Leveraging Artificial Intelligence for effective breast cancer management in low-resource centres: a pilot project of a TNM staging machine

Leveraging artificial intelligence (AI) in breast cancer management offers opportunities to address a major public health challenge in low-resource settings. However, the costs of infrastructure, data acquisition, and large-scale modelling remain significant barriers to implementation. To demonstrate the feasibility of integrating AI into a clinical workflow in a low-resource centre, we piloted a hybrid, guideline-informed machine learning model that maps clinical features to breast cancer (BC) stage. We implemented a decision tree classifier (DTC) using open-source Python libraries to reconstruct deterministic mapping of the AJCC 8th Edition staging system from structured clinical features. The model was trained on a complete set of feature-to-stage combinations derived from the AJCC framework. Performance was evaluated using held-out data and retrospective anonymized clinical records, with emphasis on functional correctness and consistency of the feature-to-stage mapping rather than probabilistic generalization. Once the feature space was completely represented, the model consistently reconstructed the full deterministic AJCC staging logic, achieving perfect classification on the held-out and retrospective real-world datasets. This outcome reflects verification of guideline-concordant mapping behaviour, not evidence of unconstrained statistical learning or independent clinical validation. This pilot study demonstrates the feasibility of a hybrid, data-inferred approach for supporting AJCC BC staging in low-resource settings. By prioritizing functional correctness, clinical interpretability, and minimal infrastructure requirements, the proposed method offers a pragmatic foundation for scalable decision-support tools where conventional data-intensive AI solutions are not immediately practical. Independent clinical validation using prospectively collected data is required to establish real-world effectiveness.