Development and benchmarking of machine learning models to classify patients suitable for outpatient lower extremity joint arthroplasty
Authors' conclusion
Machine learning models may utilize electronic health records to screen arthroplasty procedures for outpatient eligibility. Tree-based models demonstrated superior performance in this study.
Study context
- Population
- Knee and hip arthroplasty patients, single institution, October 2013 to November 20215,523 records kept
- Participants
- 7,322 participants
- Comparison
- Machine learning classifiers vs baseline
- Outcome
- F1-score, ROC AUC and area under the precision-recall curve for predicting same-day discharge
Methodological details10-fold stratified cross-validation, single institution
- 10-fold stratified cross-validation
- Single institution
Performance
As reported in the abstract
Classification of same-day discharge
F1-score improvement
0.174
Balanced random forest vs Baseline
F1-score improvement
0.031
Balanced random forest vs Logistic regression
Balanced random forest
- F1-score
- 0.347
- AUC
- 0.734
Balanced random forest
F1-score0.347
AUC0.734
Further findings
Feature importance (SHAP)
- Top determinant features: patient sex, surgical approach, surgery type, BMI
Original publication
Development and benchmarking of machine learning models to classify patients suitable for outpatient lower extremity joint arthroplasty. J Clin Anesth. Published 16 May 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →