AI and Cancer Research ── Breast ── 2026-10-26
Breast cancer: treatment, trials and AI
- Abstract
Background: Accurate breast cancer mortality prediction is constrained by severe class imbalance and the sensitivity-specificity trade-off, which frequently induces high false-positive rates in clinical settings. To address this limitation, we propose a rigorously calibrated Static Stacking Ensemble framework.Methods: The model was developed from the METABRIC cohort (N=1816) using 5-fold stratified cross-validation, while the SEER dataset (N=4024) was reserved for independent external validation. The architecture combines out-of-fold (OOF) predictions from deep neural network, XGBoost, and random forest base learners. To replace the standard fixed decision threshold (default τ=0.5), we used a data-driven threshold based on Youden’s J statistic.Results: Baseline models had important limitations, including a specificity of 0.1997 in standalone deep learning and weight collapse in evolutionary ensembles. At the optimized decision threshold (τ=0.7378), the proposed model balanced sensitivity and specificity. Evaluated on the independent cohort, the model yielded a specificity of 0.6996, a sensitivity of 0.7298, an accuracy of 0.7212, and an F1 score of 0.7894. The framework achieved a ROC-AUC of 0.6925, an area under the precision-recall curve (PR-AUC) of 0.8914, and a Brier score of 0.1563.Conclusion: Combining OOF prediction stacking with an optimized decision threshold provides a calibrated clinical decision support system (CDSS). This framework minimizes prognostic false alarms and robustly stratifies high-risk patients across heterogeneous cohorts, significantly outperforming isolated baseline architectures.
Journal IF-equivalent: 0.5 (OpenAlex 2-year mean citedness, value as of 2026-10-25, retrieved 2026-10-26; not the official Clarivate IF)Reference: Abdollahi J, Gardaneh M. Calibrated Stacking Ensemble Framework for High-Fidelity Breast Cancer Mortality Prediction. Arch Breast Cancer. 2026 Oct 23;13(4):474-85. doi:10.32768/abc.7294618352-074.Checked: Abstract only - Abstract
BACKGROUND: Breast cancer is the most frequently diagnosed malignancy in women globally, accounting for approximately 2.30 million new cases and 666,103 deaths in 2022. Artificial intelligence technologies including machine learning, deep learning, radiomics, and natural language processing have been widely applied across the breast cancer care continuum. However, no systematic review restricted to primary studies with patient-level or image-level data has synthesized evidence spanning all clinical domains simultaneously. OBJECTIVES: To evaluate the performance, methodological quality, and clinical relevance of artificial intelligence primary studies across screening and diagnostic imaging, digital pathology, radiomics, perioperative management, prognostic prediction, and natural language processing in breast cancer. METHODS: PubMed/MEDLINE, Scopus, Web of Science, Embase, and the Cochrane Library were searched from January 2015 to May 2026. Eligibility was restricted to original primary research reporting patient-level or image-level data; secondary syntheses were explicitly excluded. Risk of bias was assessed using QUADAS-2 and PROBAST. Reporting followed PRISMA 2020 guidelines. RESULTS: Of 5,275 identified records, 17 primary studies met eligibility criteria. In screening mammography (seven studies; n = 7,540 to 144,231 examinations), deep learning achieved area under the receiver operating characteristic curve (AUC) values of 0.876 to 0.939; a randomised trial in 80,033 women demonstrated non-inferiority to double reading with a 44.3% reduction in radiologist workload. In digital pathology (five studies; 270 to 44,732 whole-slide images), deep learning achieved AUC 0.985 to 0.994, exceeding unassisted pathologist performance in lymph node metastasis detection. Magnetic resonance imaging radiomics predicted pathological complete response in HER2-positive breast cancer with AUC 0.80. No breast cancer-specific perioperative primary study met the eligibility criteria. Deep survival networks achieved concordance index 0.626 versus 0.612 for Cox regression in breast cancer cohorts. Natural language processing identified metastatic recurrence in breast cancer clinical records with sensitivity 83% and specificity 73%. Risk of bias was low to moderate for imaging and pathology studies and high for small sample radiomics studies. Density-stratified performance was reported by only two imaging studies, and receptor status prediction from haematoxylin and eosin staining had not been validated as a safe alternative to immunohistochemistry. CONCLUSIONS: Artificial intelligence demonstrated clinically meaningful performance across the five domains with eligible primary studies. Breast cancer-specific perioperative and intensive care primary studies were absent, representing the largest evidence gap. External validation, demographic diversity, and prospective clinical integration evidence remained the most consistently identified limitations. Prediction of molecular receptor status without immunohistochemistry requires rigorous prospective safety validation before it can complement standard testing.
Journal IF-equivalent: 0.0 (OpenAlex 2-year mean citedness, value as of 2026-10-25, retrieved 2026-10-26; not the official Clarivate IF)Reference: [No authors listed]. Artificial Intelligence in Breast Cancer Diagnosis, Treatment, and Perioperative Management: A Systematic Review of Current Applications and Future Perspectives. Clin Ter. [Epub ahead of print]. doi:10.7417/CT.2026.2172. PMID: 42922241.Checked: Abstract only