| Popis: |
Traumatic brain injury (TBI) is defined as blunt and penetrating injury to the head and/or brain caused by an external force that leads to temporary or permanent impairments to the brain function. Accurate measurement of prediction for the outcomes of affected individual is highly desirable to plan and optimize treatment decision. The clinical experts predict the outcomes of brain injury patients with a high degree of accuracy based on their experience and the standardized Glasgow Outcome Scale (GOS). The GOS has been used over the past 40 years and it plays an important role in developing the understanding of brain injury. Recent developments in Artificial Intelligence (AI) have heightened the need for developing predictive models using machine learning (ML) methods especially for TBI patients who require life-saving interventions. ML is a subfield of AI which allows the computer algorithms to learn patterns by studying data directly without being explicitly programmed. This paper compares the different ways in which predictive models evaluate the potential of ML for TBI outcome prediction. A literature survey of latest articles from 2016 to 2018 reveals that the predictions of existing predictive models compute different prediction performances in terms of accuracy, sensitivity, specificity and area under receiving operator characteristic (ROC) curve (AUC). Depending on the specific prediction task evaluated and the type of input features included, Artificial Neural Network (ANN) creates a powerful model to predict outcomes of TBI with profound accuracy compared to other ML models. Although ANNs are considered as “black-box” in computational models, their benefits in clinical medicine have infinite potentials in evidence-based medicine practice because ANNs can be trained on new patient information. Moreover, the existing predictive models show that ML can be leveraged to more accurately predict the outcomes of TBI patients. Most importantly, predictive models can provide real-time clinical utilization that leads to greater accuracy and higher predictive value for patients suffered from traumatic brain injury. |
