Counterfactual Modeling with Fine-Tuned LLMs for Health Intervention Design and Sensor Data Augmentation

TL;DR

This paper evaluates large language models for generating counterfactual explanations in health data, demonstrating their effectiveness in intervention design and data augmentation to improve model robustness.

Contribution

It introduces a fine-tuned LLM approach for generating high-quality, clinically actionable counterfactuals that enhance health intervention strategies and data efficiency.

Findings

Fine-tuned LLMs produce plausible, valid counterfactuals with up to 99% plausibility.

Counterfactual data augmentation restores classifier performance with 20% F1 score recovery.

LLMs outperform optimization-based baselines in generating coherent, actionable counterfactuals.

Abstract

Counterfactual explanations (CFEs) provide human-centric interpretability by identifying the minimal, actionable changes required to alter a machine learning model's prediction. Therefore, CFs can be used as (i) interventions for abnormality prevention and (ii) augmented data for training robust models. We conduct a comprehensive evaluation of CF generation using large language models (LLMs), including GPT-4 (zero-shot and few-shot) and two open-source models-BioMistral-7B and LLaMA-3.1-8B, in both pretrained and fine-tuned configurations. Using the multimodal AI-READI clinical dataset, we assess CFs across three dimensions: intervention quality, feature diversity, and augmentation effectiveness. Fine-tuned LLMs, particularly LLaMA-3.1-8B, produce CFs with high plausibility (up to 99%), strong validity (up to 0.99), and realistic, behaviorally modifiable feature adjustments. When used for data augmentation under controlled label-scarcity settings, LLM-generated CFs substantially restore classifier performance, yielding an average 20% F1 recovery across three scarcity scenarios. Compared with optimization-based baselines such as DiCE, CFNOW, and NICE, LLMs offer a flexible, model-agnostic approach that generates more clinically actionable and semantically coherent counterfactuals. Overall, this work demonstrates the promise of LLM-driven counterfactuals for both interpretable intervention design and data-efficient model training in sensor-based digital health. Impact: SenseCF fine-tunes an LLM to generate valid, representative counterfactual explanations and supplement minority class in an imbalanced dataset for improving model training and boosting model robustness and predictive performance