Explainable machine learning-based preliminary screening for viral encephalitis by blood routine analysis
-
By
-
Bo Lv
-
Jie Pan
-
Aiming Shi
-
Dongxing Wang
-
June 19, 2026
-
Clinical Scorecard: Interpretable Machine Learning for Initial Screening of Viral Encephalitis Using Routine Blood Tests
At a Glance
| Category | Detail |
|---|
| Condition | Viral Encephalitis |
| Key Mechanisms | Machine learning model utilizing routine blood analysis for risk stratification. |
| Target Population | Patients with suspected viral encephalitis. |
| Care Setting | Emergency or primary care settings. |
Key Highlights
- XGBoost model achieved an AUC of 0.949 in training and 0.900 in testing.
- Serum albumin and white blood cell counts were significant predictors.
- Model transparency ensured through SHAP analysis.
Guideline-Based Recommendations
Diagnosis
- Diagnosis based on clinical manifestations and routine blood analysis.
Management
- Utilize machine learning models for preliminary risk assessment.
Monitoring & Follow-up
- Monitor serum albumin and white blood cell counts for VE prediction.
Risks
- Timely diagnosis is critical for effective management of VE.
Patient & Prescribing Data
Patients suspected of having viral encephalitis.
Routine blood analysis can serve as a rapid diagnostic tool.
Clinical Best Practices
- Integrate machine learning models into diagnostic workflows.
- Ensure model interpretability to foster clinician trust.
Related Resources & Content
