PSO-XAI: A PSO-Enhanced Explainable AI Framework for Reliable Breast Cancer Detection

TL;DR

The paper tackles reliable breast cancer detection by integrating Particle Swarm Optimization for feature selection with explainable AI (SHAP) explanations across 29 diverse classifiers on the WDBC dataset. It presents a comprehensive methodology including data cleaning, multi-objective PSO optimization, and a rigorous cross-validation framework to ensure generalization and interpretability. Key contributions include a balanced PSO fitness design, adaptive parameter control, substantial performance gains over baselines, and model-agnostic explanations that align with clinical reasoning. The findings demonstrate high predictive accuracy with strong interpretability, offering a practical and trustworthy framework for clinical decision support, with future work extending to multi-modal data and real-world validation.

Abstract

Breast cancer is considered the most critical and frequently diagnosed cancer in women worldwide, leading to an increase in cancer-related mortality. Early and accurate detection is crucial as it can help mitigate possible threats while improving survival rates. In terms of prediction, conventional diagnostic methods are often limited by variability, cost, and, most importantly, risk of misdiagnosis. To address these challenges, machine learning (ML) has emerged as a powerful tool for computer-aided diagnosis, with feature selection playing a vital role in improving model performance and interpretability. This research study proposes an integrated framework that incorporates customized Particle Swarm Optimization (PSO) for feature selection. This framework has been evaluated on a comprehensive set of 29 different models, spanning classical classifiers, ensemble techniques, neural networks, probabilistic algorithms, and instance-based algorithms. To ensure interpretability and clinical relevance, the study uses cross-validation in conjunction with explainable AI methods. Experimental evaluation showed that the proposed approach achieved a superior score of 99.1\% across all performance metrics, including accuracy and precision, while effectively reducing dimensionality and providing transparent, model-agnostic explanations. The results highlight the potential of combining swarm intelligence with explainable ML for robust, trustworthy, and clinically meaningful breast cancer diagnosis.

Figures (6)

• Figure 1: Overall Workflow of the study
• Figure 2: Boxplots of selected breast cancer features highlighting the presence and proportion of outliers across distributions.
• Figure 3: Class distribution of the breast cancer dataset. The bar chart shows the absolute counts of benign and malignant samples.
• Figure 4: Confusion Matrix evaluation of top performing models
• Figure 5: Performance evaluation of top 5 machine learning models using ROC curves (left) and Precision-Recall curves (right). The ROC curves illustrate the trade-off between true positive rate and false positive rate, obtained through 10-fold cross-validation, while the Precision-Recall curves display the precision-recall performance. All models achieve excellent performance with AUC scores ranging from 0.985 to 0.994 and Average Precision (AP) scores from 0.980 to 0.992. Models evaluated include Multi-Layer Perceptron, Logistic Regression, Linear SVC, K-Nearest Neighbors, and Support Vector Classifier.