Interpretable Synthetic Medical Tabular Data Generation for Clinical Decision Support Using Fuzzy…

Operators should replace probabilistic or deep generative synthesizers with a Fuzzy Cognitive Map (FCM) pipeline when they need interpretable, causally grounded synthetic medical tabular data.

Examples

• DCR Baseline Protection scores consistently exceed those of TVAE.

Known limitations

• Existing probabilistic and deep generative models often lack interpretability and fail to preserve clinically meaningful dependencies, which is the gap this approach targets.

Examples

• Developing and validating computer-based medical systems (CBMSs) when real clinical data is restricted by ethical or data availability limitations.

Research · arXiv cs.AI · Publisher website: arxiv.org · Last verified 8 November 2026 · Catalog ID 132847