Machine learning-based triage model for elderly traumatic brain injury patients in Chinese emergency department
-
By
-
Yanya Lin
-
Chengda Lin
-
Jianhui Chen
-
Shijun Chen
-
Jianhuang Huang
-
Jianxiong Hu
-
May 25, 2026
-
Clinical Scorecard: Triage Model Utilizing Machine Learning for Older Adults with Traumatic Brain Injury in Chinese Emergency Settings
At a Glance
| Category | Detail |
|---|
| Condition | Traumatic Brain Injury (TBI) in elderly patients |
| Key Mechanisms | Machine learning model (XGBoost) for predicting ICU triage disposition |
| Target Population | Elderly patients aged ≥60 years with TBI |
| Care Setting | Emergency departments |
Key Highlights
- Developed an XGBoost-based triage model using data from 413 elderly TBI patients.
- Model achieved an AUC of 0.93, with high recall (0.9288) and precision (0.9274).
- Key predictors included symptoms, CT hematoma density, and contusion severity.
- Decision curve analysis suggested higher theoretical net benefit compared to traditional strategies.
- Further prospective validation against patient-centered outcomes is required.
Guideline-Based Recommendations
Diagnosis
- Confirm diagnosis of TBI through clinical evaluation and head CT imaging.
Management
- Utilize machine learning models to support triage decisions in elderly TBI patients.
Monitoring & Follow-up
- Monitor for rapid deterioration in elderly patients despite initial mild symptoms.
Risks
- Traditional triage methods may lead to over- or under-treatment in elderly patients.
Patient & Prescribing Data
Elderly TBI patients aged ≥60 years.
Consider anticoagulant use and comorbidities in triage decisions.
Clinical Best Practices
- Implement machine learning models to enhance triage accuracy in emergency settings.
- Ensure timely assessment and triage decisions within 48 hours of ED admission.
Related Resources & Content
