Improving detection of impacted animal bones on lateral neck radiograph using a deep learning artificial intelligence algorithm.

OBJECTIVE: We aimed to develop a deep learning artificial intelligence (AI) algorithm to detect impacted animal bones on lateral neck radiographs and to assess its effectiveness for improving the interpretation of lateral neck radiographs. METHODS: Lateral neck radiographs were retrospectively collected for patients with animal bone impaction between January 2010 and March 2020. Radiographs were then separated into training, validation, and testing sets. A total of 1733 lateral neck radiographs were used to develop the deep learning algorithm. The testing set was assessed for the stand-alone deep learning AI algorithm and for human readers (radiologists, radiology residents, emergency physicians, ENT physicians) with and without the aid of the AI algorithm. Another radiograph cohort, collected from April 1, 2020, to June 30, 2020, was analyzed to simulate clinical application by comparing the deep learning AI algorithm with radiologists' reports. RESULTS: In the testing set, the sensitivity, specificity, and accuracy of the AI model were 96%, 90%, and 93% respectively. Among the human readers, all physicians of different subspecialties achieved a higher accuracy with AI-assisted reading than without. In the simulation set, among the 20 cases positive for animal bones, the AI model accurately identified 3 more cases than the radiologists' reports. CONCLUSION: Our deep learning AI model demonstrated a higher sensitivity for detection of animal bone impaction on lateral neck radiographs without an increased false positive rate. The application of this model in a clinical setting may effectively reduce time to diagnosis, accelerate workflow, and decrease the use of CT.

Deep Learning-Based Automatic Detection of ASPECTS in Acute Ischemic Stroke: Improving Stroke Assessment on CT Scans.

(1) Background: The Alberta Stroke Program Early CT Score (ASPECTS) is a standardized scoring tool used to evaluate the severity of acute ischemic stroke (AIS) on non-contrast CT (NCCT). Our aim in this study was to automate ASPECTS. (2) Methods: We utilized a total of 258 patient images with suspected AIS symptoms. Expert ASPECTS readings on NCCT were used as ground truths. A deep learning-based automatic detection (DLAD) algorithm was developed for automated ASPECTS scoring based on 168 training patient images using a convolutional neural network (CNN) architecture. An additional 90 testing patient images were used to evaluate the performance of the DLAD algorithm, which was then compared with ASPECTS readings on NCCT as performed by physicians. (3) Results: The sensitivity, specificity, and accuracy of DLAD for the prediction of ASPECTS were 65%, 82%, and 80%, respectively. These results demonstrate that the DLAD algorithm was not inferior to radiologist-read ASPECTS on NCCT. With the assistance of DLAD, the individual sensitivity of the ER physician, neurologist, and radiologist improved. (4) Conclusion: The proposed DLAD algorithm exhibits a reasonable ability for ASPECTS scoring on NCCT images in patients presenting with AIS symptoms. The DLAD algorithm could be a valuable tool to improve and accelerate the decision-making process of front-line physicians.

Predicting in-hospital mortality in adult non-traumatic emergency department patients: a retrospective comparison of the Modified Early Warning Score (MEWS) and machine learning approach.

BACKGROUND: A feasible and accurate risk prediction systems for emergency department (ED) patients is urgently required. The Modified Early Warning Score (MEWS) is a wide-used tool to predict clinical outcomes in ED. Literatures showed that machine learning (ML) had better predictability in specific patient population than traditional scoring system. By analyzing a large multicenter dataset, we aim to develop a ML model to predict in-hospital morality of the adult non traumatic ED patients for different time stages, and comparing performance with other ML models and MEWS. METHODS: A retrospective observational cohort study was conducted in five Taiwan EDs including two tertiary medical centers and three regional hospitals. All consecutively adult (>17 years old) non-traumatic patients admit to ED during a 9-year period (January first, 2008 to December 31th, 2016) were included. Exclusion criteria including patients with (1) out-of-hospital cardiac arrest and (2) discharge against medical advice and transferred to other hospital (3) missing collect variables. The primary outcome was in-hospital mortality and were categorized into 6, 24, 72, 168 hours mortality. MEWS was calculated by systolic blood pressure, pulse rate, respiratory rate, body temperature, and level of consciousness. An ensemble supervised stacking ML model was developed and compared to sensitive and unsensitive Xgboost, Random Forest, and Adaboost. We conducted a performance test and examine both the area under the receiver operating characteristic (AUROC) and the area under the precision and recall curve (AUPRC) as the comparative measures. RESULT: After excluding 182,001 visits (7.46%), study group was consisted of 24,37,326 ED visits. The dataset was split into 67% training data and 33% test data for ML model development. There was no statistically difference found in the characteristics between two groups. For the prediction of 6, 24, 72, 168 hours in-hospital mortality, the AUROC of MEW and ML mode was 0.897, 0.865, 0.841, 0.816 and 0.939, 0.928, 0.913, 0.902 respectively. The stacking ML model outperform other ML model as well. For the prediction of in-hospital mortality over 48-hours, AUPRC performance of MEWS drop below 0.1, while the AUPRC of ML mode was 0.317 in 6 hours and 0.2150 in 168 hours. For each time frame, ML model achieved statistically significant higher AUROC and AUPRC than MEWS (all P < 0.001). Both models showed decreasing prediction ability as time elapse, but there was a trend that the gap of AUROC values between two model increases gradually (P < 0.001). Three MEWS thresholds (score >3, >4, and >5) were determined as baselines for comparison, ML mode consistently showed improved or equally performance in sensitivity, PPV, NPV, but not in specific. CONCLUSION: Stacking ML methods improve predicted in-hospital mortality than MEWS in adult non-traumatic ED patients, especially in the prediction of delayed mortality.

Short-Term Effects of Particulate Matter and Its Constituents on Emergency Room Visits for Chronic Obstructive Pulmonary Disease: A Time-Stratified Case-Crossover Study in an Urban Area.

BACKGROUND: PM2.5 exposure is associated with pulmonary and airway inflammation, and the health impact might vary by PM2.5 constitutes. This study evaluated the effects of increased short-term exposure to PM2.5 constituents on chronic obstructive pulmonary disease (COPD)-related emergency department (ED) visits and determined the susceptible groups. METHODS: This retrospective observational study performed in a medical center from 2007 to 2010, and enrolled non-trauma patients aged >20 years who visited the emergency department (ED) and were diagnosed as COPD. Concentrations of PM2.5, PM10, and the four PM2.5 components, including organic carbon (OC), elemental carbon (EC), nitrate (NO3-), and sulfate (SO42-), were collected by three PM supersites in Kaohsiung City. We used an alternative design of the Poisson time series regression models called a time-stratified and case-crossover design to analyze the data. RESULTS: Per interquartile range (IQR) increment in PM2.5 level on lag 2 were associated with increments of 6.6% (95% confidence interval (CI), 0.5-13.0%) in risk of COPD exacerbation. An IQR increase in elemental carbon (EC) was significantly associated with an increment of 3.0% (95% CI, 0.1-5.9%) in risk of COPD exacerbation on lag 0. Meanwhile, an IQR increase in sulfate, nitrate, and OC levels was not significantly associated with COPD. Patients were more sensitive to the harmful effects of EC on COPD during the warm season (interaction p = 0.019). The risk of COPD exacerbation after exposure to PM2.5 was higher in individuals who are currently smoking, with malignancy, or during cold season, but the differences did not achieve statistical significance. CONCLUSION: PM2.5 and EC may play an important role in COPD events in Kaohsiung, Taiwan. Patients were more susceptible to the adverse effects of EC on COPD on warm days.

Benefits of Early In-Hospital Antivenom Administration to Patients with Protobothrops mucrosquamatus Envenomation.

Protobothrops mucrosquamatus is one of the common venomous snakes in Southeast Asia. This retrospective cohort study conducted in six medical institutions in Taiwan aimed to obtain information on the optimal management strategies for P. mucrosquamatus snakebite envenomation. Data were extracted from the Chang Gung Research Database from January 2006 to December 2016. The association between early antivenom administration and patient demographics, pain requiring treatment with analgesic injections, and hospital length of stay was analyzed. A total of 195 patients were enrolled; 130 were administered antivenom within 1 hour after emergency department arrival (early group), whereas 65 were treated later than 1 hour after arrival (late group). No in-hospital mortality was identified. The difference in surgical intervention rates between the early and late groups was statistically insignificant (P = 0.417). Compared with the early group, the late group showed a higher rate of antivenom skin test performance (46.9% versus 63.1%, respectively, P = 0.033), longer hospital stay (42 ± 62 hours versus 99 ± 70 hours, respectively, P = 0.016), and higher rate of incidences of pain requiring treatment with analgesic injections (29.2% versus 46.2%, respectively, P = 0.019). After adjusting for confounding factors, early antivenom administration was associated with decreased pain requiring treatment with analgesic injections (adjusted odds ratio: 0.51, 95% CI: 0.260-0.985). Antivenom administration within 1 hour of arrival was associated with a decreased likelihood of experiencing pain and hospital length of stay in patients with P. mucrosquamatus snakebites. Antivenom skin testing was associated with delays in antivenom administration.

Medical Liability of Residents in Taiwan Criminal Court: An Analysis of Closed Malpractice Cases.

OBJECTIVE: By analyzing closed criminal malpractice claims involving resident physicians, we aimed to clarify the characteristics of litigations and examine the litigious errors leading to guilty verdicts. DESIGN: A retrospective descriptive study. Setting/Study Participants. The verdicts pertaining to physicians recorded on the national database of the Taiwan justice system were reviewed. Main Outcome Measures. The characteristics of litigations were documented. Negligence and guilty verdicts were further analyzed to identify litigious errors. RESULTS: Between January 1, 2000, and December 31, 2014, from a total of 436 closed criminal malpractice cases, 40 included resident physicians. Five (12.5%) cases received guilty verdicts with mean imprisonment sentences of 5.4 ± 4.1 months. An average of 77.2 months was required for the final adjudication, and surgery residents were involved most frequently (38.9%). Attending physicians were codefendants in 82.5% of cases and were declared guilty in 60% of them. Sepsis (37.5%) was the most common disease in the 40 cases examined, followed by operation/procedure complications (25%). Performance errors (70%) were more than twice as common than diagnostic errors (30%), but the percentage of guilty verdicts in performance error cases was much lower (7.1% vs. 25%). Four negligence cases received nonguilty verdicts, which were mostly due to lack of causation. CONCLUSION: Closed criminal malpractice cases involving residents took on average 6.22 years to conclude. Performance errors accounted for 70% of cases, with treatment of sepsis and operation/procedure complications predominant. To reduce medicolegal risk, residents should learn experiences from analyzing malpractice cases to avoid similar litigious pitfalls.

Comparison of Injury Severity Score, Glasgow Coma Scale, and Revised Trauma Score in Predicting the Mortality and Prolonged ICU Stay of Traumatic Young Children: A Cross-Sectional Retrospective Study.

INTRODUCTION: The purpose of this study was to examine the capacity of commonly used trauma scoring systems such as the Glasgow Coma Scale (GCS), Injury Severity Score (ISS), and Revised Trauma Score (RTS) to predict outcomes in young children with traumatic injuries. METHODS: This retrospective study was conducted for the period from 2009 to 2016 in Kaohsiung Chang Gung Memorial Medical Hospital, a level I trauma center. We included all children under the age of 6 years admitted to the hospital via the emergency department with any traumatic injury and compared the trauma scores of GCS, ISS, and RTS on patients' outcome. The primary outcomes were mortality and prolonged Intensive Care Unit (ICU) stay, with the latter defined as an ICU stay longer than 14 days. The secondary outcome was the hospital length of stay (HLOS). Receiver operating characteristic (ROC) analysis was also adopted with the value of the area under the ROC curve (AUC) for comparing trauma score prediction with patient mortality. Cutoff values from each trauma score for mortality prediction were also measured by determining the point along the ROC curve where Youden's index was maximum. RESULTS: We included a total of 938 patients in this study, with a mean age of 3.1 ± 1.82 years. The mortality rate was 0.9%, and 93 (9.9%) patients had a prolonged ICU stay. An elevated ISS (34 ± 19.9 vs. 5 ± 5.1, p=0.004), lower GCS (8 ± 5.0 vs. 15 ± 1.3, p=0.006), and lower RTS (5.58 ± 1.498 vs. 7.64 ± 0.640, p=0.006) were all associated with mortality. All three scores were considered to be independent risk factors of mortality and prolonged ICU stay and had a linear correlation with increased HLOS. With regard to predicting mortality, ISS has the highest AUC value (ISS: 0.975; GCS: 0.864; and RTS: 0.899). The prediction cutoff values of ISS, GCS, and RTS on mortality were 15, 11, and 7, respectively. CONCLUSION: Regarding traumatic injuries in young children, worse ISS, GCS, and RTS were all associated with increased mortality, prolonged ICU stay, and longer hospital LOS. Of these scoring systems, ISS was the best at predicting mortality.