English

Language Model Training Paradigms for Clinical Feature Embeddings

Machine Learning 2024-02-07 v2 Computation and Language

Abstract

In research areas with scarce data, representation learning plays a significant role. This work aims to enhance representation learning for clinical time series by deriving universal embeddings for clinical features, such as heart rate and blood pressure. We use self-supervised training paradigms for language models to learn high-quality clinical feature embeddings, achieving a finer granularity than existing time-step and patient-level representation learning. We visualize the learnt embeddings via unsupervised dimension reduction techniques and observe a high degree of consistency with prior clinical knowledge. We also evaluate the model performance on the MIMIC-III benchmark and demonstrate the effectiveness of using clinical feature embeddings. We publish our code online for replication.

Keywords

Cite

@article{arxiv.2311.00768,
  title  = {Language Model Training Paradigms for Clinical Feature Embeddings},
  author = {Yurong Hu and Manuel Burger and Gunnar Rätsch and Rita Kuznetsova},
  journal= {arXiv preprint arXiv:2311.00768},
  year   = {2024}
}

Comments

Poster at "NeurIPS 2023 Workshop: Self-Supervised Learning - Theory and Practice"

R2 v1 2026-06-28T13:08:58.362Z