Machine learning model predicts in-hospital complications after traumatic

People who are admitted to the hospital with a sudden traumatic brain injury can face another danger: complications that develop during their hospital stay. These problems can slow recovery and lead to worse brain-related outcomes. A new study published in Brain Informatics reports that a simple model using four pieces of information collected at hospital admission can predict which patients are at high risk of these complications. The logistic regression model had an area under the curve (AUC) of 0.952, with a 95% confidence interval (CI) of 0.905–0.992.

The study, led by Mei-Hua Wang and colleagues, looked back at the medical records of 186 adults with acute traumatic brain injury who were admitted to a Neurocritical Care Unit. The researchers aimed to develop and test a practical model that could predict complications during the hospital stay using information available when patients were admitted.

Of the 186 patients in the study, 136, or 73.12%, developed complications during their hospital stay. The authors note that this high rate shows how often acute traumatic brain injury is accompanied by problems that can lengthen hospital stays and lead to worse neurological outcomes.

To develop the prediction models, the researchers compared five methods: logistic regression, Lasso regression, random forest, support vector machine, and decision tree. They used these methods to identify possible predictors and build models to estimate the risk of complications. They then assessed how well each model distinguished between patients at high and low risk, how closely its predictions matched the actual results, and whether using it could provide meaningful benefits in clinical care.

Four factors measured or assessed at hospital admission were found to be independent predictors of complications during the hospital stay: HbA1c, the Glasgow Coma Scale (GCS) score, stress-induced hyperglycemia (SIH), and severe injuries outside the brain. Because these factors can be checked soon after a patient arrives, the model can help doctors assess risk early instead of relying on information collected over the following days.

According to the study results, the final model using these four factors was good at distinguishing patients who developed complications from those who did not. Its predictions also matched the observed outcomes well, and the model showed potential value for clinical care. Of the five methods tested, logistic regression performed best, with an AUC of 0.952 and a 95% confidence interval of 0.905 to 0.992. An AUC this high means the model was very good at separating patients who later developed complications from those who did not in the group studied.

The authors conclude that a simple, practical model using HbA1c, GCS score, SIH, and severe injuries outside the brain can effectively predict the risk of complications during the hospital stay for patients with acute traumatic brain injury (TBI). They suggest that the model could help doctors assess risk early and make clinical decisions.

However, the findings have several limitations. The study looked back at past medical records and was carried out at only one hospital, in a single Neurocritical Care Unit. The authors say that further prospective studies involving multiple hospitals are needed to test and improve the model. A retrospective study can identify links between factors and outcomes and show how well a model predicts results in the group studied. However, it cannot prove that the model will work just as well in other hospitals, healthcare systems, or patient groups. The reported AUC reflects the results in this particular dataset, and it remains unknown how well the model will perform when tested with independent data from other settings.

The study was published as open-access research in Brain Informatics on 18 September 2026. The version released is an accepted manuscript made available early so readers can access the peer-reviewed research sooner. It may undergo further editing before the final Version of Record is published.

Subject of Research: Predicting complications during the hospital stay after acute traumatic brain injury using machine learning models

Article Title: Early prediction of in-hospital complications following acute traumatic brain injury: a retrospective study using multiple machine learning models

Article References: Wang, M.-H., Li, R., Fang, J., Deng, X., Zhang, J., Guo, Y., Chen, L., Chen, L., Yu, C., Chen, Y., Wu, G., Hu, J., Yao, H., Yuan, Q., & Yang, L. (2026). Early prediction of in-hospital complications following acute traumatic brain injury: a retrospective study using multiple machine learning models. Brain Informatics. https://doi.org/10.1186/s40708-026-00334-w

Image Credits: AI-generated image

DOI: 10.1186/s40708-026-00334-w

Keywords: traumatic brain injury, complications during hospitalization, machine learning, logistic regression, Glasgow Coma Scale, HbA1c, stress-induced hyperglycemia, risk prediction, Neurocritical Care Unit, retrospective study, early prediction