Improving Clinical Validity in Synthetic Electronic Health Record Generation Using Best-of-N Sampling: Comparative Evaluation Study
JMIR AI · 30.07.2026 · 5 · e90590-e90590
DOI: 10.2196/90590 ↗Kurz gefasst
Fachartikel in der Zeitschrift JMIR AI, online erschienen am 30.07.2026. Verfasst von Md Akmol Masud und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electrical and Computer Engineering, Queen's University, 99 University Ave, Kingston, ON, K7L 3N6, Canada, +880 1304963440. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.
Abstract (Originalsprache)
Background
Synthetic electronic health record generation is limited not only by statistical fidelity but also by clinical validity. Records that appear statistically plausible may still violate hard structural, physiological, or relational constraints.
Objective
This study evaluated best-of-N constraint-minimizing selection as an inference-time strategy for improving clinical validity and characterized when such selection succeeds or fails as a function of the generator’s valid support mass ( p valid ).
Methods
We evaluated Wasserstein generative adversarial network with gradient penalty (WGAN-GP) and conditional tabular generative adversarial network (CTGAN) on 3 public clinical tabular datasets: stroke prediction (n=5110), diabetes health indicators (n=100,000), and cardiovascular disease (n=68,599). For each generator, we compared random sampling, naive clipping, and best-of-N selection (N ∈ {8, 16, 128}). Validity was assessed via dataset-specific rule violations; fidelity via Kolmogorov-Smirnov (KS) statistics and correlation preservation; utility via TSTR (train-on-synthetic-test-on-real) AUC (area under the receiver operating characteristic curve); and privacy via membership inference attack (MIA) AUC and distance to closest record (DCR).
Results
Best-of-N effectiveness was governed by the base generator’s valid support mass. In stroke, WGAN-GP had p valid =0, and all selection strategies, including strict rejection across 5000 draws, produced 500/500 (100%) invalid samples. In cardiovascular data, WGAN-GP had p valid =0.12; best-of-16 reduced violations from 441/500 (88.2%) to 65/500 (13.0%), and best-of-128 eliminated them entirely (0/500), matching theoretical predictions exactly. In diabetes, WGAN-GP random sampling was already fully valid (0/500 violations), and best-of-N primarily shifted utility. Across all 3 datasets, CTGAN+best-of-16 achieved 0/500 violations with strong fidelity (mean KS≤0.12) and utility (TSTR AUC up to 0.81). MIA AUC remained near-random guessing (approximately 0.50) throughout, though DCR-based proximity concentration increased markedly under CTGAN selection (reaching 22.6% on stroke and 100% on cardiovascular data).
Conclusions
Best-of-N functions as a bounded-budget feasibility filter rather than a universal repair mechanism. Its effectiveness depends critically on the generator’s valid support mass. CTGAN+best-of-16 offered the strongest overall trade-off across clinical validity, fidelity, utility, and privacy in our experiments.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Electrical and Computer Engineering, Queen's University, 99 University Ave, Kingston, ON, K7L 3N6, Canada, +880 1304963440.
- Department of Biostatistics, Virginia Commonwealth University, Richmond, VA, United States.
Zitieren
APA
Masud, M. A., & Hasan, M. (2026). Improving Clinical Validity in Synthetic Electronic Health Record Generation Using Best-of-N Sampling: Comparative Evaluation Study. JMIR AI, 5, e90590-e90590. https://doi.org/10.2196/90590
Vancouver
Masud MA, Hasan M. Improving Clinical Validity in Synthetic Electronic Health Record Generation Using Best-of-N Sampling: Comparative Evaluation Study. JMIR AI. 2026;5:e90590-e90590. doi:10.2196/90590
BibTeX
@article{Masud2026,
author = {Masud, Md Akmol and Hasan, Mahmud},
title = {Improving Clinical Validity in Synthetic Electronic Health Record Generation Using Best-of-N Sampling: Comparative Evaluation Study},
journal = {JMIR AI},
year = {2026},
volume = {5},
pages = {e90590--e90590},
doi = {10.2196/90590}
}
EUVIMED