Relation Extraction Using Large Language Models: A Case Study on Acupuncture Point Locations

In acupuncture therapy, the accurate location of acupoints is essential for its effectiveness. The advanced language understanding capabilities of large language models (LLMs) like Generative Pre-trained Transformers (GPT) present a significant opportunity for extracting relations related to acupoint locations from textual knowledge sources. This study aims to compare the performance of GPT with traditional deep learning models (Long Short-Term Memory (LSTM) and Bidirectional Encoder Representations from Transformers for Biomedical Text Mining (BioBERT)) in extracting acupoint-related location relations and assess the impact of pretraining and fine-tuning on GPT's performance. We utilized the World Health Organization Standard Acupuncture Point Locations in the Western Pacific Region (WHO Standard) as our corpus, which consists of descriptions of 361 acupoints. Five types of relations ('direction_of,' 'distance_of,' 'part_of,' 'near_acupoint,' and 'located_near') (n= 3,174) between acupoints were annotated. Five models were compared: BioBERT, LSTM, pre-trained GPT-3.5, fine-tuned GPT-3.5, as well as pre-trained GPT-4. Performance metrics included micro-average exact match precision, recall, and F1 scores. Our results demonstrate that fine-tuned GPT-3.5 consistently outperformed other models in F1 scores across all relation types. Overall, it achieved the highest micro-average F1 score of 0.92. This study underscores the effectiveness of LLMs like GPT in extracting relations related to acupoint locations, with implications for accurately modeling acupuncture knowledge and promoting standard implementation in acupuncture training and practice. The findings also contribute to advancing informatics applications in traditional and complementary medicine, showcasing the potential of LLMs in natural language processing.

Zero-shot Clinical Entity Recognition using ChatGPT

In this study, we investigated the potential of ChatGPT, a large language model developed by OpenAI, for the clinical named entity recognition task defined in the 2010 i2b2 challenge, in a zero-shot setting with two different prompt strategies. We compared its performance with GPT-3 in a similar zero-shot setting, as well as a fine-tuned BioClinicalBERT model using a set of synthetic clinical notes from MTSamples. Our findings revealed that ChatGPT outperformed GPT-3 in the zero-shot setting, with F1 scores of 0.418 (vs.0.250) and 0.620 (vs. 0.480) for exact- and relaxed-matching, respectively. Moreover, prompts affected ChatGPT's performance greatly, with relaxed-matching F1 scores of 0.628 vs.0.541 for two different prompt strategies. Although ChatGPT's performance was still lower than that of the supervised BioClinicalBERT model (i.e., relaxed-matching F1 scores of 0.628 vs. 0.870), our study demonstrates the great potential of ChatGPT for clinical NER tasks in a zero-shot setting, which is much more appealing as it does not require any annotation.