LLM-generated summaries can produce patient embeddings that are more 'portable' and robust to hospital distribution shifts than specialized clinical models.
March 26, 2026
Original Paper
Can we generate portable representations for clinical time series data using LLMs?
arXiv · 2603.23987
The Takeaway
It challenges the need for complex, domain-specific time-series models for clinical transfer. By leveraging LLMs to map irregular data to natural language, researchers found representations that maintain performance across different ICU cohorts with minimal-to-no retraining.
From the abstract
Deploying clinical ML is slow and brittle: models that work at one hospital often degrade under distribution shifts at the next. In this work, we study a simple question -- can large language models (LLMs) create portable patient embeddings i.e. representations of patients enable a downstream predictor built on one hospital to be used elsewhere with minimal-to-no retraining and fine-tuning. To do so, we map from irregular ICU time series onto concise natural language summaries using a frozen LLM