FCA-guided method creates clear counterfactuals for breast cancer diagnosis
FCA-Guided Counterfactual Explanations for Multi-Modal Breast Cancer Diagnosis: A Framework Achieving Perfect Validity with Emergent Sparsity
Artificial Intelligence
Summary
Breast cancer diagnosis models use many types of data but can be hard to explain in ways doctors trust. The authors show a new method that guides explanations with a structured concept map, ensuring all suggested changes actually switch the diagnosis while changing the fewest possible factors. Compared to other explanation tools, this method delivered perfectly valid and simpler alternative diagnosis scenarios. This helps make AI decisions more understandable and useful in clinical settings.
What this means in practice
- •For clinical ai developers: Generate clinically meaningful alternative diagnosis cases that highlight minimal necessary data changes to flip breast cancer model decisions.
- •For medical data scientists: Enhance interpretability of multi-modal cancer diagnostic models by incorporating formal concept analysis to guide counterfactual explanations.
Authors
Abdullahi Isa, Souley Boukari, Muhammad Aliyu
Abstract
Deep learning models for multi-modal breast cancer diagnosis achieve high predictive accuracy but remain clinically unacceptable without actionable, counterfactual explanations. Attribution-based methods (LIME, SHAP) are categorically inapplicable to this purpose, as they generate no alternative instances and thus cannot be evaluated on counterfactual quality metrics. This investigation provides empirical evidence that FCA-Guided Counterfactual (FCA-CF) framework that uses a Formal Concept Analysis (FCA) concept lattice as a hard structural constraint on counterfactual search, operating over a multi-modal TCGA-BRCA dataset. We benchmark against four genuine counterfactual methods: Wachter-style CF, DiCE, FACE, and NICE, evaluated on 60 benign-predicted TCGA-BRCA instances. The FCA-CF framework achieves Validity = 1.0000 (100% of counterfactuals successfully flip the prediction), Sparsity = 2.37 features changed (best among all valid methods), and Proximity = 0.900 (normalised L2-based, matching NICE as joint best). The classifier achieves Accuracy = 0.980, F1 = 0.976, ROC-AUC = 0.9947. Ablation analysis confirms that the FCA lattice constraint is the primary sparsity driver (removing it increases sparsity by +40%, p < 0.001, Cohen's d = 0.78), while Phase C greedy refinement accounts for the largest individual contribution (+113% sparsity increase when disabled, p < 0.001, d = 5.01). FCA-guided counterfactual generation achieves a clinically important Pareto-dominant outcome; it is simultaneously the sparsest and among the most proximate of all valid methods, with perfect validity. The emergent sparsity property arising from lattice topology rather than numerical penalty terms constitutes a structurally novel contribution to the counterfactual explanation literature.