Improving the Realism of Synthetic Clinical Benchmarks Under Utility Constraints

Source

Source name: arXiv
Source kind: paper
Authors / organization: Omid Bazgir, Md Nasir, Jacob Hoffman, Yang Yang, Manu Agrawal, Anusua Trivedi, Vinay Rao Dandin, Chris Gibbons, Christine Swisher
Venue: arXiv: cs.AI
Published: 2026-08-06
Collected: 2026-08-07
URL: http://arxiv.org/abs/2608.06265v1
PDF: https://arxiv.org/pdf/2608.06265v1
DOI: UNKNOWN

Plain-Language Brief

Researchers propose a method to improve the realism of synthetic clinical benchmarks for AI agents in healthcare settings without compromising their downstream utility checks. They develop a framework that ensures dataset changes increase realism while maintaining an operational utility floor. Their results demonstrate that two deterministic revisions improve the realism of the baseline benchmark, while a naive densification control does not.

What the Document Says

The study focuses on improving synthetic clinical benchmarks for enterprise AI agents in privacy-sensitive healthcare settings.
The baseline benchmark is unrealistic, with high missingness rates and low actionability.
Two deterministic revisions improve the realism of the panels while maintaining the current utility floor.
A naive densification control does not lead to improvements in realism.
Internal benchmark realism and source fidelity are related but distinct objectives.

Key Concepts Mentioned

synthetic clinical benchmarks
enterprise AI agents
healthcare settings
utility constraints
realism improvement
missingness structure
simplicity
structural plausibility
population alignment

Claims Made by the Document

Synthetic clinical benchmarks can be unrealistic, especially in privacy-sensitive healthcare settings.
Two deterministic revisions improve the realism of the panels while maintaining the current utility floor.
Internal benchmark realism and source fidelity are related but distinct objectives.

Evidence Used

The text mentions that the baseline benchmark is unrealistic, with high missingness rates and low actionability.
It is stated that two deterministic revisions improve the realism of the panels while maintaining the current utility floor.
The text suggests that internal benchmark realism and source fidelity are related but distinct objectives.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Synthetic clinical benchmarks for enterprise AI agents can pass existing utility checks and still remain structurally unrealistic, especially in privacy-sensitive healthcare settings where operational data are hard to access. We study how to improve such benchmarks without breaking the downstream utility checks already used in practice. We formulate benchmark revision as utility-constrained realism improvement: dataset changes should increase realism while staying above an operational utility floor. We instantiate this idea on a care-gap benchmark derived from Synthea-generated patients exercised through demonstration electronic health record workflows and then processed by the same downstream pipeline as operational data. Realism is measured through missingness structure, simplicity, structural plausibility, and population alignment. The baseline benchmark is extremely thin: sampled-pair missingness is 79.44%, only 12.75% of rows are actionable, 38.94% of patients have zero actionable measures, and top-three token concentration reaches 100.0%. Two deterministic revisions improve these panels while remaining above the current utility floor, whereas a naive densification control preserves unrealistic templating. We further show that internal benchmark realism and source fidelity to an aggregate operational reference are related but distinct objectives. These results suggest that synthetic benchmark quality should be optimized explicitly, with utility treated as one constraint rather than as sufficient evidence of realism.