Application of transfer learning for biomedical signals: A comprehensive review of the last decade (2014–2024)
Article
Tao, Xiaohui, Jafari, Mahboobeh, Barua, Prabal, Tan, Ru-San and Acharya, U.Rajendra. 2025. "Application of transfer learning for biomedical signals: A comprehensive review of the last decade (2014–2024)." Information Fusion. 118. https://doi.org/10.1016/j.inffus.2025.102982
| Article Title | Application of transfer learning for biomedical signals: A comprehensive review of the last decade (2014–2024) |
|---|---|
| ERA Journal ID | 20983 |
| Article Category | Article |
| Authors | Tao, Xiaohui, Jafari, Mahboobeh, Barua, Prabal, Tan, Ru-San and Acharya, U.Rajendra |
| Journal Title | Information Fusion |
| Journal Citation | 118 |
| Article Number | 102982 |
| Number of Pages | 38 |
| Year | 2025 |
| Publisher | Elsevier |
| Place of Publication | Netherlands |
| ISSN | 1566-2535 |
| 1872-6305 | |
| Digital Object Identifier (DOI) | https://doi.org/10.1016/j.inffus.2025.102982 |
| Web Address (URL) | https://www.sciencedirect.com/science/article/pii/S1566253525000557 |
| Abstract | Precise and timely disease diagnosis is essential for making effective treatment decisions and halting disease progression. Biomedical signals offer the potential for non-invasive diagnosis of diverse conditions, enhancing the ability to predict clinical outcomes and plan treatments more effectively. These signals have garnered significant attention, particularly in +conjunction with artificial intelligence (AI)-powered models, such as conventional machine learning (ML) and deep learning (DL), demonstrating promising outcomes. However, DL models, which have become the de facto standard in medical data analysis, encounter challenges such as inadequate data availability, improper distribution, and storage limitations. To mitigate these issues, transfer learning (TL) has been employed to transfer knowledge from one domain to a related domain, enabling models to be fine-tuned with small-scale data while ensuring adaptability across diverse contexts, including variations in subjects, datasets, and sessions. This review presents a detailed and systematic overview of studies from the current decade that have employed TL models for healthcare-related applications using biomedical signals. In the introduction section, we explain the importance of employing TL techniques on biomedical signals in various domains, including disease diagnosis and prediction, and brain-computer interfaces (BCIs). The following section presents TL strategies. Another section is dedicated to searching and selection of articles based on the PRISMA method from reference databases including IEEE, Scopus, Web of Science, and PubMed. In this review, we examined 239 Q1 articles. Review articles published using TL techniques with biomedical signals are discussed in a separate section. In this review, we have studied the papers that have utilized TL techniques with various biosignals for various applications. Following this, we discuss the key challenges and future directions for the field based on the reviewed articles and conclude with a summary of key findings. Based on our study, EEG signals were the most frequently utilized in TL methods, particularly in the context of Brain-Computer Interface (BCI) applications, followed by applications in epilepsy detection. Additionally, domain adaptation methods are widely used in biomedical signals to address variations in data distribution caused by differences in subjects, devices, datasets, and recording conditions. These methods aim to align source and target domains, enabling models to generalize effectively across diverse datasets. This study provides a comprehensive review of current TL methods, offering useful insights for choosing the most suitable TL techniques for specific applications. It aims to deal with problems like data scarcity, domain mismatches, real-time issues, and hardware resource constraints in real-world scenarios. © 2025 The Author(s) |
| Keywords | Artificial intelligence; Transfer learning; Pre-trained; Biomedical signals; Machine learning; Deep learning; Disease diagnosis; BCIs |
| Contains Sensitive Content | Does not contain sensitive content |
| ANZSRC Field of Research 2020 | 461103. Deep learning |
| Byline Affiliations | School of Mathematics, Physics and Computing |
| Kumamoto University, Japan | |
| University of Technology Sydney | |
| Australian International Institute of Higher Education, Australia | |
| Cogninet Australia, Australia | |
| University of New England | |
| SRM Institute of Science and Technology, India | |
| Duke-NUS Medical School, Singapore | |
| National Heart Centre, Singapore |
Permalink -
https://research.usq.edu.au/item/zx188/application-of-transfer-learning-for-biomedical-signals-a-comprehensive-review-of-the-last-decade-2014-2024
Download files
3
total views0
total downloads3
views this month0
downloads this month