Survey on Data-Centric Foundation Models in Healthcare AI
A survey published on arXiv examines the role of data-centric approaches in foundation models for computational healthcare. Foundation models, a suite of AI techniques, are reshaping healthcare by emphasizing better data characterization, quality, and scale. The study covers challenges in obtaining high-quality clinical data, including quantity, annotation, patient privacy, and ethics. It explores data-centric methods from model pre-training to inference, and discusses AI security, assessment, and alignment with human values. The survey offers a forward-looking perspective on using foundation models to improve patient outcomes and clinical workflows.
Key facts
- The survey is titled 'Data-Centric Foundation Models in Computational Healthcare: A Survey'.
- It was published on arXiv with ID 2401.02458.
- The announcement type is replace-cross.
- Foundation models are described as an emerging suite of AI techniques.
- The survey emphasizes a data-centric AI paradigm guided by pre-training data and human instructions.
- Challenges in healthcare AI include data quantity, annotation, patient privacy, and ethics.
- The survey covers data-centric approaches from model pre-training to inference.
- Key perspectives discussed include AI security, assessment, and alignment with human values.
Entities
Institutions
- arXiv