To evaluate the predictive performance of multiclass ensemble models specifically for diabetes-related high health care usage and assess their economic impact.
Key Findings:
Boosted tree models achieved the highest performance with multiclass area under the receiver operating curve scores of 0.6877 (95% CI 0.6927-0.7255) for LOS and 0.7601 (95% CI 0.7301-0.7654) for ED visits.
Interpretation:
Ensemble models effectively predict multilevel health care usage, indicating their utility in guiding targeted interventions and resource allocation.
Limitations:
Potential overfitting due to exclusion of the 2019-2020 dataset, which may limit the generalizability of the findings.
Conclusion:
Ensemble models can support targeted interventions in diabetes-related health programs and may lead to significant cost savings, highlighting their importance in diabetes management.
