English

Resource-Conscious Modeling for Next- Day Discharge Prediction Using Clinical Notes

Artificial Intelligence 2026-04-07 v1

Abstract

Timely discharge prediction is essential for optimizing bed turnover and resource allocation in elective spine surgery units. This study evaluates the feasibility of lightweight, fine-tuned large language models (LLMs) and traditional text-based models for predicting next-day discharge using postoperative clinical notes. We compared 13 models, including TF-IDF with XGBoost and LGBM, and compact LLMs (DistilGPT-2, Bio_ClinicalBERT) fine-tuned via LoRA. TF-IDF with LGBM achieved the best balance, with an F1-score of 0.47 for the discharge class, a recall of 0.51, and the highest AUC-ROC (0.80). While LoRA improved recall in DistilGPT2, overall transformer-based and generative models underperformed. These findings suggest interpretable, resource-efficient models may outperform compact LLMs in real-world, imbalanced clinical prediction tasks.

Keywords

Cite

@article{arxiv.2604.03498,
  title  = {Resource-Conscious Modeling for Next- Day Discharge Prediction Using Clinical Notes},
  author = {Ha Na Cho and Sairam Sutari and Alexander Lopez and Hansen Bow and Kai Zheng},
  journal= {arXiv preprint arXiv:2604.03498},
  year   = {2026}
}
R2 v1 2026-07-01T11:53:33.272Z