Dataset Information

Derivation and validation of different machine-learning models in mortality prediction of trauma in motorcycle riders: a cross-sectional retrospective study in southern Taiwan.

ABSTRACT: OBJECTIVES:This study aimed to build and test the models of machine learning (ML) to predict the mortality of hospitalised motorcycle riders. SETTING:The study was conducted in a level-1 trauma centre in southern Taiwan. PARTICIPANTS:Motorcycle riders who were hospitalised between January 2009 and December 2015 were classified into a training set (n=6306) and test set (n=946). Using the demographic information, injury characteristics and laboratory data of patients, logistic regression (LR), support vector machine (SVM) and decision tree (DT) analyses were performed to determine the mortality of individual motorcycle riders, under different conditions, using all samples or reduced samples, as well as all variables or selected features in the algorithm. PRIMARY AND SECONDARY OUTCOME MEASURES:The predictive performance of the model was evaluated based on accuracy, sensitivity, specificity and geometric mean, and an analysis of the area under the receiver operating characteristic curves of the two different models was carried out. RESULTS:In the training set, both LR and SVM had a significantly higher area under the receiver operating characteristic curve (AUC) than DT. No significant difference was observed in the AUC of LR and SVM, regardless of whether all samples or reduced samples and whether all variables or selected features were used. In the test set, the performance of the SVM model for all samples with selected features was better than that of all other models, with an accuracy of 98.73%, sensitivity of 86.96%, specificity of 99.02%, geometric mean of 92.79% and AUC of 0.9517, in mortality prediction. CONCLUSION:ML can provide a feasible level of accuracy in predicting the mortality of motorcycle riders. Integration of the ML model, particularly the SVM algorithm in the trauma system, may help identify high-risk patients and, therefore, guide appropriate interventions by the clinical staff.

SUBMITTER: Kuo PJ

PROVIDER: S-EPMC5781097 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundMotorcycles are one of the most commonly used transportation modes in low and middle-income countries. In India, motorized two-wheelers comprise 70% of the total vehicle population, and motorcycle users are considered the most vulnerable road users. It is essential to understand the risky riding behaviour and associated factors among the motorcyclists to develop evidence-based traffic safety programs targeting motorcycle riders. The purpose of the current study was two-fold. First, it aimed to determine the appropriate structure of a modified version of the MRBQ among young riders in Manipal, India. Second, it assessed to what extent MRBQ factors were associated with self-reported crash involvement and violations.MethodsThe motorcycle rider behaviour questionnaire (MRBQ) is a 43-item scale that assesses five aspects of risky motorcycle rider behaviour, i.e., violations, control errors, traffic errors, stunts, and protective equipment. The MRBQ, along with measures of socio-demographic variables and the number of motorcycle crashes, was filled out by 300 young motorcycle riders who were in the age group of 18-25 years and had been riding for at least the past three years (93% males, 92.3% students).ResultsFive factors emerged out of the MRBQ after an exploratory factor analysis: traffic errors, control errors, stunts, protective equipment, and violations. Cronbach's alpha for these factors ranged from .66 to .82. Reports of performing stunts and committing violations were positively associated with self-reported near-crash experiences over the past three months. Riders reporting stunts, violations and using a motorcycle of 125-200 cc reported having received more fines in the last three months. These findings were confirmed in both univariate and multivariate binary logistic regression models.ConclusionThe study assessed the factor structure of a modified version MRBQ and the extracted factors associations with self-reported crash involvement. The factor structure revealed in the current study is consistent with MRBQ factor structures found in other countries. However, the support for a relationship between MRBQ factors and self-reported crashes was less significant. The findings suggest that if replicated by future studies, local policymakers are advised to focus on the five MRBQ factors while planning future interventions to achieve a reduction in the number of road crashes among motorcyclists.

Project description:Purpose: This study aimed to determine the patterns associated with adult mandibular fractures from a Level-I trauma center in southern Taiwan. Methods: The data of adult trauma patients admitted between 1 January 2009 and 31 December 2014 were retrieved from the Trauma Registry System and retrospectively reviewed. Fracture site and cause of injury were categorized into groups for comparison, and corresponding odds ratios (ORs) and 95% confidence intervals (CIs) were obtained by multivariate logistic regression. Results: Motorcycle accidents were the most common cause of mandibular fractures (76.3%), followed by falls (10.9%), motor vehicle accidents (4.8%), and being struck by/against objects (4.5%). Of the 503 cases of mandibular fractures, the condylar neck and head were the most common sites (32.0%), followed by the parasymphysis (21.7%), symphysis (19.5%), angle and ramus (17.5%), and body (9.3%). The location of mandibular fractures in patients who had motorcycle accidents was similar to that in all patients. Motor vehicle accidents resulted in a significantly higher number of body fractures (OR 3.3, 95% CI 1.24-8.76, p = 0.017) and struck injury in a significantly higher number of angle and ramus fractures (OR 3.9, 95% CI 1.48-10.26, p = 0.006) compared to motorcycle accidents. The helmet-wearing status and body weight were not associated with the location of mandibular fractures in motorcycle accidents. Conclusions: Our study revealed that the anatomic fracture sites of mandible were specifically related to different etiologies. In southern Taiwan, motorcycle accidents accounted for the major cause of mandibular fractures and were associated with the condylar neck and head as the most frequent fracture sites. In contrast, motor vehicle accidents and struck injuries tended to cause more body fracture as well as angle and ramus fracture compared to motorcycle accidents. Furthermore, the status of helmet-wearing and body weight were not associated with the location of mandible fractures caused by motorcycle accidents.

Project description:Background: This study aimed to profile the epidemiology of injury among preschool-aged and school-aged children in comparison to those in adults. Methods: According to the Trauma Registry System of a level I trauma center, the medical data were retrieved from 938 preschool-aged children (aged less than seven years), 670 school-aged children (aged 7⁻12 years), and 16,800 adults (aged 20⁻64 years) between 1 January 2009 and 31 December 2016. Two-sided Pearson’s, chi-squared, and Fisher’s exact tests were used to compare categorical data. A one-way analysis of variance (ANOVA) with the Games-Howell post-hoc test was used to assess the differences in continuous variables among different groups of patients. The mortality outcomes of different subgroups were assessed by a multivariable regression model under the adjustment of sex, injury mechanisms, and injury severity. Results: InFsupppjury mechanisms in preschool-aged and school-aged children were remarkably different from that in adults; in preschool-aged children, burns were the most common cause of injury requiring hospitalization (37.4%), followed by falls (35.1%) and being struck by/against objects (11.6%). In school-aged children, injuries were most commonly sustained from falls (47.8%), followed by bicycle accidents (14%) and being struck by/against objects (12.5%). Compared to adults, there was no significant difference of the adjusted mortality of the preschool-aged children (AOR = 0.9; 95% CI 0.38⁻2.12; p = 0.792) but there were lower adjusted odds of mortality of the school-aged children (AOR = 0.4; 95% CI 0.10⁻0.85; p = 0.039). The school-aged children had lower odds of mortality than adults (OR, 0.2; 95% CI, 0.06⁻0.74; p = 0.012), but such lower odds of risk of mortality were not found in preschool-aged children (OR, 0.7; 95% CI, 0.29⁻1.81; p = 0.646). Conclusions: This study suggests that specific types of injuries from different injury mechanisms are predominant among preschool-aged and school-aged children. The school-aged children had lower odds of mortality than adults; nonetheless there was no difference in mortality rates of preschool-aged children than adults, with or without controlling for sex, injury mechanisms and ISS. These results highlight the importance of injury prevention, particularly for preschool-aged children in Southern Taiwan.

Project description:OBJECTIVES:Current mortality prediction models used in the intensive care unit (ICU) have a limited role for specific diseases such as influenza, and we aimed to establish an explainable machine learning (ML) model for predicting mortality in critically ill influenza patients using a real-world severe influenza data set. STUDY DESIGN:A cross-sectional retrospective multicentre study in Taiwan SETTING: Eight medical centres in Taiwan. PARTICIPANTS:A total of 336 patients requiring ICU-admission for virology-proven influenza at eight hospitals during an influenza epidemic between October 2015 and March 2016. PRIMARY AND SECONDARY OUTCOME MEASURES:We employed extreme gradient boosting (XGBoost) to establish the prediction model, compared the performance with logistic regression (LR) and random forest (RF), demonstrated the feature importance categorised by clinical domains, and used SHapley Additive exPlanations (SHAP) for visualised interpretation. RESULTS:The data set contained 76 features of the 336 patients with severe influenza. The severity was apparently high, as shown by the high Acute Physiology and Chronic Health Evaluation II score (22, 17 to 29) and pneumonia severity index score (118, 88 to 151). XGBoost model (area under the curve (AUC): 0.842; 95% CI 0.749 to 0.928) outperformed RF (AUC: 0.809; 95% CI 0.629 to 0.891) and LR (AUC: 0.701; 95% CI 0.573 to 0.825) for predicting 30-day mortality. To give clinicians an intuitive understanding of feature exploitation, we stratified features by the clinical domain. The cumulative feature importance in the fluid balance domain, ventilation domain, laboratory data domain, demographic and symptom domain, management domain and severity score domain was 0.253, 0.113, 0.177, 0.140, 0.152 and 0.165, respectively. We further used SHAP plots to illustrate associations between features and 30-day mortality in critically ill influenza patients. CONCLUSIONS:We used a real-world data set and applied an ML approach, mainly XGBoost, to establish a practical and explainable mortality prediction model in critically ill influenza patients.

Dataset Information

Derivation and validation of different machine-learning models in mortality prediction of trauma in motorcycle riders: a cross-sectional retrospective study in southern Taiwan.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets