Explainable artificial intelligence in breast cancer detection and risk prediction: A systematic scoping review

SHAP is the dominant explainability tool in breast cancer AI research: Among 30 peer-reviewed studies examining explainable AI in breast cancer detection and risk prediction (2017-2023), SHAP (SHapley Additive exPlanations) emerged as the most widely used technique for making AI decisions transparent to clinicians.

Black-box AI algorithms create trust barriers in clinical settings: The review identifies a critical problem—complex AI algorithms used in medical diagnosis operate as "black boxes" that clinicians cannot understand or trust, which XAI methods are designed to address by revealing how these systems reach their conclusions.

Explainability enables clinicians to validate AI reasoning: By explaining which biomarkers and features drive AI predictions for diagnosis, classification, and survival analysis, XAI techniques help clinicians verify whether the AI is using medically sound logic rather than spurious patterns.

Model-agnostic approaches offer practical flexibility: SHAP's popularity stems from its ability to explain any type of AI model and its relative ease of implementation, making it a practical solution for healthcare institutions adopting AI without requiring specialized technical expertise.

Explainability is essential for responsible AI deployment in healthcare: The systematic focus on XAI methods in breast cancer research reflects growing recognition that trustworthy, understandable AI—not just accurate AI—is necessary for clinical adoption and patient safety.