Prognosis prediction in traumatic brain injury patients using machine learning algorithms

Abstract

Predicting treatment outcomes in traumatic brain injury (TBI) patients is challenging worldwide. The present study aimed to achieve the most accurate machine learning (ML) algorithms to predict the outcomes of TBI treatment by evaluating demographic features, laboratory data, imaging indices, and clinical features. We used data from 3347 patients admitted to a tertiary trauma centre in Iran from 2016 to 2021. After the exclusion of incomplete data, 1653 patients remained. We used ML algorithms such as random forest (RF) and decision tree (DT) with ten-fold cross-validation to develop the best prediction model. Our findings reveal that among different variables included in this study, the motor component of the Glasgow coma scale, the condition of pupils, and the condition of cisterns were the most reliable features for predicting in-hospital mortality, while the patients’ age takes the place of cisterns condition when considering the long-term survival of TBI patients. Also, we found that the RF algorithm is the best model to predict the short-term mortality of TBI patients. However, the generalized linear model (GLM) algorithm showed the best performance (with an accuracy rate of 82.03 ± 2.34) in predicting the long-term survival of patients. Our results showed that using appropriate markers and with further development, ML has the potential to predict TBI patients’ survival in the short- and long-term.

Predictive value of the Trauma Rating Index in Age, Glasgow Coma Scale, Respiratory rate and Systolic blood pressure score (TRIAGES) for the short-term mortality of older patients with isolated traumatic brain injury: a retrospective cohort study

“… 11 12 Recent studies employing machine learning-based prognostic models for TBI identify GCS motor scores as reliable predictive indicators. 13 In our prior study, we also observed that the GCS exhibited good predictive power, with a slightly higher AUROC than TRIAGES. This finding emphasises the essential role of the GCS in the TRIAGES scoring system.…”

“…Moreover, modified versions of the GCS, like the GCS-P, exhibit enhanced predictive accuracy for TBI prognosis 11 12. Recent studies employing machine learning-based prognostic models for TBI identify GCS motor scores as reliable predictive indicators 13. In our prior study, we also observed that the GCS exhibited good predictive power, with a slightly higher AUROC than TRIAGES.…”

“…A parallel finding was observed in the study of the Base deficit, International normalized ratio, and Glasgow Coma Scale (BIG) scores, where integrating an age-adjusted BIG model significantly improved the predictive accuracy for TBI prognosis 17. In a study on machine learning of prediction models for patients with TBI, age was similarly considered a significant predictor 13. This phenomenon could be attributed to the more significant brain atrophy and intracranial space observed in older individuals, rendering them more vulnerable to intracranial haemorrhage 18.…”

“… 17 In a study on machine learning of prediction models for patients with TBI, age was similarly considered a significant predictor. 13 This phenomenon could be attributed to the more significant brain atrophy and intracranial space observed in older individuals, rendering them more vulnerable to intracranial haemorrhage. 18 Interestingly, despite similar levels of intracranial injury severity, older patients often present with higher GCS.…”

Method to aid the diagnosis of prostate cancer using machine learning and clinical data

“…K samples that are closer to the test sample are chosen from the training dataset to classify a test sample. For classi cation tasks, the dominant label among the target labels of the K chosen training samples is chosen as the predicted label for the test sample [20].…”