📄 other
Comparison of parsimonious and deep learning models for predicting surgical resource utilization in total hip arthroplasty: a retrospective cohort study
This retrospective cohort study of over 300,000 total hip arthroplasty cases found that while deep learning models outperformed simple statistical approaches in predicting surgical duration and length of stay with lower mean absolute error, the resulting gains in clinically relevant operational accuracy were minimal, suggesting that patient-level data alone may have reached a practical ceiling for improving resource utilization in standardized procedures.