Integrative Machine Learning Approach for Predicting Resistance to First-generation Receptor Ligands in Acromegaly
Authors' conclusion
The CatBoost-based calculator effectively predicts fgSRLs treatment response in acromegaly patients. Prospective validation is required before clinical implementation.
Study context
- Population
- Patients with acromegaly treated with first-generation SRLs (first-generation somatostatin receptor ligands (fgSRLs)), Mass General Brigham-affiliated hospitals, January 2010-July 2024
- Participants
- 111 patients
- Comparison
- fgSRLs-responsive vs fgSRLs-resistant
- fgSRLs-responsiven = 64reference
- fgSRLs-resistantn = 47
- Outcome
- Model performance for predicting fgSRLs resistance (AUROC)
Methodological detailsmachine learning, single-centre, 10 machine learning (ML) algorithms evaluated, SHAP analysis, web-based calculator
- Machine learning
- Single-centre
- 10 ML algorithms evaluated
- SHAP analysis
- Web-based calculator
Performance
As reported in the abstract
CatBoost model performance (optimal model)
- AUROC
- 0.896 (95% CI 0.751-0.990)
- Accuracy
- 82.4%
- Precision
- 86.7%
- Specificity
- 88.2%
- F1 score
- 81.2%
Further findings
Key predictors of fgSRLs resistance (SHAP)
- Pre-fgSRLs GH, Knosp grade, pre-fgSRLs IGF-1 index, T2-weighted MRI density and comorbidity burden
Calibration and clinical utility
- Brier score (excellent calibration)
- 0.131
- Clinical utility via decision curve analysis (DCA)
Original publication
Integrative Machine Learning Approach for Predicting Resistance to First-generation Receptor Ligands in Acromegaly. J Clin Endocrinol Metab. Published 26 June 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →