The predictive probability of mortality in the presence of full-thickness burns

A comparative study of the predictive value of four models for death in patients with severe burns

OBJECTIVE

To assess the prognostic value of the Ryan score, Belgian Outcome of Burn Injury (BOBI) score,revised Baux (rBaux) score, and a new model (a Logit(P)-based scoring method created in 2020) for predicting mortality risk in patients with extremely severe burns and to conduct a comparative analysis.

METHODS

A retrospective analysis was conducted on 599 burn patients who met the inclusion criteria and were admitted to the burn unit of the First Affiliated Hospital of Nanchang University from 2017 to 2022. Relevant information was collected, and receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were plotted for each of the four models in assessing mortality in these burn patients using both age-stratified and unstratified forms. The ROC curve section was further compared with the area under the curve (AUC), optimal cutoff value, as well as its sensitivity and specificity. Additionally, the quality of the AUC was assessed using the Delong test.

RESULT

Among the patients who met the inclusion criteria, 532 were in the survival group and 67 in the death group. Irrespective of age stratification, the novel model exhibited superior performance with an AUC of 0.868 (95% CI: 0.838-0.894) among all four models predicting mortality risk in included patients, and also demonstrated better AUC quality than other models; the calibration curves showed that the accuracy of all four models was good; the DCA curves showed that the clinical utility of the novel model and rBuax score were better. In the comparison of four scoring models across different age groups, the new model demonstrated the largest AUC in both 0-19 years (0.954, 95% CI 0.914-0.979) and 20-59 years groups (0.838, 95% CI 0.793-0.877), while rBuax score exhibited the highest AUC in ≥ 60 years group (0.708, 95% CI of 0.602-0.800). The calibration curves showed that the four models exhibited greater accuracy within the age range of 20-59 years, while the DCA curves indicated that both the novel model and rBuax score scale displayed better prediction in both the 20-59 and ≥ 60 years groups.

CONCLUSIONS

All four models demonstrate accurate and effective prognostication for patients with severe burns. Both the novel model and rBaux score exhibit enhanced prediction utility. In terms of the model itself alone, the new model is not simpler than, for example, the rBaux score, and whether it can be applied clinicallyinvolves further study.

Machine learning models for early prediction of mortality risk in patients with burns: A single center experience

Mortality rate is considered as the most important outcome measure for assessing the severity of burn injury. A scale or model that accurately predicts burn mortality can be useful to determine the clinical course of burn injuries, discuss treatment options and rehabilitation with patients and their families, and evaluate novel, innovative interventions for the injuries. This study aimed to use machine learning models to predict the mortality risk of patients with burns after their first admission to the center and to compare the performances of these models. Overall, 1064 patients hospitalized in burn intensive care and burn service units between 2016 and 2022 were included in the study. In total, 40 parameters, including demographic characteristics and biochemical parameters of all patients, were analyzed in the study. Furthermore, the dataset was randomly divided into two clusters with 70% of the data used for artificial neural networks (ANNs) training and 30% for model success testing. The ANN model proposed in this study showed high success across all machine learning methods tried in different variants, with an accuracy of 95.92% in the test set. Machine learning models can be used to predict the mortality risk of patients with burns. This study may help validate the use of machine learning models for applications in clinical practice. Conducting multicenter studies will further contribute to the literature.