Predictive Modeling for Breast Cancer Classification in the Context of Bangladeshi Patients: A Supervised Machine Learning Approach with Explainable AI

Machine learning can significantly improve breast cancer diagnosis speed: The study demonstrated that XGBoost achieved 97% accuracy in classifying breast cancer cases from 500 Bangladeshi patients, suggesting that AI-assisted diagnosis could help reduce the time-consuming manual diagnostic process and enable earlier detection.

Explainability is essential for clinical trust: The researchers used SHAP analysis to interpret how the XGBoost model made its predictions and which patient features most influenced its decisions. This transparency is critical for doctors to understand and trust AI recommendations before using them in real clinical settings.

Different algorithms perform differently on the same task: The study compared five machine learning methods (decision tree, random forest, logistic regression, naive Bayes, and XGBoost), finding that XGBoost outperformed others. This highlights that choosing the right algorithm matters and that one-size-fits-all approaches to AI in healthcare may not be optimal.

Population-specific datasets improve relevance: By using data from 500 patients at Dhaka Medical College Hospital rather than relying solely on international datasets, the researchers created a model potentially better suited to Bangladeshi patients' characteristics, addressing the important issue of AI bias across different populations and healthcare contexts.

Multiple performance metrics matter beyond accuracy alone: The study evaluated precision, recall, and F-1 scores alongside accuracy, recognizing that in cancer diagnosis, different metrics capture different types of errors—some measures may be more important than others depending on whether false negatives or false positives pose greater clinical risks.