Applications and prospects of transfer learning in rare diseases research
10.12173/j.issn.1005-0698.202412136
- VernacularTitle:迁移学习在罕见病研究中的应用与展望
- Author:
Xueying ZHENG
1
;
Guoyou QIN
;
Yongfu YU
Author Information
1. 复旦大学公共卫生学院生物统计教研室(上海 200032)
- Publication Type:Journal Article
- Keywords:
Rare diseases;
Scarce data;
Transfer learning
- From:
Chinese Journal of Pharmacoepidemiology
2025;34(8):986-992
- CountryChina
- Language:Chinese
-
Abstract:
Transfer learning is a method of learning new tasks in related domain using existing knowledge from source data.In rare disease research,data are often limited.Transfer learning can effectively use data from other related diseases or fields to enhance model performance and research efficiency.This approach helps researchers rapidly identify characteristics and develop potential treatments of rare disease.Currently,transfer learning has been applied in the systematic characterization and drug development of rare diseases.It also shows potential in optimizing rare disease classification,accelerating early diagnosis,and supporting multi-task research.However,challenges arise in the application of transfer learning in rare disease research.In the future,if transfer learning can be combined with techniques such as reinforcement learning,federated learning,and deep learning,greater breakthroughs are expected to be achieved in the field of rare diseases.
