FairCauseSyn: Towards Causally Fair LLM-Augmented Synthetic Data Generation

2025-06-23

AI Summary

Existing synthetic data methods in healthcare predominantly target counterfactual fairness, neglecting causal fairness modeling and failing to preserve the underlying causal structure of real medical data. Method: We propose the first table-based synthetic data generation framework integrating large language models (LLMs) with causal inference, explicitly modeling causal pathways involving sensitive attributes. Our approach employs an LLM-enhanced generator coupled with causally constrained adversarial training to ensure causal structure invariance. Contribution/Results: Evaluated on real-world healthcare datasets, our method achieves <10% deviation on causal fairness metrics; predictive models trained on the synthetic data exhibit a 70% reduction in bias with respect to sensitive attributes. This work pioneers the unification of LLM-driven generation and causal fairness in medical tabular data synthesis, substantially enhancing fairness and trustworthiness in health research and clinical decision-making.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Causal Learning
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Synthetic data generation creates data based on real-world data using generative models. In health applications, generating high-quality data while maintaining fairness for sensitive attributes is essential for equitable outcomes. Existing GAN-based and LLM-based methods focus on counterfactual fairness and are primarily applied in finance and legal domains. Causal fairness provides a more comprehensive evaluation framework by preserving causal structure, but current synthetic data generation methods do not address it in health settings. To fill this gap, we develop the first LLM-augmented synthetic data generation method to enhance causal fairness using real-world tabular health data. Our generated data deviates by less than 10% from real data on causal fairness metrics. When trained on causally fair predictors, synthetic data reduces bias on the sensitive attribute by 70% compared to real data. This work improves access to fair synthetic data, supporting equitable health research and healthcare delivery.

Problem

• Ensuring causal fairness in synthetic health data generation
• Addressing bias in sensitive attributes for equitable outcomes
• Developing LLM-augmented methods for causally fair synthetic data

Innovation

• LLM-augmented synthetic data generation
• Enhances causal fairness in health data
• Reduces bias by 70% on sensitive attributes