mlsanityData to 7 Oct 2026

Critical Challenges and Guidelines in Evaluating Synthetic Tabular Data: A Systematic Review

arXiv 2504.18544 ↗First posted 10 Apr 2025cs.LG cs.AI cs.CYcited by 0

Not on the map. Why?

Abstract (arXiv)

Generating synthetic tabular health data is challenging, and evaluating their quality is equally, if not more, complex. This systematic review highlights the critical importance of rigorous evaluation of synthetic health data to ensure reliability, clinical relevance, and appropriate use. From an initial identification of 2067 relevant papers published in the last ten years, 134 studies were selected for detailed analysis. Our review identifies key challenges, including lack of consensus on evaluation methods, inconsistent application of evaluation metrics, limited involvement of domain experts, inadequate reporting of dataset characteristics, and limited reproducibility of results. In response, we provide a structured consolidation of synthetic data generation and evaluation methods into taxonomies, alongside practical guidelines to support more robust and standardised evaluation practices. These findings aim to support the responsible development and use of synthetic health data, aligned with emerging expectations around transparency, reproducibility, and governance, ultimately enabling the community to fully harness its transformative potential and accelerate innovation.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 26 of its 158 references that are in the map. First posted on arXiv 10 Apr 2025.

Builds on

26 of 158 references.

References · 26

  1. 2019BERTScore: Evaluating Text Generation with BERTcited by 5,462
  2. 2022TabDDPM: Modelling Tabular Data with Diffusion Modelscited by 373
  3. 2021How Faithful is your Synthetic Data? Sample-level Metrics for Evaluating and Auditing Generative Modelscited by 193
  4. 2021DECAF: Generating Fair Synthetic Data Using Causally-Aware Generative Networkscited by 109
  5. 2024TabDiff: a Mixed-type Diffusion Model for Tabular Data Generationcited by 60
  6. 2021Generating Synthetic Mixed-type Longitudinal Electronic Health Records for Artificial Intelligent Applicationscited by 47
  7. 2023Reliable Generation of Privacy-preserving Synthetic Electronic Health Record Time Series via Diffusion Modelscited by 24
  8. 2020Estimating Individual Treatment Effects with Time-Varying Confounderscited by 21
  9. 2025TimeDP: Learning to Generate Multi-Domain Time Series with Domain Promptscited by 21
  10. 2022The Health Gym: Synthetic Health-Related Datasets for the Development of Reinforcement Learning Algorithmscited by 19
  11. 2022Multi-Label Clinical Time-Series Generation via Conditional GANcited by 16
  12. 2024Diffusion Models for Tabular Data Imputation and Synthetic Data Generationcited by 15
14 more
  1. 2022Conditional Generation of Medical Time Series for Extrapolation to Underrepresented Populationscited by 14
  2. 2025Population Aware Diffusion for Time Series Generationcited by 14
  3. 2018Measuring the quality of Synthetic data for use in competitionscited by 11
  4. 2024Synthesizing Multimodal Electronic Health Records via Predictive Diffusion Modelscited by 7
  5. 2024SynEHRgy: Synthesizing Mixed-Type Structured Electronic Health Records using Decoder-Only Transformerscited by 5
  6. 2025TabRep: Training Tabular Diffusion Models with a Simple and Effective Continuous Representationcited by 5
  7. 2016Patient Flow Prediction via Discriminative Learning of Mutually-Correcting Processescited by 4
  8. 2023MedDiffusion: Boosting Health Risk Prediction via Diffusion-based Data Augmentationcited by 4
  9. 2024Protect and Extend -- Using GANs for Synthetic Data Generation of Time-Series Medical Recordscited by 4
  10. 2021Synthesizing time-series wound prognosis factors from electronic medical records using generative adversarial networkscited by 3
  11. 2025Synthetic Data Generation for Augmenting Small Samplescited by 3
  12. 2014Predicting epidemic risk from past temporal contact datacited by 1
  13. 2021Impulse data models for the inverse problem of electrocardiographycited by 1
  14. 2024Fairness-Optimized Synthetic EHR Generation for Arbitrary Downstream Predictive Taskscited by 1

Built on by

Not cited yet.