Incorporating Environmental and Social Factors Improves Machine Learning Predictions for Pneumonia Readmissions
-
By
-
Jack A. Cummins
-
Feifan Liu
-
June 15, 2026
Clinical Scorecard: Incorporating Environmental and Social Factors Improves Machine Learning Predictions for Pneumonia Readmissions
At a Glance
| Category | Detail |
|---|
| Condition | Pneumonia Readmissions |
| Key Mechanisms | Incorporation of residential greenness (NDVI) into predictive models |
| Target Population | Patients with pneumonia at risk of readmission |
| Care Setting | Clinical settings utilizing electronic health records and machine learning |
Key Highlights
- Integration of NDVI improves predictive models for pneumonia readmissions.
- Study utilized a cohort of 22,600 patients with rigorous feature selection.
- NDVI serves as a proxy for social determinants of health.
- Potential for NDVI to reduce algorithmic performance disparities across demographics.
- Future research suggested to assess NDVI's impact on marginalized cohorts.
Guideline-Based Recommendations
Diagnosis
Management
Monitoring & Follow-up
Risks
Patient & Prescribing Data
Patients with pneumonia at risk of readmission
NDVI may enhance predictive accuracy for readmissions.
Clinical Best Practices
- Consider environmental factors like NDVI in predictive modeling.
- Utilize ablation analysis to assess the impact of features in machine learning models.
- Evaluate predictive performance across sociodemographic subgroups.
Related Resources & Content
