Advancing Breast Cancer Care Through Transparent AI and Federated Learning: Integrating Radiological, Histopathological, and Clinical Data for Diagnosis, Recurrence Prediction, and Survivorship

The integration of artificial intelligence (AI) in breast cancer care shows immense potential for improving patient outcomes. However, challenges persist in model transparency, data privacy, and holistic utilization of diverse medical data. This Research Topic aims to explore novel approaches to advancing breast cancer care through transparent AI techniques and federated learning, focusing on the comprehensive integration of radiological, histopathological, and clinical data.
We encourage research investigating AI models that combine multiple data modalities to enhance accuracy and transparency in breast cancer screening, diagnosis, and prognosis. Central to research are explainable AI (XAI) techniques to address AI transparency in healthcare. By making AI predictions interpretable, healthcare professionals understand and validate AI-generated diagnoses and recommendations. This transparency is crucial for building trust and facilitating integration into clinical workflows. XAI techniques also allow the identification and mitigation of potential biases, ensuring more equitable outcomes across diverse patient populations. The topic encompasses the entire spectrum of breast cancer care, from developing models for predicting cancer recurrence and optimizing screening protocols, to guiding survivorship care. By integrating clinical notes and patient histories, these models can generate personalized risk assessments and follow-up recommendations. While, the explainability ensures that both healthcare providers and patients can understand and trust AI-generated insights, fostering shared decision-making in long-term care planning.
Finally, to address data siloes and privacy concerns, we encourage research exploring federated and in general decentralized learning frameworks. This approach enables collaborative model training across multiple healthcare institutions without centralized data storage. By keeping patient data localized while allowing models to learn from diverse datasets, federated learning enhances model generalization and robustness, which is particularly beneficial for improving breast cancer detection and recurrence prediction across varied patient demographics and clinical settings.
Please note: Manuscripts consisting solely of bioinformatics, computational analysis, or predictions of public databases which are not accompanied by validation (independent clinical or patient cohort, or biological validation in vitro or in vivo, which are not based on public databases) are not suitable for publication in this journal.