Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 02026 | |
Number of page(s) | 7 | |
Section | Machine Learning in Healthcare and Finance | |
DOI | https://doi.org/10.1051/itmconf/20257002026 | |
Published online | 23 January 2025 |
Application of federated learning in predicting breast cancer
Intelligence Science and Technology, Sun Yat-sen University, 518000 Shenzhen, China
Corresponding author: chaijr@mail2.sysu.edu.cn
The prediction and diagnosis of breast cancer relies on multimodal data, such as imaging, genetic information, and patient lifestyle habits. Federated learning provides a framework to protect data privacy, allowing multiple institutions to share model training without sharing the original data. This paper proposes a breast cancer prediction model combined with federated learning, where each participant trains the model locally using multimodal data such as imaging, genes, and treatment history. During the local training process, the data is normalized and feature extracted, initially classified using support vector machines (SVM) or penalized logistic regression and optimized using stochastic gradient descent (SGD). Subsequently, each participant then sends the updated model parameters to the central server, where the FedAvg algorithm combines them to produce a global model. The model achieves data protection, also accurately predicts the progression and recurrence risk of breast cancer. Although federated learning effectively solves the privacy protection problem, the issues of data heterogeneity and model interpretability still need to be addressed. In the future, interpretability technologies (such as SHAP and LIME) and transfer learning can be combined to improve the transparency and adaptability of the model.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.