Effective Natural Language Processing Algorithms for Gout Flare Early Alert from Chief Complaints

Abstract

AbstractEarly identification of acute gout is crucial, enabling healthcare professionals to implement targeted interventions for rapid pain relief and preventing disease progression, ensuring improved long-term joint function. In this study, we comprehensively explored the potential of gout flare (GF) early detection based on nurse chief complaint notes in the Emergency Department (ED). Addressing the challenge of identifying GFs prospectively during an ED visit, where documentation is typically minimal, our research focuses on employing alternative Natural Language Processing (NLP) techniques to enhance the detection accuracy. We investigate GF detection algorithms using both sparse representations by traditional NLP methods and dense encodings by medical domain-specific Large Language Models (LLMs), distinguishing between generative and discriminative models. Three methods are used to alleviate the issue of severe data imbalance, including over-sampling, class weights, and focal loss. Extensive empirical studies are done on the Gout Emergency Department Chief Complaint Corpora. Sparse text representations like tf-idf proved to produce strong performance, achieving higher than 0.75 F1 Score. The best deep learning models are RoB-ERTa-Large-PM-M3-Voc and BioGPT, with the best F1 Scores on each dataset with a 0.8 on the 2019 dataset and a 0.85 F1 Score the 2020 dataset. We concluded that although discriminative LLMs performed better for this classification task, compared to generative LLMs, a combination of using generative models as feature extractors and employing support vector machine for classification yields promising results comparable to those obtained with discriminative models.