Comparative Evaluation of Machine Learning and Specialist Physicians in Breast Care Triaging: A Real-World Observational Study

Abstract

Objective: To evaluate the diagnostic accuracy and efficiency of a proprietary breast-specific machine learning (ML) model-built upon the open-source Open Triage platform-in comparison to specialist physicians, using standardized real-world clinical data for breast referral triaging. Materials and Methods: A retrospective observational study was conducted using 174 standardized breast cases obtained from proprietary industry datasets, spanning 46 disease types, 23 of which were cancers. The cohort ranged from 19 to 75 years (mean: 39.4±12.0). Physicians and an ML model each generated three diagnostic predictions per case. Both modalities were compared after benchmarking their predictions against a gold-standard diagnosis established by imaging and biopsy. Performance was evaluated using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operating characteristic (ROC) analysis. Time efficiency was also assessed to compare diagnostic turnaround times between physician- and ML-generated predictions. Results: 0.442). Both groups achieved comparable specificity and NPV values. ROC analysis showed an AUC of 0.91 for the ML model's first prediction versus 0.83 for the doctor's first prediction, indicating superior predictive power of the ML model. Conclusion: The ML model demonstrated diagnostic accuracy comparable to or better than that of physicians while significantly reducing the time required. These findings suggest that AI-powered triage tools could enhance the efficiency and standardization of breast triage.

Ähnliche Arbeiten