Explainable machine learning-based mortality risk stratification for older adults with COVID-19: pinpointing core immunological biomarkers and revealing dose-threshold effects
-
By
-
Lin Luo
-
Lin Wang
-
Hao Wang
-
Hui Li
-
Ting Liu
-
Sha Yu
-
May 25, 2026
-
Clinical Scorecard: Machine Learning Approaches for Mortality Risk Assessment in Elderly COVID-19 Patients: Identifying Key Immunological Biomarkers and Dose-Response Relationships
At a Glance
| Category | Detail |
|---|
| Condition | COVID-19 mortality risk in elderly patients |
| Key Mechanisms | Utilization of routine hematological indicators for risk prediction |
| Target Population | Elderly COVID-19 patients |
| Care Setting | Clinical settings requiring early risk stratification |
Key Highlights
- 2393 COVID-19 patients were analyzed to develop a machine learning model.
- The LGBM model achieved an AUC of 0.973 and a recall of 0.924.
- Top 10 key features for mortality risk included CRP, D-dimer, and age.
- Non-linear associations observed with CRP and D-dimer levels.
- A simplified model reduced training time by 58.31% without compromising performance.
Guideline-Based Recommendations
Diagnosis
- Utilize routine hematological indicators for initial assessment.
Management
- Implement machine learning models for early risk stratification.
Monitoring & Follow-up
- Regularly assess key biomarkers such as CRP and D-dimer.
Risks
- Older adults are at significantly higher risk of COVID-19 mortality.
Patient & Prescribing Data
Elderly patients with COVID-19
Focus on monitoring hematological indicators to guide treatment decisions.
Clinical Best Practices
- Incorporate machine learning models in clinical practice for risk assessment.
- Prioritize resource allocation based on mortality risk predictions.
Related Resources & Content
