Why the study?
The relative benefits of artificial neural networks compared with traditional methods in health services research remain unclear.
Do Artificial Neural Network models improve the prediction of high Medicare costs, 30-day readmissions, and preventable hospitalizations compared to logistic regression in Medicare beneficiaries aged 65 and older?
Population
254,748 Medicare beneficiaries aged 65 and older
Comparison
Logistic regression vs artificial neural network models
Design
Cross-sectional study
Key result
Logistic regression and ANN models showed comparable predictive performance for high Medicare costs, 30-day readmissions, and preventable hospitalizations among older adults.
Authors
Loading...
ANN offers no advantage over logistic regression in older Medicare patients; reinforces traditional models as sufficient for claims-based risk prediction.
Do Artificial Neural Network models improve the prediction of high Medicare costs, 30-day readmissions, and preventable hospitalizations compared to logistic regression in Medicare beneficiaries aged 65 and older?
Artificial neural networks and traditional logistic regression models demonstrate comparable performance in predicting healthcare utilization and costs among older adults using structured Medicare data.
Chen et al. (2026) studied this question. Logistic regression and ANN models showed comparable predictive performance for high Medicare costs, 30-day readmissions, and preventable hospitalizations among older adults.