Ensemble Machine Learning Model Incorporating Radiomics and Body Composition for Predicting Intraoperative Hemodynamic Instability (HDI) in PPGL
Authors' conclusion
The EL model combining preoperative computed tomography-based body composition, tumor radiomics, and clinical data could potentially help predict intraoperative HDI in patients with PPGL.
Study context
- Population
- Patients with PPGL confirmed by surgical pathology; HDI vs hemodynamic stability (HDS) by intraoperative systolic blood pressure; training cohort, validation cohortHDI 101 · vs HDS 98 · training cohort 140 · validation cohort 59
- Participants
- 199 consecutive patients
- Exposure
- Ensemble machine learning model vs multivariate logistic regression model
- Outcome
- Discrimination of intraoperative HDI vs HDS (AUC)
Methodological detailsprediction model development and validation, consecutive patients
- Prediction model development and validation
- Consecutive patients
Main finding
As reported in the abstract
AUC for intraoperative HDI
| Group | AUC, training cohort | AUC, validation cohort |
|---|---|---|
| EL model | 96.2% (95% CI, 93.5%-99.0%) | 93.7% (95% CI, 88.0%-99.4%) |
| Logistic regression model | 74.4% (95% CI, 66.1%-82.6%) | 74.2% (95% CI, 61.1%-87.3%) |
AUC, training cohort
EL model96.2% (95% CI, 93.5%-99.0%)
Logistic regression model74.4% (95% CI, 66.1%-82.6%)
AUC, validation cohort
EL model93.7% (95% CI, 88.0%-99.4%)
Logistic regression model74.2% (95% CI, 61.1%-87.3%)
Observational data: associations, not effects.
Further findings
- Favorable calibration performance and clinical applicability of the EL model
Original publication
Ensemble Machine Learning Model Incorporating Radiomics and Body Composition for Predicting Intraoperative HDI in PPGL. J Clin Endocrinol Metab. Published 14 September 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →