Classification of kinetic-related injury in hospital triage data using NLP

2025-09-05

AI Summary

This study addresses the classification of kinetic-energy-related injuries in hospital triage texts, tackling three key challenges: data privacy sensitivity, high annotation costs, and constrained edge-computing resources. We propose a lightweight two-stage fine-tuning paradigm: first, preliminary fine-tuning of a pre-trained large language model on 2K open-source samples using GPU; second, efficient secondary adaptation on one thousand anonymized hospital records executed entirely on CPU. This approach eliminates reliance on high-end hardware or extensive expert annotations, substantially lowering deployment barriers. Experimental results demonstrate that the model maintains high classification accuracy under low-resource conditions while ensuring patient data privacy, computational efficiency, and clinical applicability. The method provides a practical, deployable solution for intelligent triage in resource-limited primary healthcare settings.

Technology Category

• Machine Learning: Hardware-aware ML
• Natural Language Processing: Safety and Robustness
• Data Mining & Knowledge Management: Intelligent Query Processing

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Triage notes, created at the start of a patient's hospital visit, contain a wealth of information that can help medical staff and researchers understand Emergency Department patient epidemiology and the degree of time-dependent illness or injury. Unfortunately, applying modern Natural Language Processing and Machine Learning techniques to analyse triage data faces some challenges: Firstly, hospital data contains highly sensitive information that is subject to privacy regulation thus need to be analysed on site; Secondly, most hospitals and medical facilities lack the necessary hardware to fine-tune a Large Language Model (LLM), much less training one from scratch; Lastly, to identify the records of interest, expert inputs are needed to manually label the datasets, which can be time-consuming and costly. We present in this paper a pipeline that enables the classification of triage data using LLM and limited compute resources. We first fine-tuned a pre-trained LLM with a classifier using a small (2k) open sourced dataset on a GPU; and then further fine-tuned the model with a hospital specific dataset of 1000 samples on a CPU. We demonstrated that by carefully curating the datasets and leveraging existing models and open sourced data, we can successfully classify triage data with limited compute resources.

Problem

• Classifying kinetic-related injury in triage notes
• Overcoming privacy and hardware constraints for NLP
• Reducing expert labeling costs with limited resources

Innovation

• Fine-tuned pre-trained LLM with classifier
• Used small open dataset on GPU first
• Further fine-tuned with hospital data on CPU