To analyze the safety of stem-cell mobilization (SCM) in multiple myeloma patients and develop machine learning models to predict adverse events (AEs) and their onset timing.
Key Findings:
97% of patients achieved successful stem cell collection.
69% experienced severe AEs necessitating hospitalization.
Risk-stratified outpatient protocols could reduce bed usage by at least one third.
Classification models accurately predicted some AE types (e.g., elevated creatinine, ROC-AUC 1.0), but neutropenic fever prediction was challenging (ROC-AUC 0.67).
Regression models forecasted AE onset with a mean error of just over one day.
Interpretation:
The study outlines a data-driven approach for safely implementing outpatient SCM, optimizing resource allocation in clinical practice.
Limitations:
The datasets analyzed are not publicly available due to privacy and ethical restrictions.
Machine learning models may not generalize to all patient populations.
Conclusion:
Outpatient SCM can be safely adopted with proper risk assessment and management strategies.
