ABSTRACT
Recently, resuscitative endovascular balloon occlusion of the aorta (REBOA) had been introduced as an innovative procedure for severe hemorrhage in the abdomen or pelvis. We aimed to investigate risk factors associated with mortality after REBOA and construct a model for predicting mortality. This multicenter retrospective study collected data from 251 patients admitted at five regional trauma centers across South Korea from 2015 to 2022. The indications for REBOA included patients experiencing hypovolemic shock due to hemorrhage in the abdomen, pelvis, or lower extremities, and those who were non-responders (systolic blood pressure (SBP) < 90 mmHg) to initial fluid treatment. The primary and secondary outcomes were mortality due to exsanguination and overall mortality, respectively. After feature selection using the least absolute shrinkage and selection operator (LASSO) logistic regression model to minimize overfitting, a multivariate logistic regression (MLR) model and nomogram were constructed. In the MLR model using risk factors selected in the LASSO, five risk factors, including initial heart rate (adjusted odds ratio [aOR], 0.99; 95% confidence interval [CI], 0.98-1.00; p = 0.030), initial Glasgow coma scale (aOR, 0.86; 95% CI 0.80-0.93; p < 0.001), RBC transfusion within 4 h (unit, aOR, 1.12; 95% CI 1.07-1.17; p < 0.001), balloon occlusion type (reference: partial occlusion; total occlusion, aOR, 2.53; 95% CI 1.27-5.02; p = 0.008; partial + total occlusion, aOR, 2.04; 95% CI 0.71-5.86; p = 0.187), and post-REBOA systolic blood pressure (SBP) (aOR, 0.98; 95% CI 0.97-0.99; p < 0.001) were significantly associated with mortality due to exsanguination. The prediction model showed an area under curve, sensitivity, and specificity of 0.855, 73.2%, and 83.6%, respectively. Decision curve analysis showed that the predictive model had increased net benefits across a wide range of threshold probabilities. This study developed a novel intuitive nomogram for predicting mortality in patients undergoing REBOA. Our proposed model exhibited excellent performance and revealed that total occlusion was associated with poor outcomes, with post-REBOA SBP potentially being an effective surrogate measure.
Subject(s)
Aorta , Balloon Occlusion , Hospital Mortality , Nomograms , Resuscitation , Humans , Balloon Occlusion/methods , Male , Female , Retrospective Studies , Middle Aged , Resuscitation/methods , Adult , Endovascular Procedures/methods , Risk Factors , Wounds and Injuries/mortality , Wounds and Injuries/complications , Wounds and Injuries/therapy , Aged , Republic of Korea/epidemiology , Hemorrhage/mortality , Hemorrhage/therapy , Hemorrhage/etiology , Logistic ModelsABSTRACT
Flail chest is a severe injury to the chest wall and is related to adverse outcomes. A flail chest is classified as the physiologic, paradoxical motion of a chest wall or flail segment of rib fracture (RFX). We hypothesized that patients with paradoxical chest wall movement would present different clinical features from patients with a flail segment. This retrospective observational study included patients with blunt chest trauma who visited our level 1 trauma center between January 2019 and October 2022 and were diagnosed with one or more flail segments by computed tomography. The primary outcome of our study was a clinically diagnosed visible, paradoxical chest wall motion. We used the least absolute shrinkage and selection operator (LASSO) logistic regression model to minimize overfitting. After a feature selection using the LASSO regression model, we constructed a multivariable logistic regression (MLR) model and nomogram. A total of five risk factors were selected in the LASSO model and applied to the multivariable logistic regression model. Of these, four risk factors were statistically significant: the total number of RFX (adjusted OR [aOR], 1.28; 95% confidence interval [CI], 1.09-1.49; p = 0.002), number of segmental RFX including Grade III fractures (aOR, 1.78; 95% CI, 1.14-2.79; p = 0.012), laterally located primary fracture lines (aOR, 4.00; 95% CI, 1.69-9.43; p = 0.002), and anterior-lateral flail segments (aOR, 4.20; 95% CI, 1.60-10.99; p = 0.004). We constructed a nomogram to predict the personalized probability of the flail motion. A novel nomogram was developed in patients with flail segments of traumatic RFX to predict paradoxical chest wall motion. The number of RFX, Grade III segmental RFX, and the location of the RFX were significant risk factors.
Subject(s)
Flail Chest , Rib Fractures , Thoracic Injuries , Thoracic Wall , Wounds, Nonpenetrating , Humans , Rib Fractures/diagnostic imaging , Retrospective Studies , Nomograms , Fracture Fixation, Internal/methodsABSTRACT
Sarcopenia, a generalized loss of skeletal muscle mass that is primarily evident in the respiratory musculature, is associated with adverse outcomes in critically ill patients. However, the relationship between sarcopenia and ventilation-weaning outcomes has not yet been fully studied in patients with brain injuries. In this study, we examined the effect of reduced respiratory muscle mass on ventilation weaning in patients with brain injury. This observational study retrospectively reviewed the medical records of 73 patients with brain injury between January 2017 and December 2019. Thoracic skeletal muscle volumes were measured from thoracic CT images using the institute's three-dimensional modeling software program of our institute. The thoracic skeletal muscle volumes index (TSMVI) was normalized by dividing muscle volume by the square of patient height. Sarcopenia was defined as a TSMVI of less than the 50th sex-specific percentile. Among 73 patients with brain injury, 12 (16.5%) failed to wean from mechanical ventilation. The patients in the weaning-failure group had significantly higher sequential organ failure assessment scores [7.8â ±â 2.7 vs 6.1â ±â 2.2, Pâ =â .022] and lower thoracic skeletal muscle volume indexes [652.5â ±â 252.4 vs 1000.4â ±â 347.3, Pâ =â .002] compared with those in the weaning-success group. In multivariate analysis, sarcopenia was significantly associated with an increased risk of weaning failure (odds ratio 12.72, 95% confidence interval 2.87-70.48, Pâ =â .001). Our study showed a significant association between the TSMVI and ventilation weaning outcomes in patients with brain injury.
Subject(s)
Brain Injuries , Respiratory Insufficiency , Sarcopenia , Male , Female , Humans , Ventilator Weaning/methods , Retrospective Studies , Sarcopenia/etiology , Sarcopenia/complications , Respiration, Artificial/adverse effects , Muscle, Skeletal/diagnostic imaging , Respiratory Insufficiency/etiology , Brain , Brain Injuries/complicationsABSTRACT
Background and Objectives: Angioembolization has emerged as an effective therapeutic approach for pelvic hemorrhages; however, its exact effect size concerning the level of embolized artery remains uncertain. Therefore, we conducted this systematic review and meta-analysis to investigate the effect size of embolization-related pelvic complications after nonselective angioembolization compared to that after selective angioembolization in patients with pelvic injury accompanying hemorrhage. Materials and Methods: Relevant articles were collected by searching the PubMed, EMBASE, and Cochrane databases until 24 June 2023. Meta-analyses were conducted using odds ratios (ORs) for binary outcomes. Quality assessment was conducted using the risk of bias tool in non-randomized studies of interventions. Results: Five studies examining 357 patients were included in the meta-analysis. Embolization-related pelvic complications did not significantly differ between patients with nonselective and selective angioembolization (OR 1.581, 95% confidence interval [CI] 0.592 to 4.225, I2 = 0%). However, in-hospital mortality was more likely to be higher in the nonselective group (OR 2.232, 95% CI 1.014 to 4.913, I2 = 0%) than in the selective group. In the quality assessment, two studies were found to have a moderate risk of bias, whereas two studies exhibited a serious risk of bias. Conclusions: Despite the favorable outcomes observed with nonselective angioembolization concerning embolization-related pelvic complications, determining the exact effect sizes was limited owing to the significant risk of bias and heterogeneity. Nonetheless, the low incidence of ischemic pelvic complications appears to be a promising result.
Subject(s)
Embolization, Therapeutic , Hemorrhage , Humans , Hemorrhage/etiology , Hemorrhage/therapy , Arteries , Databases, Factual , Embolization, Therapeutic/adverse effects , Hospital MortalityABSTRACT
BACKGROUND: Within the trauma system, the emergency department (ED) is the hospital's first contact and is vital for allocating medical resources. However, there is generally limited information about patients that die in the ED. OBJECTIVE: The aim of this study was to develop an artificial intelligence (AI) model to predict trauma mortality and analyze pertinent mortality factors for all patients visiting the ED. METHODS: We used the Korean National Emergency Department Information System (NEDIS) data set (N=6,536,306), incorporating over 400 hospitals between 2016 and 2019. We included the International Classification of Disease 10th Revision (ICD-10) codes and chose the following input features to predict ED patient mortality: age, sex, intentionality, injury, emergent symptom, Alert/Verbal/Painful/Unresponsive (AVPU) scale, Korean Triage and Acuity Scale (KTAS), and vital signs. We compared three different feature set performances for AI input: all features (n=921), ICD-10 features (n=878), and features excluding ICD-10 codes (n=43). We devised various machine learning models with an ensemble approach via 5-fold cross-validation and compared the performance of each model with that of traditional prediction models. Lastly, we investigated explainable AI feature effects and deployed our final AI model on a public website, providing access to our mortality prediction results among patients visiting the ED. RESULTS: Our proposed AI model with the all-feature set achieved the highest area under the receiver operating characteristic curve (AUROC) of 0.9974 (adaptive boosting [AdaBoost], AdaBoost + light gradient boosting machine [LightGBM]: Ensemble), outperforming other state-of-the-art machine learning and traditional prediction models, including extreme gradient boosting (AUROC=0.9972), LightGBM (AUROC=0.9973), ICD-based injury severity scores (AUC=0.9328 for the inclusive model and AUROC=0.9567 for the exclusive model), and KTAS (AUROC=0.9405). In addition, our proposed AI model outperformed a cutting-edge AI model designed for in-hospital mortality prediction (AUROC=0.7675) for all ED visitors. From the AI model, we also discovered that age and unresponsiveness (coma) were the top two mortality predictors among patients visiting the ED, followed by oxygen saturation, multiple rib fractures (ICD-10 code S224), painful response (stupor, semicoma), and lumbar vertebra fracture (ICD-10 code S320). CONCLUSIONS: Our proposed AI model exhibits remarkable accuracy in predicting ED mortality. Including the necessity for external validation, a large nationwide data set would provide a more accurate model and minimize overfitting. We anticipate that our AI-based risk calculator tool will substantially aid health care providers, particularly regarding triage and early diagnosis for trauma patients.
Subject(s)
Artificial Intelligence , Fractures, Bone , Humans , Retrospective Studies , Republic of Korea , Emergency Service, HospitalABSTRACT
The direct consequences of chest trauma may cause adverse outcomes. Therefore, the early detection of high-risk patients and appropriate interventions can improve patient outcomes. This study aimed to investigate the risk factor for overall pulmonary complications in patients with blunt traumatic rib fractures. Prospectively recorded data of patients with blunt chest trauma in a level 1 trauma center between January 2019 and October 2022 were retrospectively analyzed. The primary outcomes were one or more pulmonary complications. To minimize the overfitting of the prediction model, we used the least absolute shrinkage and selection operator (LASSO) logistic regression. We input selected features using LASSO regression into the multivariable logistic regression model (MLR). We also constructed a nomogram to calculate approximate individual probability. Altogether, 542 patients were included. The LASSO regression model identified age, injury severity score (ISS), and flail motion of the chest wall as significant risk factors. In the MLR analysis, age (adjusted OR [aOR] 1.06; 95% confidence interval [CI] 1.03-1.08; p < 0.001), ISS (aOR 1.10; 95% CI 1.05-1.16; p < 0.001), and flail motion (aOR 8.82; 95% CI 4.13-18.83; p < 0.001) were significant. An MLR-based nomogram predicted the individual risk, and the area under the receiver operating characteristic curve was 0.826. We suggest a novel nomogram with good performance for predicting adverse pulmonary outcomes. The flail motion of the chest wall may be the most significant risk factor for pulmonary complications.
Subject(s)
Rib Fractures , Thoracic Injuries , Wounds, Nonpenetrating , Humans , Rib Fractures/complications , Thoracic Injuries/complications , Retrospective Studies , Nomograms , Wounds, Nonpenetrating/complicationsABSTRACT
Hypovolemia may be underestimated due to compensatory mechanisms. In this systematic review and meta-analysis, we investigated the diagnostic accuracy of a flat inferior vena cava (IVC) on computed tomography (CT) for predicting the development of shock and mortality in trauma patients. Relevant studies were obtained by searching PubMed, EMBASE, and Cochrane databases (articles up to 16 September 2022). The number of 2-by-2 contingency tables for the index test were collected. We adopted the Bayesian bivariate random-effects meta-analysis model. Twelve studies comprising a total of 1706 patients were included. The flat IVC on CT showed 0.46 pooled sensitivity (95% credible interval [CrI] 0.32-0.63), 0.87 pooled specificity (95% CrI 0.78-0.94), and 0.78 pooled AUC (95% CrI 0.58-0.93) for the development of shock. The flat IVC for mortality showed 0.48 pooled sensitivity (95% CrI 0.21-0.94), 0.70 pooled specificity (95% CrI 0.47-0.88), and 0.60 pooled AUC (95% CrI 0.26-0.89). Regarding the development of shock, flat IVC provided acceptable accuracy with high specificity. Regarding in-hospital mortality, the flat IVC showed poor accuracy. However, these results should be interpreted with caution due to the high risk of bias and substantial heterogeneity in some included studies.
ABSTRACT
BACKGROUND: Physical trauma-related mortality places a heavy burden on society. Estimating the mortality risk in physical trauma patients is crucial to enhance treatment efficiency and reduce this burden. The most popular and accurate model is the Injury Severity Score (ISS), which is based on the Abbreviated Injury Scale (AIS), an anatomical injury severity scoring system. However, the AIS requires specialists to code the injury scale by reviewing a patient's medical record; therefore, applying the model to every hospital is impossible. OBJECTIVE: We aimed to develop an artificial intelligence (AI) model to predict in-hospital mortality in physical trauma patients using the International Classification of Disease 10th Revision (ICD-10), triage scale, procedure codes, and other clinical features. METHODS: We used the Korean National Emergency Department Information System (NEDIS) data set (N=778,111) compiled from over 400 hospitals between 2016 and 2019. To predict in-hospital mortality, we used the following as input features: ICD-10, patient age, gender, intentionality, injury mechanism, and emergent symptom, Alert/Verbal/Painful/Unresponsive (AVPU) scale, Korean Triage and Acuity Scale (KTAS), and procedure codes. We proposed the ensemble of deep neural networks (EDNN) via 5-fold cross-validation and compared them with other state-of-the-art machine learning models, including traditional prediction models. We further investigated the effect of the features. RESULTS: Our proposed EDNN with all features provided the highest area under the receiver operating characteristic (AUROC) curve of 0.9507, outperforming other state-of-the-art models, including the following traditional prediction models: Adaptive Boosting (AdaBoost; AUROC of 0.9433), Extreme Gradient Boosting (XGBoost; AUROC of 0.9331), ICD-based ISS (AUROC of 0.8699 for an inclusive model and AUROC of 0.8224 for an exclusive model), and KTAS (AUROC of 0.1841). In addition, using all features yielded a higher AUROC than any other partial features, namely, EDNN with the features of ICD-10 only (AUROC of 0.8964) and EDNN with the features excluding ICD-10 (AUROC of 0.9383). CONCLUSIONS: Our proposed EDNN with all features outperforms other state-of-the-art models, including the traditional diagnostic code-based prediction model and triage scale.
Subject(s)
Artificial Intelligence , Humans , Hospital Mortality , Trauma Severity Indices , Injury Severity Score , Republic of Korea , Retrospective StudiesABSTRACT
Background and Objectives: Traumatic duodenal injury is a rare disease with limited evidence. We aimed to evaluate the risk factors for postoperative leakage and outcomes of pyloric exclusion after duodenal grade 2 and 3 injury. Materials and Methods: We reviewed a prospectively collected trauma database for the period January 2004-December 2020. Patients with grade 2 and 3 traumatic duodenal injury were included. To identify the risk factors for postoperative leakage, we used a stepwise multivariable logistic regression model and a least absolute shrinkage and selection operator (LASSO) logistic model. We constructed a receiver operator characteristic (ROC) curve to predict risk factors for postoperative leakage. Results: During the 17-year period, 179,887 trauma patients were admitted to a regional trauma center in Korea. Of these patients, 74 (0.04%) had duodenal injuries. A total of 49 consecutive patients had grade 2 and 3 traumatic duodenal injuries and underwent laparotomy. The incidence of postoperative leakage was 32.6% (16/49). Overall mortality was 18.4% (9/49). A stepwise multivariable logistic regression and LASSO logistic regression model showed that time from injury to initial operation was the sole statistically significant risk factor. The ROC curve at the optimal threshold of 15.77 h showed the following: area under ROC curve, 0.782; sensitivity, 68.8%; specificity, 87.9%; positive predictive value, 73.3%; and negative predictive value, 85.3%. There was no significant difference in outcomes between primary repair alone and pyloric exclusion. Conclusions: Time from injury to initial operation may be the sole significant risk factor for postoperative duodenal leakage. Pyloric exclusion may not be able to prevent postoperative leakage.
Subject(s)
Duodenum , Trauma Centers , Duodenum/injuries , Duodenum/surgery , Humans , Postoperative Period , Retrospective Studies , Risk FactorsABSTRACT
In this systematic review and meta-analysis, we aimed to investigate the efficacy and safety of laparoscopy for pediatric patients with abdominal trauma. Relevant articles were obtained by searching the MEDLINE PubMed, EMBASE, and Cochrane databases until 7 December 2021. Meta-analyses were performed using odds ratio (OR) for binary outcomes, standardized mean differences (SMDs) for continuous outcome measures, and overall proportion for single proportional outcomes. Nine studies examining 12,492 patients were included in our meta-analysis. Our meta-analysis showed younger age (SMD -0.47, 95% confidence interval (CI) -0.52 to -0.42), lower injury severity score (SMD -0.62, 95% CI -0.67 to -0.57), shorter hospital stay (SMD -0.55, 95% CI -0.60 to -0.50), less complications (OR 0.375, 95% CI 0.309 to 0.455), and lower mortality rate (OR 0.055, 95% CI 0.0.28 to 0.109) in the laparoscopy group compared to the laparotomy group. The majority of patients were able to avoid laparotomy (0.816, 95% CI 0.800 to 0.833). There were no missed injuries during the laparoscopic procedures in seven eligible studies. Laparoscopy for stable pediatric patients showed favorable outcomes in terms of morbidity and mortality. There were no missed injuries, and laparotomy could be avoided for the majority of patients.
ABSTRACT
Artificial intelligence (AI) technologies have been applied in various medical domains to predict patient outcomes with high accuracy. As AI becomes more widely adopted, the problem of model bias is increasingly apparent. In this study, we investigate the model bias that can occur when training a model using datasets for only one particular gender and aim to present new insights into the bias issue. For the investigation, we considered an AI model that predicts severity at an early stage based on the medical records of coronavirus disease (COVID-19) patients. For 5,601 confirmed COVID-19 patients, we used 37 medical records, namely, basic patient information, physical index, initial examination findings, clinical findings, comorbidity diseases, and general blood test results at an early stage. To investigate the gender-based AI model bias, we trained and evaluated two separate models-one that was trained using only the male group, and the other using only the female group. When the model trained by the male-group data was applied to the female testing data, the overall accuracy decreased-sensitivity from 0.93 to 0.86, specificity from 0.92 to 0.86, accuracy from 0.92 to 0.86, balanced accuracy from 0.93 to 0.86, and area under the curve (AUC) from 0.97 to 0.94. Similarly, when the model trained by the female-group data was applied to the male testing data, once again, the overall accuracy decreased-sensitivity from 0.97 to 0.90, specificity from 0.96 to 0.91, accuracy from 0.96 to 0.91, balanced accuracy from 0.96 to 0.90, and AUC from 0.97 to 0.95. Furthermore, when we evaluated each gender-dependent model with the test data from the same gender used for training, the resultant accuracy was also lower than that from the unbiased model.
ABSTRACT
The aim of the study is to develop artificial intelligence (AI) algorithm based on a deep learning model to predict mortality using abbreviate injury score (AIS). The performance of the conventional anatomic injury severity score (ISS) system in predicting in-hospital mortality is still limited. AIS data of 42,933 patients registered in the Korean trauma data bank from four Korean regional trauma centers were enrolled. After excluding patients who were younger than 19 years old and those who died within six hours from arrival, we included 37,762 patients, of which 36,493 (96.6%) survived and 1269 (3.4%) deceased. To enhance the AI model performance, we reduced the AIS codes to 46 input values by organizing them according to the organ location (Region-46). The total AIS and six categories of the anatomic region in the ISS system (Region-6) were used to compare the input features. The AI models were compared with the conventional ISS and new ISS (NISS) systems. We evaluated the performance pertaining to the 12 combinations of the features and models. The highest accuracy (85.05%) corresponded to Region-46 with DNN, followed by that of Region-6 with DNN (83.62%), AIS with DNN (81.27%), ISS-16 (80.50%), NISS-16 (79.18%), NISS-25 (77.09%), and ISS-25 (70.82%). The highest AUROC (0.9084) corresponded to Region-46 with DNN, followed by that of Region-6 with DNN (0.9013), AIS with DNN (0.8819), ISS (0.8709), and NISS (0.8681). The proposed deep learning scheme with feature combination exhibited high accuracy metrics such as the balanced accuracy and AUROC than the conventional ISS and NISS systems. We expect that our trial would be a cornerstone of more complex combination model.
Subject(s)
Wounds and Injuries/mortality , Abbreviated Injury Scale , Artificial Intelligence/statistics & numerical data , Benchmarking/statistics & numerical data , Databases, Factual/statistics & numerical data , Hospital Mortality , Humans , Injury Severity Score , Trauma Centers/statistics & numerical dataABSTRACT
Background and Objective: Breast mass lesions are common; however, determining the malignant potential of the lesion can be ambiguous. Recently, to evaluate breast mass lesions, vacuum-assisted excision (VAE) biopsy has been widely used for both diagnostic and therapeutic purposes. This study aimed to investigate the therapeutic role of VAE. Materials and Methods: Relevant articles were obtained by searching PubMed and EMBASE on 3 September 2021. Meta-analyses were performed using odds ratios and proportions. To assess heterogeneity, we conducted a subgroup analysis and meta-regression tests. Results: Finally, 26 studies comprising 18,170 patients were included. All of these were observational studies. The meta-analysis showed that the complete resection rate of VAE was 0.930. In the meta-regression test, there was no significant difference. The meta-analysis showed a recurrence rate of 0.039 in the VAE group. The meta-regression test showed no statistical significance. Postoperative hematoma, pain, and ecchymosis after VAE were 0.092, 0.082, and 0.075, respectively. Conclusion: VAE for benign breast lesions showed favorable outcomes with respect to complete resection and complications. This meta-analysis suggested that VAE for low-risk benign breast lesions is a reasonable option for both diagnostic and therapeutic purposes.
Subject(s)
Breast Neoplasms , Breast , Biopsy, Needle , Breast/surgery , Breast Neoplasms/surgery , Female , Humans , Image-Guided Biopsy , Retrospective Studies , VacuumABSTRACT
The efficacy and safety of laparoscopy for blunt trauma remain controversial. This systemic review and meta-analysis aimed to evaluate the usefulness of laparoscopy in blunt trauma. The PubMed, EMBASE, and Cochrane databases were searched up to 23 February 2021. Meta-analyses were performed using odds ratios (ORs), standardized mean differences (SMDs), and overall proportions. Overall, 19 studies with a total of 1520 patients were included. All patients were hemodynamically stable. In the laparoscopy group, meta-analysis showed lesser blood loss (SMD -0.28, 95% confidence interval (CI) -0.51 to -0.05, I2 = 62%) and shorter hospital stay (SMD -0.67, 95% CI -0.90 to -0.43, I2 = 47%) compared with the laparotomy group. Pooled prevalence of missed injury (0.003 (95% CI 0 to 0.023), I2 = 0%), nontherapeutic laparotomy (0.004 (95% CI 0.001 to 0.026), I2 = 0%), and mortality (0.021 (95% CI 0.010 to 0.043), I2 = 0%) were very low in blunt trauma. In subgroup analysis, recently published studies (2011-present) showed lesser conversion rate (0.115 (95% CI 0.067 to 0.190) vs. 0.391 (95% CI 0.247 to 0.556), test for subgroup difference: p < 0.01). This meta-analysis suggests that laparoscopy is a safe and feasible option in hemodynamic stable patients with blunt abdominal trauma.
ABSTRACT
BACKGROUND: The number of deaths from COVID-19 continues to surge worldwide. In particular, if a patient's condition is sufficiently severe to require invasive ventilation, it is more likely to lead to death than to recovery. OBJECTIVE: The goal of our study was to analyze the factors related to COVID-19 severity in patients and to develop an artificial intelligence (AI) model to predict the severity of COVID-19 at an early stage. METHODS: We developed an AI model that predicts severity based on data from 5601 COVID-19 patients from all national and regional hospitals across South Korea as of April 2020. The clinical severity of COVID-19 was divided into two categories: low and high severity. The condition of patients in the low-severity group corresponded to no limit of activity, oxygen support with nasal prong or facial mask, and noninvasive ventilation. The condition of patients in the high-severity group corresponded to invasive ventilation, multi-organ failure with extracorporeal membrane oxygenation required, and death. For the AI model input, we used 37 variables from the medical records, including basic patient information, a physical index, initial examination findings, clinical findings, comorbid diseases, and general blood test results at an early stage. Feature importance analysis was performed with AdaBoost, random forest, and eXtreme Gradient Boosting (XGBoost); the AI model for predicting COVID-19 severity among patients was developed with a 5-layer deep neural network (DNN) with the 20 most important features, which were selected based on ranked feature importance analysis of 37 features from the comprehensive data set. The selection procedure was performed using sensitivity, specificity, accuracy, balanced accuracy, and area under the curve (AUC). RESULTS: We found that age was the most important factor for predicting disease severity, followed by lymphocyte level, platelet count, and shortness of breath or dyspnea. Our proposed 5-layer DNN with the 20 most important features provided high sensitivity (90.2%), specificity (90.4%), accuracy (90.4%), balanced accuracy (90.3%), and AUC (0.96). CONCLUSIONS: Our proposed AI model with the selected features was able to predict the severity of COVID-19 accurately. We also made a web application so that anyone can access the model. We believe that sharing the AI model with the public will be helpful in validating and improving its performance.
Subject(s)
Artificial Intelligence , COVID-19/epidemiology , Adolescent , Adult , Aged , Aged, 80 and over , COVID-19/mortality , Child , Child, Preschool , Female , Humans , Infant , Infant, Newborn , Male , Middle Aged , Models, Statistical , Mortality , Republic of Korea/epidemiology , Research Design , Retrospective Studies , Risk Factors , SARS-CoV-2 , Young AdultABSTRACT
Background and objective: The early detection of underlying hemorrhage of pelvic trauma has been a critical issue. The aim of this study was to systematically determine the diagnostic accuracy of computed tomography (CT) for detecting severe pelvic hemorrhage. Materials and Methods: Relevant articles were obtained by searching PubMed, EMBASE, and Cochrane databases through 28 November 2020. Diagnostic test accuracy results were reviewed to obtain the sensitivity, specificity, diagnostic odds ratio, and summary receiver operating characteristic curve of CT for the diagnosis in pelvic trauma patients. The positive finding on CT was defined as the contrast extravasation. As the reference standard, severe pelvic hemorrhage was defined as an identification of bleeding at angiography or by direct inspection using laparotomy that required hemostasis by angioembolization or surgery. A subgroup analysis was performed according to the CT modality that is divided by the number of detector rows. Result: Thirteen eligible studies (29 subsets) were included in the present meta-analysis. Pooled sensitivity of CT was 0.786 [95% confidence interval (CI), 0.574-0.909], and pooled specificity was 0.944 (95% CI, 0.900-0.970). Pooled sensitivity of the 1-4 detector row group and 16-64 detector row group was 0.487 (95% CI, 0.215-0.767) and 0.915 (95% CI, 0.848-0.953), respectively. Pooled specificity of the 1-4 and 16-64 detector row groups was 0.956 (95% CI, 0.876-0.985) and 0.906 (95% CI, 0.828-0.951), respectively. Conclusion: Multi-detector CT with 16 or more detector rows has acceptable high sensitivity and specificity. Extravasation on CT indicates severe hemorrhage in patients with pelvic trauma.
Subject(s)
Pelvis , Tomography, X-Ray Computed , Angiography , Hemorrhage/diagnostic imaging , Hemorrhage/etiology , Humans , Pelvis/diagnostic imaging , Sensitivity and SpecificityABSTRACT
This systematic review and meta-analysis aimed to investigate the ultrasonographic variation of the diameter of the inferior vena cava (IVC), internal jugular vein (IJV), subclavian vein (SCV), and femoral vein (FV) to predict fluid responsiveness in critically ill patients. Relevant articles were obtained by searching PubMed, EMBASE, and Cochrane databases (articles up to 21 October 2021). The number of true positives, false positives, false negatives, and true negatives for the index test to predict fluid responsiveness was collected. We used a hierarchical summary receiver operating characteristics model and bivariate model for meta-analysis. Finally, 30 studies comprising 1719 patients were included in this review. The ultrasonographic variation of the IVC showed a pooled sensitivity and specificity of 0.75 and 0.83, respectively. The area under the receiver operating characteristics curve was 0.86. In the subgroup analysis, there was no difference between patients on mechanical ventilation and those breathing spontaneously. In terms of the IJV, SCV, and FV, meta-analysis was not conducted due to the limited number of studies. The ultrasonographic measurement of the variation in diameter of the IVC has a favorable diagnostic accuracy for predicting fluid responsiveness in critically ill patients. However, there was insufficient evidence in terms of the IJV, SCV, and FV.
ABSTRACT
BACKGROUND: COVID-19, which is accompanied by acute respiratory distress, multiple organ failure, and death, has spread worldwide much faster than previously thought. However, at present, it has limited treatments. OBJECTIVE: To overcome this issue, we developed an artificial intelligence (AI) model of COVID-19, named EDRnet (ensemble learning model based on deep neural network and random forest models), to predict in-hospital mortality using a routine blood sample at the time of hospital admission. METHODS: We selected 28 blood biomarkers and used the age and gender information of patients as model inputs. To improve the mortality prediction, we adopted an ensemble approach combining deep neural network and random forest models. We trained our model with a database of blood samples from 361 COVID-19 patients in Wuhan, China, and applied it to 106 COVID-19 patients in three Korean medical institutions. RESULTS: In the testing data sets, EDRnet provided high sensitivity (100%), specificity (91%), and accuracy (92%). To extend the number of patient data points, we developed a web application (BeatCOVID19) where anyone can access the model to predict mortality and can register his or her own blood laboratory results. CONCLUSIONS: Our new AI model, EDRnet, accurately predicts the mortality rate for COVID-19. It is publicly available and aims to help health care providers fight COVID-19 and improve patients' outcomes.
Subject(s)
COVID-19/mortality , Adult , Aged , Artificial Intelligence , China , Female , Hospitalization , Humans , Male , Middle Aged , Neural Networks, Computer , Republic of Korea , SARS-CoV-2ABSTRACT
In this study, we report a case of failed angioembolization of a ruptured liver hemangioma complicated by iatrogenic injury of the subclavian vein during catheter insertion. A 30-year-old woman experienced blunt trauma upon falling from her bed. Laceration of a seemingly preexisting hepatic hemangioma was diagnosed. No other injury was detected during a preoperative diagnostic workup. Subclavian vein catheterization was performed, followed by angioembolization to control bleeding due to the ruptured hemangioma. After angioembolization, the patient's systolic blood pressure and hemoglobin levels were 70 mmHg and 5.3 g/dL, respectively. She underwent emergency laparotomy. During the surgery, a large volume of blood in the abdominal cavity due to profuse bleeding from the ruptured hemangioma was observed. Because of a hemothorax found on chest radiography, we performed thoracoscopy, which revealed a large volume of blood in the right thoracic cavity and perforation of the subclavian vein by the catheter. After the damage-control surgery, the patient recovered safely. In this case, ruptured liver hemangioma complicated by subclavian vein catheter-related injury was treated safely using damage-control surgery. The catheter-related injury could be identified and treated using thoracoscopy.
Subject(s)
Catheterization/adverse effects , Embolization, Therapeutic/adverse effects , Hemangioma , Liver Neoplasms , Subclavian Vein/injuries , Adult , Catheterization/instrumentation , Catheters/adverse effects , Embolization, Therapeutic/instrumentation , Female , Hemangioma/physiopathology , Hemangioma/therapy , Humans , Iatrogenic Disease , Liver Neoplasms/physiopathology , Liver Neoplasms/therapy , Rupture, Spontaneous/physiopathology , Rupture, Spontaneous/therapy , Treatment FailureABSTRACT
BACKGROUND: Coronavirus disease (COVID-19) has spread explosively worldwide since the beginning of 2020. According to a multinational consensus statement from the Fleischner Society, computed tomography (CT) is a relevant screening tool due to its higher sensitivity for detecting early pneumonic changes. However, physicians are extremely occupied fighting COVID-19 in this era of worldwide crisis. Thus, it is crucial to accelerate the development of an artificial intelligence (AI) diagnostic tool to support physicians. OBJECTIVE: We aimed to rapidly develop an AI technique to diagnose COVID-19 pneumonia in CT images and differentiate it from non-COVID-19 pneumonia and nonpneumonia diseases. METHODS: A simple 2D deep learning framework, named the fast-track COVID-19 classification network (FCONet), was developed to diagnose COVID-19 pneumonia based on a single chest CT image. FCONet was developed by transfer learning using one of four state-of-the-art pretrained deep learning models (VGG16, ResNet-50, Inception-v3, or Xception) as a backbone. For training and testing of FCONet, we collected 3993 chest CT images of patients with COVID-19 pneumonia, other pneumonia, and nonpneumonia diseases from Wonkwang University Hospital, Chonnam National University Hospital, and the Italian Society of Medical and Interventional Radiology public database. These CT images were split into a training set and a testing set at a ratio of 8:2. For the testing data set, the diagnostic performance of the four pretrained FCONet models to diagnose COVID-19 pneumonia was compared. In addition, we tested the FCONet models on an external testing data set extracted from embedded low-quality chest CT images of COVID-19 pneumonia in recently published papers. RESULTS: Among the four pretrained models of FCONet, ResNet-50 showed excellent diagnostic performance (sensitivity 99.58%, specificity 100.00%, and accuracy 99.87%) and outperformed the other three pretrained models in the testing data set. In the additional external testing data set using low-quality CT images, the detection accuracy of the ResNet-50 model was the highest (96.97%), followed by Xception, Inception-v3, and VGG16 (90.71%, 89.38%, and 87.12%, respectively). CONCLUSIONS: FCONet, a simple 2D deep learning framework based on a single chest CT image, provides excellent diagnostic performance in detecting COVID-19 pneumonia. Based on our testing data set, the FCONet model based on ResNet-50 appears to be the best model, as it outperformed other FCONet models based on VGG16, Xception, and Inception-v3.