PaReGTA: An LLM-based EHR Data Encoding Approach to Capture Temporal Information
This work addresses the loss of temporal information in EHR data for healthcare applications, offering a model-agnostic and interpretable approach that is incremental in leveraging pre-trained LLMs for domain adaptation.
The authors tackled the problem of capturing temporal information in EHR data by proposing PaReGTA, an LLM-based encoding framework that converts events into text and uses fine-tuning and pooling to create patient representations, resulting in improved performance for migraine type classification on a dataset of 39,088 patients compared to sparse baselines.
Temporal information in structured electronic health records (EHRs) is often lost in sparse one-hot or count-based representations, while sequence models can be costly and data-hungry. We propose PaReGTA, an LLM-based encoding framework that (i) converts longitudinal EHR events into visit-level templated text with explicit temporal cues, (ii) learns domain-adapted visit embeddings via lightweight contrastive fine-tuning of a sentence-embedding model, and (iii) aggregates visit embeddings into a fixed-dimensional patient representation using hybrid temporal pooling that captures both recency and globally informative visits. Because PaReGTA does not require training from scratch but instead utilizes a pre-trained LLM, it can perform well even in data-limited cohorts. Furthermore, PaReGTA is model-agnostic and can benefit from future EHR-specialized sentence-embedding models. For interpretability, we introduce PaReGTA-RSS (Representation Shift Score), which quantifies clinically defined factor importance by recomputing representations after targeted factor removal and projecting representation shifts through a machine learning model. On 39,088 migraine patients from the All of Us Research Program, PaReGTA outperforms sparse baselines for migraine type classification while deep sequential models were unstable in our cohort.