Taxonomy of synthetic data in healthcare
This taxonomy helps researchers understand and organize different types of synthetic data used in healthcare. It looks at how much synthetic data is used, what kinds of data formats can be created, and how synthetic data can improve dataset quality while protecting privacy.
At a glance
Use when
Designing or evaluating synthetic data generation methods in healthcare; comparing trade-offs between privacy, utility, and data type
Avoid when
When specific algorithmic guidance for synthetic data generation is required; in regulatory contexts requiring validated data generation protocols
Inputs
Healthcare datasets requiring privacy-preserving data sharing or augmentation
Outputs
Classification of synthetic data approaches based on proportion, modality, and transformation goals
How it works
The taxonomy classifies synthetic data in healthcare along three dimensions: Data Proportion (ratio of synthetic to real data in datasets), Data Modality (types of data formats such as structured, imaging, or text that can be synthesized), and Data Transformation (use of synthetic data to enhance utility, privacy, or fairness). It provides a structured framework to navigate current research and applications, highlighting trade-offs, challenges, and overlaps across categories.
- Project
- INSAFEDARE
- Funding
- Horizon Europe
- Project status
- Ongoing
- HTA domains
- Aspects Beyond HTA
- Categories
- Data ClassificationSynthetic Data
- Technology
- Non-specific
- Assumptions
- Synthetic data can sufficiently mimic real data distributions while preserving privacy and utility
- Strengths
- Provides a clear, multi-dimensional framework to compare synthetic data methods; supports decision-making in research and development; addresses privacy challenges in healthcare data sharing
- Limitations
- Does not specify technical implementation details for generating synthetic data; framework applicability may vary across data types and use cases
- Also known as
- Synthetic Data Taxonomy for Healthcare, INSAFEDARE Synthetic Data Taxonomy
Questions this answers
- › What types of synthetic data can be used in healthcare research?
- › How much synthetic data can be included in a dataset?
- › What are the privacy and utility trade-offs when using synthetic data?
- › Which healthcare data formats can be synthesized?
- › How can synthetic data improve the quality of real-world datasets?
- › What are the challenges in generating and using synthetic healthcare data?
References & sources
- paperarxiv.org ↗
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.