cd /news/machine-learning/generating-1000000-synthetic-medical… · home › topics › machine-learning › article
[ARTICLE · art-173435] src=dev.to ↗ pub= topic=machine-learning verified=true sentiment=↑ positive

Generating 1,000,000 Synthetic Medical Records in Seconds

A developer built a two-stage synthetic data engine that combines class-conditional Gaussian copulas with decision trees to generate up to 1,000,000 realistic tabular medical records in seconds, avoiding the GPU cost and clinical-rule violations of deep-learning synthesizers like CTGAN and diffusion models. Tested on an Apache 2.0 asthma dataset, the engine reported mean distribution differences under 0.6%, a correlation difference of 0.0225, zero boundary or non-asthmatic rule violations, zero exact clones of real patients, and a Random Forest TSTR ROC-AUC gap of 0.0013. The generator is published as a live Hugging Face Space alongside a 10k-patient Kaggle benchmark dataset and companion notebooks.

recommendation

For tabular synthetic data, try class-conditional Gaussian copulas with decision-tree rules before reaching for CTGAN or diffusion models; the source reports sub-0.6% distribution error and zero clinical-rule violations.

more in playbook: Fine-tuning and data
by read1 min views6 publishedNov 20, 2026

Generating realistic tabular medical data usually comes with a frustrating trade-off: deep learning models (CTGAN, Diffusion) are slow, GPU-heavy, and prone to violating clinical rules.

To solve this, we built a transparent, ultra-fast Two-Stage Generative Engine combining Class-Conditional Gaussian Copulas and Decision Trees.

Asthma_Control_Level, perfectly decoupling non-asthmatics.

Pillar Metric Result Status
Statistical Fidelity Mean Diffs / Correlation Diff < 0.6% / 0.0225 PASS
Clinical Logic Boundary Violations / Non-Asthmatic Rule 0 Violations / 100% 'N/A' PASS
Privacy Guarantee Exact Clones vs Real Data 0 Duplicates PASS

| ML Utility (TSTR) | Random Forest $\Delta \text{ROC-AUC}$ | 0.0013 | PASS | Provenance note: Fitted on Arif Miah's (miadul) Apache 2.0 dataset on Kaggle. Replicates synthetic data distributions, not human biology.

Live Hugging Face Space (Try generating cohorts up to 1M):

[https://huggingface.co/spaces/abdurrahman-hakan-acar/synthetic-asthma-generator](https://huggingface.co/spaces/abdurrahman-hakan-acar/synthetic-asthma-generator)

**Kaggle Benchmark Dataset (10k Patients - Usability 9.41):**

[https://www.kaggle.com/datasets/abdurrahmanhakanacar/synthetic-asthma-medical-dataset-10k-patients](https://www.kaggle.com/datasets/abdurrahmanhakanacar/synthetic-asthma-medical-dataset-10k-patients)

Kaggle EDA & Baseline ML Notebook:

https://www.kaggle.com/code/abdurrahmanhakanacar/synthetic-asthma-data-eda-risk-factor-modeling Kaggle On-Demand Generator Notebook:

https://www.kaggle.com/code/abdurrahmanhakanacar/synthetic-asthma-data-generator-on-demand-custom What is your preferred approach for tabular data synthesis: statistical copulas or deep neural nets? Let's discuss below!

── more in #machine-learning 4 stories · sorted by recency
── more on @hugging face 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/generating-1000000-s…] indexed:0 read:1min 2026-11-20 · —