: A Comparative Study in Prehospital STEMI Diagnosis.

PURPOSE: Prehospital telecardiology facilitates early ST-elevation myocardial infarction (STEMI) detection, yet its widespread implementation remains challenging. Extracting digital STEMI biomarkers from printed electrocardiograms (ECGs) using phone cameras could offer an affordable and scalable solution. This study assessed the feasibility of this approach with real-world prehospital ECGs. MATERIALS AND METHODS: Patients suspected of having STEMI by emergency medical technicians (EMTs) were identified from a policy research dataset. A deep learning-based ECG analyzer (QCG™ analyzer) extracted a STEMI biomarker (qSTEMI) from prehospital ECGs. The biomarker was compared to a group of human experts, including five emergency medical service directors (board-certified emergency physicians) and three interventional cardiologists based on their consensus score (number of participants answering "yes" for STEMI). Non-inferiority of the biomarker was tested using a 0.100 margin of difference in sensitivity and specificity. RESULTS: Among 53 analyzed patients (24 STEMI, 45.3%), the area under the receiver operating characteristic curve of qSTEMI and consensus score were 0.815 (0.691-0.938) and 0.736 (0.594-0.879), respectively (p=0.081). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of qSTEMI were 0.750 (0.583-0.917), 0.862 (0.690-0.966), 0.826 (0.679-0.955), and 0.813 (0.714-0.929), respectively. For the consensus score, sensitivity, specificity, PPV, and NPV were 0.708 (0.500-0.875), 0.793 (0.655-0.966), 0.750 (0.600-0.941), and 0.760 (0.655-0.880), respectively. The 95% confidence interval of sensitivity and specificity differences between qSTEMI and consensus score were 0.042 (-0.099-0.182) and 0.103 (-0.043-0.250), respectively, confirming qSTEMI's non-inferiority. CONCLUSION: The digital STEMI biomarker, derived from printed prehospital ECGs, demonstrated non-inferiority to expert consensus, indicating a promising approach for enhancing prehospital telecardiology.

Impact of an Emergency Department Isolation Policy for Patients With Suspected COVID-19 on Door-to-Electrocardiography Time and Clinical Outcomes in Patients With Acute Myocardial Infarction.

BACKGROUND: Rapid electrocardiography diagnosis within 10 minutes of presentation is critical for acute myocardial infarction (AMI) patients in the emergency department (ED). However, the coronavirus disease 2019 (COVID-19) pandemic has significantly impacted the emergency care system. Screening for COVID-19 symptoms and implementing isolation policies in EDs may delay the door-to-electrocardiography (DTE) time. METHODS: We conducted a cross-sectional study of 1,458 AMI patients who presented to a single ED in South Korea from January 2019 to December 2021. We used multivariate logistic regression analysis to assess the impact of COVID-19 pandemic and ED isolation policies on DTE time and clinical outcomes. RESULTS: We found that the mean DTE time increased significantly from 5.5 to 11.9 minutes (P < 0.01) in ST segment elevation myocardial infarction (STEMI) patients and 22.3 to 26.7 minutes (P < 0.01) in non-ST segment elevation myocardial infarction (NSTEMI) patients. Isolated patients had a longer mean DTE time compared to non-isolated patients in both STEMI (9.2 vs. 24.4 minutes) and NSTEMI (22.4 vs. 61.7 minutes) groups (P < 0.01). The adjusted odds ratio (aOR) for the effect of COVID-19 duration on DTE ≥ 10 minutes was 1.93 (95% confidence interval [CI], 1.51-2.47), and the aOR for isolation status was 5.62 (95% CI, 3.54-8.93) in all patients. We did not find a significant association between in-hospital mortality and the duration of COVID-19 (aOR, 0.9; 95% CI, 0.52-1.56) or isolation status (aOR, 1.62; 95% CI, 0.71-3.68). CONCLUSION: Our study showed that ED screening or isolation policies in response to the COVID-19 pandemic could lead to delays in DTE time. Timely evaluation and treatment of emergency patients during pandemics are essential to prevent potential delays that may impact their clinical outcomes.

Hyperkalemia Detection in Emergency Departments Using Initial ECGs: A Smartphone AI ECG Analyzer vs. Board-Certified Physicians.

BACKGROUND: Hyperkalemia is a potentially fatal condition that mandates rapid identification in emergency departments (EDs). Although a 12-lead electrocardiogram (ECG) can indicate hyperkalemia, subtle changes in the ECG often pose detection challenges. An artificial intelligence application that accurately assesses hyperkalemia risk from ECGs could revolutionize patient screening and treatment. We aimed to evaluate the efficacy and reliability of a smartphone application, which utilizes camera-captured ECG images, in quantifying hyperkalemia risk compared to human experts. METHODS: We performed a retrospective analysis of ED hyperkalemic patients (serum potassium ≥ 6 mmol/L) and their age- and sex-matched non-hyperkalemic controls. The application was tested by five users and its performance was compared to five board-certified emergency physicians (EPs). RESULTS: Our study included 125 patients. The area under the curve (AUC)-receiver operating characteristic of the application's output was nearly identical among the users, ranging from 0.898 to 0.904 (median: 0.902), indicating almost perfect interrater agreement (Fleiss' kappa 0.948). The application demonstrated high sensitivity (0.797), specificity (0.934), negative predictive value (NPV) (0.815), and positive predictive value (PPV) (0.927). In contrast, the EPs showed moderate interrater agreement (Fleiss' kappa 0.551), and their consensus score had a significantly lower AUC of 0.662. The physicians' consensus demonstrated a sensitivity of 0.203, specificity of 0.934, NPV of 0.527, and PPV of 0.765. Notably, this performance difference remained significant regardless of patients' sex and age (P < 0.001 for both). CONCLUSION: Our findings suggest that a smartphone application can accurately and reliably quantify hyperkalemia risk using initial ECGs in the ED.

A simple scoring rule to predict survival to discharge after out of hospital cardiac arrest at the time of ED arrival.

BACKGROUND: It is important to be able to predict the chance of survival to hospital discharge upon ED arrival in order to determine whether to continue or terminate resuscitation efforts after out of hospital cardiac arrest. This study was conducted to develop and validate a simple scoring rule that could predict survival to hospital discharge at the time of ED arrival. METHODS: This was a multicenter retrospective cohort study based on a nationwide registry (Korean Cardiac Arrest Research Consortium) of out of hospital cardiac arrest (OHCA). The study included adult OHCA patients older than 18 years old, who visited one of 33 tertiary hospitals in South Korea from September 1st, 2015 to June 30th, 2020. Among 12,321 screened, 5471 patients were deemed suitable for analysis after exclusion. Pre-hospital ROSC, pre-hospital witness, shockable rhythm, initial pH, and age were selected as the independent variables. The dependent variable was set to be the survival to hospital discharge. Multivariable logistic regression (LR) was performed, and the beta-coefficients were rounded to the nearest integer to formulate the scoring rule. Several machine learning algorithms including the random forest classifier (RF), support vector machine (SVM), and K-nearest neighbor classifier (K-NN) were also trained via 5-fold cross-validation over a pre-specified grid, and validated on the test data. The prediction performances and the calibration curves of each model were obtained. Pre-processing of the registry was done using R, model training & optimization using Python. RESULTS: A total of 5471 patients were included in the analysis. The AUROC of the scoring rule over the test data was 0.7620 (0.7311-0.7929). The AUROCs of the machine learning classifiers (LR, SVM, k-NN, RF) were 0.8126 (0.7748-0.8505), 0.7920 (0.7512-0.8329), 0.6783 (0.6236-0.7329), and 0.7879 (0.7465-0.8294), respectively. CONCLUSION: A simple scoring rule consisting of five, binary variables could aid in the prediction of the survival to hospital discharge at the time of ED arrival, showing comparable results to conventional machine learning classifiers.

Feasibility of Artificial Intelligence-Based Electrocardiography Analysis for the Prediction of Obstructive Coronary Artery Disease in Patients With Stable Angina: Validation Study.

BACKGROUND: Despite accumulating research on artificial intelligence-based electrocardiography (ECG) algorithms for predicting acute coronary syndrome (ACS), their application in stable angina is not well evaluated. OBJECTIVE: We evaluated the utility of an existing artificial intelligence-based quantitative electrocardiography (QCG) analyzer in stable angina and developed a new ECG biomarker more suitable for stable angina. METHODS: This single-center study comprised consecutive patients with stable angina. The independent and incremental value of QCG scores for coronary artery disease (CAD)-related conditions (ACS, myocardial injury, critical status, ST-elevation myocardial infarction, and left ventricular dysfunction) for predicting obstructive CAD confirmed by invasive angiography was examined. Additionally, ECG signals extracted by the QCG analyzer were used as input to develop a new QCG score. RESULTS: Among 723 patients with stable angina (median age 68 years; male: 470/723, 65%), 497 (69%) had obstructive CAD. QCG scores for ACS and myocardial injury were independently associated with obstructive CAD (odds ratio [OR] 1.09, 95% CI 1.03-1.17 and OR 1.08, 95% CI 1.02-1.16 per 10-point increase, respectively) but did not significantly improve prediction performance compared to clinical features. However, our new QCG score demonstrated better prediction performance for obstructive CAD (area under the receiver operating characteristic curve 0.802) than the original QCG scores, with incremental predictive value in combination with clinical features (area under the receiver operating characteristic curve 0.827 vs 0.730; P<.001). CONCLUSIONS: QCG scores developed for acute conditions show limited performance in identifying obstructive CAD in stable angina. However, improvement in the QCG analyzer, through training on comprehensive ECG signals in patients with stable angina, is feasible.

Artificial intelligence versus physicians on interpretation of printed ECG images: Diagnostic performance of ST-elevation myocardial infarction on electrocardiography.

BACKGROUND: Smartphone-based ECG analyzer using camera input can be useful as everyone have it. The purpose of this study was to evaluate whether such a system can outperform clinicians in detecting ST-elevation myocardial infarction (STEMI) regardless of image acquisition conditions. METHODS: We retrospectively enrolled suspected STEMI patients in an emergency department from January to October 2021. A multifaceted cardiovascular assessment system (Quantitative ECG, QCG™) using ECG images to produce a quantitative score (QCG score, ranging from 0 to 100) was compared to human experts of 7 emergency physicians and 3 cardiologists. Voting scores (number of participants answering "yes" for STEMI) were calculated for comparison. The system's robustness was evaluated using an equivalence test where we prove its performance metric (area under the curve of the receiver operating characteristic curve, AUC-ROC) changes within a predetermined equivalence range (-0.01 to 0.01) in 6 different environments (A combination of three different smartphones and two image sources including computer screen and paper). RESULTS: 187 patients (96 STEMI, 51.3%) were analyzed. AUC-ROC of QCG score was 0.919 (0.880-0.957). AUC-ROCs of voting scores, 0.856 (0.799-0.913) for all clinicians, 0.843 (0.786-0.900) for emergency physicians, 0.817 (0.756-0.877) for cardiologists, and 0.848 (0.790-0.905) for high-performance group were significantly lower compared to that of QCG score. The change in AUC-ROC by image acquisition condition was negligible with a narrow confidence interval within -0.01 to 0.01 confirming the equivalence. CONCLUSIONS: Image-based AI system can outperform clinicians in STEMI diagnosis and its performance was robust to change in image acquisition conditions.

Trends in Utilization of Aortic Valve Replacement for Severe Aortic Stenosis.

BACKGROUND: Despite the rapid growth of aortic valve replacement (AVR) for aortic stenosis (AS), limited data suggest symptomatic severe AS remains undertreated. OBJECTIVES: This study sought to investigate temporal trends in AVR utilization among patients with a clinical indication for AVR. METHODS: Patients with severe AS (aortic valve area <1 cm2) on transthoracic echocardiograms from 2000 to 2017 at 2 large academic medical centers were classified based on clinical guideline indications for AVR and divided into 4 AS subgroups: high gradient with normal left ventricular ejection fraction (LVEF) (HG-NEF), high gradient with low LVEF (HG-LEF), low gradient with normal LVEF (LG-NEF), and low gradient with low LVEF (LG-LEF). Utilization of AVR was examined and predictors identified. RESULTS: Of 10,795 patients, 6,150 (57%) had an indication or potential indication for AVR, of whom 2,977 (48%) received AVR. The frequency of AVR varied by AS subtype with LG groups less likely to receive an AVR (HG-NEF: 70%, HG-LEF: 53%, LG-NEF: 32%, LG-LEF: 38%, P < 0.001). AVR volumes grew over the 18-year study period but were paralleled by comparable growth in the number of patients with an indication for AVR. In patients with a Class I indication, younger age, coronary artery disease, smoking history, higher hematocrit, outpatient index transthoracic echocardiogram, and LVEF ≥0.5 were independently associated with an increased likelihood of receiving an AVR. AVR was associated with improved survival in each AS-subgroup. CONCLUSIONS: Over an 18-year period, the proportion of patients with an indication for AVR who did not receive AVR has remained substantial despite the rapid growth of AVR volumes.

A Retrospective Clinical Evaluation of an Artificial Intelligence Screening Method for Early Detection of STEMI in the Emergency Department.

BACKGROUND: Rapid revascularization is the key to better patient outcomes in ST-elevation myocardial infarction (STEMI). Direct activation of cardiac catheterization laboratory (CCL) using artificial intelligence (AI) interpretation of initial electrocardiography (ECG) might help reduce door-to-balloon (D2B) time. To prove that this approach is feasible and beneficial, we assessed the non-inferiority of such a process over conventional evaluation and estimated its clinical benefits, including a reduction in D2B time, medical cost, and 1-year mortality. METHODS: This is a single-center retrospective study of emergency department (ED) patients suspected of having STEMI from January 2021 to June 2021. Quantitative ECG (QCG™), a comprehensive cardiovascular evaluation system, was used for screening. The non-inferiority of the AI-driven CCL activation over joint clinical evaluation by emergency physicians and cardiologists was tested using a 5% non-inferiority margin. RESULTS: Eighty patients (STEMI, 54 patients [67.5%]) were analyzed. The area under the curve of QCG score was 0.947. Binned at 50 (binary QCG), the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were 98.1% (95% confidence interval [CI], 94.6%, 100.0%), 76.9% (95% CI, 60.7%, 93.1%), 89.8% (95% CI, 82.1%, 97.5%) and 95.2% (95% CI, 86.1%, 100.0%), respectively. The difference in sensitivity and specificity between binary QCG and the joint clinical decision was 3.7% (95% CI, -3.5%, 10.9%) and 19.2% (95% CI, -4.7%, 43.1%), respectively, confirming the non-inferiority. The estimated median reduction in D2B time, evaluation cost, and the relative risk of 1-year mortality were 11.0 minutes (interquartile range [IQR], 7.3-20.0 minutes), 26,902.2 KRW (22.78 USD) per STEMI patient, and 12.39% (IQR, 7.51-22.54%), respectively. CONCLUSION: AI-assisted CCL activation using initial ECG is feasible. If such a policy is implemented, it would be reasonable to expect some reduction in D2B time, medical cost, and 1-year mortality.

Prediction of Neurological Outcomes in Out-of-hospital Cardiac Arrest Survivors Immediately after Return of Spontaneous Circulation: Ensemble Technique with Four Machine Learning Models.

BACKGROUND: We performed this study to establish a prediction model for 1-year neurological outcomes in out-of-hospital cardiac arrest (OHCA) patients who achieved return of spontaneous circulation (ROSC) immediately after ROSC using machine learning methods. METHODS: We performed a retrospective analysis of an OHCA survivor registry. Patients aged ≥ 18 years were included. Study participants who had registered between March 31, 2013 and December 31, 2018 were divided into a develop dataset (80% of total) and an internal validation dataset (20% of total), and those who had registered between January 1, 2019 and December 31, 2019 were assigned to an external validation dataset. Four machine learning methods, including random forest, support vector machine, ElasticNet and extreme gradient boost, were implemented to establish prediction models with the develop dataset, and the ensemble technique was used to build the final prediction model. The prediction performance of the model in the internal validation and the external validation dataset was described with accuracy, area under the receiver-operating characteristic curve, area under the precision-recall curve, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Futhermore, we established multivariable logistic regression models with the develop set and compared prediction performance with the ensemble models. The primary outcome was an unfavorable 1-year neurological outcome. RESULTS: A total of 1,207 patients were included in the study. Among them, 631, 139, and 153 were assigned to the develop, the internal validation and the external validation datasets, respectively. Prediction performance metrics for the ensemble prediction model in the internal validation dataset were as follows: accuracy, 0.9620 (95% confidence interval [CI], 0.9352-0.9889); area under receiver-operator characteristics curve, 0.9800 (95% CI, 0.9612-0.9988); area under precision-recall curve, 0.9950 (95% CI, 0.9860-1.0000); sensitivity, 0.9594 (95% CI, 0.9245-0.9943); specificity, 0.9714 (95% CI, 0.9162-1.0000); PPV, 0.9916 (95% CI, 0.9752-1.0000); NPV, 0.8718 (95% CI, 0.7669-0.9767). Prediction performance metrics for the model in the external validation dataset were as follows: accuracy, 0.8509 (95% CI, 0.7825-0.9192); area under receiver-operator characteristics curve, 0.9301 (95% CI, 0.8845-0.9756); area under precision-recall curve, 0.9476 (95% CI, 0.9087-0.9867); sensitivity, 0.9595 (95% CI, 0.9145-1.0000); specificity, 0.6500 (95% CI, 0.5022-0.7978); PPV, 0.8353 (95% CI, 0.7564-0.9142); NPV, 0.8966 (95% CI, 0.7857-1.0000). All the prediction metrics were higher in the ensemble models, except NPVs in both the internal and the external validation datasets. CONCLUSION: We established an ensemble prediction model for prediction of unfavorable 1-year neurological outcomes in OHCA survivors using four machine learning methods. The prediction performance of the ensemble model was higher than the multivariable logistic regression model, while its performance was slightly decreased in the external validation dataset.

Classification of cardioembolic stroke based on a deep neural network using chest radiographs.

BACKGROUND: Although chest radiographs have not been utilised well for classifying stroke subtypes, they could provide a plethora of information on cardioembolic stroke. This study aimed to develop a deep convolutional neural network that could diagnose cardioembolic stroke based on chest radiographs. METHODS: Overall, 4,064 chest radiographs of consecutive patients with acute ischaemic stroke were collected from a prospectively maintained stroke registry. Chest radiographs were randomly partitioned into training/validation (n = 3,255) and internal test (n = 809) datasets in an 8:2 ratio. A densely connected convolutional network (ASTRO-X) was trained to diagnose cardioembolic stroke based on chest radiographs. The performance of ASTRO-X was evaluated using the area under the receiver operating characteristic curve. Gradient-weighted class activation mapping was used to evaluate the region of focus of ASTRO-X. External testing was performed with 750 chest radiographs of patients with acute ischaemic stroke from 7 hospitals. FINDINGS: The areas under the receiver operating characteristic curve of ASTRO-X were 0.86 (95% confidence interval [CI], 0.83-0.89) and 0.82 (95% CI, 0.79-0.85) during the internal and multicentre external testing, respectively. The gradient-weighted class activation map demonstrated that ASTRO-X was focused on the area where the left atrium was located. Compared with cases predicted as non-cardioembolism by ASTRO-X, cases predicted as cardioembolism by ASTRO-X had higher left atrial volume index and lower left ventricular ejection fraction in echocardiography. INTERPRETATION: ASTRO-X, a deep neural network developed to diagnose cardioembolic stroke based on chest radiographs, demonstrated good classification performance and biological plausibility. FUNDING: Grant No. 14-2020-046 and 08-2016-051 from the Seoul National University Bundang Research Fund and NRF-2020M3E5D9079768 from the National Research Foundation of Korea.

Hospitalization for acute coronary syndrome increases the long-term risk of pneumonia: a population-based cohort study.

It is well established that the risk of acute coronary syndrome (ACS) increases after respiratory infection. However, the reverse association has not been evaluated. We tested the hypothesis that the long-term risk of pneumonia is increased after a new ACS event. A matched-cohort study was conducted using a nationally representative dataset. We identified patients with admission for ACS between 2004 and 2014, without a previous history of ACS or pneumonia. Incidence density sampling was used to match patients, on the basis of age and sex, to 3 controls who were also free from both ACS and pneumonia. We examined the incidence of pneumonia after ACS until the end of the cohort observation (Dec 31, 2014). The analysis cohort consisted of 5469 ACS cases and 16,392 controls (median age, 64 years; 68.3% men). The incidence rate ratios of the first and the total pneumonia episodes in the ACS group relative to the control group was 1.25 (95% confidence interval [CI], 1.11-1.41) and 1.23(95% CI 1.11-1.36), respectively. A significant ACS-related increase in the incidence of pneumonia was observed in the Cox-regression, shared frailty, and joint frailty model analyses, with hazard ratios of 1.25 (95% CI 1.09-1.42), 1.35 (95% CI 1.15-1.58), and 1.24 (95% CI 1.10-1.39), respectively. In this population-based cohort of patients who were initially free from both ACS and pneumonia, we found that hospitalization for ACS substantially increased the long term risk of pneumonia. This should be considered when formulating post-discharge care plans and preventive vaccination strategies in patients with ACS.

The index of oxygenation to respiratory rate as a prognostic factor for mortality in Sepsis.

OBJECTIVES: An index combining respiratory rate and oxygenation (ROX) has been introduced, and the ROX index is defined as the ratio of oxygen saturation by pulse oximetry/fraction of inspired oxygen to respiratory rate. In sepsis, hypoxemia and tachypnea are commonly observed. We performed this study to investigate the association between the ROX index and 28-day mortality in patients with sepsis or septic shock. METHODS: This retrospective study included 2862 patients. The patients were divided into three groups according to the ROX index: Group I (ROX index >20), Group II (ROX index >10 and ≤ 20), and Group III (ROX index ≤10). RESULTS: The median ROX index was significantly lower in the nonsurvivors than in the survivors (12.8 and 18.2, respectively) (p < 0.001). The 28-day mortality rates in Groups I, II and III were 14.5%, 21.3% and 34.4%, respectively (p < 0.001). In the multivariable Cox regression analysis, Group III had an approximately 40% higher risk of death than Group I during the 28-day period (hazard ratio = 1.41, 95% confidence interval 1.13-1.76). The area under the curve of the ROX index was significantly higher than that of the quick Sequential Organ Failure Assessment score (p < 0.001). CONCLUSIONS: The ROX index was lower in nonsurvivors than in survivors, and a ROX index less than or equal to 10 was an independent prognostic factor for 28-day mortality in patients with sepsis or septic shock. Therefore, the ROX index could be used as a prognostic marker in sepsis.

Risk of injury after emergency department visit for acute peripheral vertigo: a matched-cohort study.

OBJECTIVE: Peripheral vertigo is one of the most common causes of the emergency department (ED) visits. It can impair balance and might predispose patients to injuries after discharge. The purpose of this study was to determine whether peripheral vertigo is associated with an increased risk of trauma. METHODS: This matched-cohort study used the nationally representative dataset of de-identified claim information of 1 million randomly sampled individuals from a real Korean population, from 2002 to 2013. The exposure cohort included patients who visited EDs for new-onset peripheral vertigo without prior or concurrent injury. Each patient was randomly matched to five unexposed individuals (also without previous injury) by incidence density sampling. The primary outcome was a new injury within 1 year. The secondary outcomes were various injury subtypes. The time-dependent effect of the exposure was modeled using the extended Cox model. Age, sex, comorbidities, and household income level were included as covariates. RESULTS: A total of 776 and 3,880 individuals were included as the exposure and comparison cohorts, respectively. The risks of trunk injury and upper extremity injury were significantly higher in the exposure cohort. Extended Cox models with multivariable adjustment showed significantly increased risk for up to 1 year, with the first 1-month; 1 month to 3 months; and 3 months to 1 year hazard ratios of 5.23 (95% confidence interval [CI], 2.83-9.64); 1.50 (95% CI, 1.02-2.20); and 1.37 (95% CI, 1.11-1.68), respectively. CONCLUSION: Patients visiting EDs for acute peripheral vertigo are at a higher risk of a new injury for up to a year.

Effect of Prehospital Epinephrine on Outcomes of Out-of-Hospital Cardiac Arrest: A Bayesian Network Approach.

BACKGROUND: The benefit of prehospital epinephrine in out-of-hospital cardiac arrest (OHCA) was shown in a recent large placebo-controlled trial. However, placebo-controlled studies cannot identify the nonpharmacologic influences on concurrent or downstream events that might modify the main effect positively or negatively. We sought to identify the real-world effect of epinephrine from a clinical registry using Bayesian network with time-sequence constraints. METHODS: We analyzed a prospective regional registry of OHCA where a prehospital advanced life support (ALS) protocol named "Smart ALS (SALS)" was gradually implemented from July 2015 to December 2016. Using Bayesian network, a causal structure was estimated. The effect of epinephrine and SALS program was modelled based on the structure using extended Cox-regression and logistic regression, respectively. RESULTS: Among 4324 patients, SALS was applied to 2351 (54.4%) and epinephrine was administered in 1644 (38.0%). Epinephrine was associated with faster ROSC rate in nonshockable rhythm (HR: 2.02, 6.94, and 7.43; 95% CI: 1.08-3.78, 4.15-11.61, and 2.92-18.91, respectively, for 1-10, 11-20, and >20 minutes) while it was associated with slower rate up to 20 minutes in shockable rhythm (HR: 0.40, 0.50, and 2.20; 95% CI: 0.21-0.76, 0.32-0.77, and 0.76-6.33). SALS was associated with increased prehospital ROSC and neurologic recovery in noncardiac etiology (HR: 5.36 and 2.05; 95% CI: 3.48-8.24 and 1.40-3.01, respectively, for nonshockable and shockable rhythm). CONCLUSIONS: Epinephrine was associated with faster ROSC rate in nonshockable rhythm but slower rate in shockable rhythm up to 20 minutes. SALS was associated with improved prehospital ROSC and neurologic recovery in noncardiac etiology.

The Association of Extreme Tachycardia and Sustained Return of Spontaneous Circulation after Nontraumatic Out-of-Hospital Cardiac Arrest.

OBJECTIVE: Heart rate (HR), an essential vital sign that reflects hemodynamic stability, is influenced by myocardial oxygen demand, coronary blood flow, and myocardial performance. HR at the time of the return of spontaneous circulation (ROSC) could be influenced by the ß1-adrenergic effect of the epinephrine administered during cardiopulmonary resuscitation (CPR), and its effect could be decreased in patients who have the failing heart. We aimed to investigate the association between HR at the time of ROSC and the outcomes of adult out-of-hospital cardiac arrest (OHCA) patients. METHODS: This study was a secondary analysis of a cardiac arrest registry from a single institution from January 2008 to July 2014. The OHCA patients who achieved ROSC at the emergency department (ED) were included, and HR was retrieved from an electrocardiogram or vital sign at the time of ROSC. The patients were categorized into four groups according to the HR (bradycardia (HR < 60), normal HR (60 ≤ HR ≤ 100), tachycardia (100 < HR < 150), and extreme tachycardia (HR ≥ 150)). The primary outcome was the rate of sustained ROSC and the secondary outcomes were the rate of one-month survival and six-month good neurologic outcome. RESULTS: A total of 330 patients were included. In the univariate logistic regression model, the rate of sustained ROSC increased by 17% as HR increased by every 10 beats per minute (bpm) (odds ratio (OR), 1.171; 95% confidence interval (CI), 1.077-1.274, p < 0.001). In the multivariate logistic regression model, extreme tachycardia was independently associated with a high probability of sustained ROSC compared to normal heart rate (OR, 15.96; 95% CI, 2.04-124.93, p=0.008). CONCLUSION: Extreme tachycardia (HR ≥ 150) at the time of ROSC is independently associated with a high probability of sustained ROSC in nontraumatic adult OHCA patients.

Moderate to severe hyperphosphataemia as an independent prognostic factor for 28-day mortality in adult patients with sepsis.

BACKGROUND: Ischaemic tissue injury caused by tissue hypoperfusion is one of the major consequences of sepsis. Phosphate concentrations are elevated in ischaemic tissue injury. This study was performed to investigate the association of phosphate concentrations with mortality in patients with sepsis. METHODS: This was a retrospective cohort study of patients with sepsis conducted at an urban, tertiary care emergency department (ED) in Korea. Patients with sepsis arriving between March 2010 and April 2017 were stratified into four groups according to the initial phosphate concentration at presentation to the ED: group I (hypophosphataemia, phosphate <2 mg/dL), group II (normophosphataemia, phosphate 2-4 mg/dL), group III (mild hyperphosphataemia, phosphate 4-6 mg/dL), group IV (moderate to severe hyperphosphataemia, phosphate ≥6 mg/dL). Multivariable Cox proportional hazard regression analyses were performed to evaluate the independent association of initial phosphate concentration with 28-day mortality. RESULTS: Of the 3034 participants in the study, the overall mortality rate was 21.9%. The 28-day mortality rates were group I (hypophosphataemia) 14.6%, group II 17.4% (normophosphataemia), group III (mild hyperphosphataemia) 29.2% and group IV (moderate to severe hyperphosphataemia) 51.4%, respectively (p<0.001). In the multivariable analyses, patients with severe hyperphosphataemia had a significantly higher risk of death than those with normal phosphate levels (HR 1.59; 95% CI 1.23 to 2.05). Mortality in the other groups was not significantly different from mortality in patients with normophosphataemia. CONCLUSIONS: Moderate to severe hyperphosphataemia was associated with 28-day mortality in patients with sepsis. Phosphate level could be used as a prognostic indicator in sepsis.