Ensemble pretrained language models to extract biomedical knowledge from literature.

OBJECTIVES: The rapid expansion of biomedical literature necessitates automated techniques to discern relationships between biomedical concepts from extensive free text. Such techniques facilitate the development of detailed knowledge bases and highlight research deficiencies. The LitCoin Natural Language Processing (NLP) challenge, organized by the National Center for Advancing Translational Science, aims to evaluate such potential and provides a manually annotated corpus for methodology development and benchmarking. MATERIALS AND METHODS: For the named entity recognition (NER) task, we utilized ensemble learning to merge predictions from three domain-specific models, namely BioBERT, PubMedBERT, and BioM-ELECTRA, devised a rule-driven detection method for cell line and taxonomy names and annotated 70 more abstracts as additional corpus. We further finetuned the T0pp model, with 11 billion parameters, to boost the performance on relation extraction and leveraged entites' location information (eg, title, background) to enhance novelty prediction performance in relation extraction (RE). RESULTS: Our pioneering NLP system designed for this challenge secured first place in Phase I-NER and second place in Phase II-relation extraction and novelty prediction, outpacing over 200 teams. We tested OpenAI ChatGPT 3.5 and ChatGPT 4 in a Zero-Shot setting using the same test set, revealing that our finetuned model considerably surpasses these broad-spectrum large language models. DISCUSSION AND CONCLUSION: Our outcomes depict a robust NLP system excelling in NER and RE across various biomedical entities, emphasizing that task-specific models remain superior to generic large ones. Such insights are valuable for endeavors like knowledge graph development and hypothesis formulation in biomedical research.

Improving large language models for clinical named entity recognition via prompt engineering.

IMPORTANCE: The study highlights the potential of large language models, specifically GPT-3.5 and GPT-4, in processing complex clinical data and extracting meaningful information with minimal training data. By developing and refining prompt-based strategies, we can significantly enhance the models' performance, making them viable tools for clinical NER tasks and possibly reducing the reliance on extensive annotated datasets. OBJECTIVES: This study quantifies the capabilities of GPT-3.5 and GPT-4 for clinical named entity recognition (NER) tasks and proposes task-specific prompts to improve their performance. MATERIALS AND METHODS: We evaluated these models on 2 clinical NER tasks: (1) to extract medical problems, treatments, and tests from clinical notes in the MTSamples corpus, following the 2010 i2b2 concept extraction shared task, and (2) to identify nervous system disorder-related adverse events from safety reports in the vaccine adverse event reporting system (VAERS). To improve the GPT models' performance, we developed a clinical task-specific prompt framework that includes (1) baseline prompts with task description and format specification, (2) annotation guideline-based prompts, (3) error analysis-based instructions, and (4) annotated samples for few-shot learning. We assessed each prompt's effectiveness and compared the models to BioClinicalBERT. RESULTS: Using baseline prompts, GPT-3.5 and GPT-4 achieved relaxed F1 scores of 0.634, 0.804 for MTSamples and 0.301, 0.593 for VAERS. Additional prompt components consistently improved model performance. When all 4 components were used, GPT-3.5 and GPT-4 achieved relaxed F1 socres of 0.794, 0.861 for MTSamples and 0.676, 0.736 for VAERS, demonstrating the effectiveness of our prompt framework. Although these results trail BioClinicalBERT (F1 of 0.901 for the MTSamples dataset and 0.802 for the VAERS), it is very promising considering few training samples are needed. DISCUSSION: The study's findings suggest a promising direction in leveraging LLMs for clinical NER tasks. However, while the performance of GPT models improved with task-specific prompts, there's a need for further development and refinement. LLMs like GPT-4 show potential in achieving close performance to state-of-the-art models like BioClinicalBERT, but they still require careful prompt engineering and understanding of task-specific knowledge. The study also underscores the importance of evaluation schemas that accurately reflect the capabilities and performance of LLMs in clinical settings. CONCLUSION: While direct application of GPT models to clinical NER tasks falls short of optimal performance, our task-specific prompt framework, incorporating medical knowledge and training samples, significantly enhances GPT models' feasibility for potential clinical applications.

Integrating Commercial and Social Determinants of Health: A Unified Ontology for Non-Clinical Determinants of Health.

The pivotal impact of Social Determinants of Health (SDoH) on people's health and well-being has been widely recognized and researched. However, the effect of Commercial Determinants of Health (CDoH) is only now garnering increased attention. Developing an ontology for CDoH can offer a systematic approach to identifying and categorizing the diverse commercial factors affecting health. These factors, including the production, distribution, and marketing of goods and services, may exert a substantial influence on health outcomes. The objectives of this research are 1) to develop an ontology for CDoH by utilizing PubMed articles and ChatGPT; 2) to foster ontology reuse by integrating CDoH with an existing SDoH ontology into a unified structure; 3) to devise an overarching conception for all nonclinical determinants of health (N-CDoH) and to create an initial ontology for N-CDoH; 4) and to validate the degree of correspondence between concepts provided by ChatGPT with the existing SDoH ontology.

Extended Analysis of Topological-Pattern-Based Ontology Enrichment.

Maintenance of biomedical ontologies is difficult. We have previously developed a topological-pattern-based method to deal with the problem of identifying concepts in a reference ontology that could be of interest for insertion into a target ontology. Assuming that both ontologies are parts of the Unified Medical Language System (UMLS), the method suggests approximate locations where the target ontology could be extended with new concepts from the reference ontology. However, the final decision about each concept has to be made by a human expert. In this paper, we describe the universe of cross-ontology topological patterns in quantitative terms. We then present a theoretical analysis of the number of potential placements of reference concepts in a path in a target ontology, allowing for new cross-ontology synonyms. This provides a rough estimate of what expert resources need to be allocated for the task. One insight in previous work on this topic was the large percentage of cases where importing concepts was impossible, due to a configuration called "alternative classification." In this paper, we confirm this observation. Our target ontology is the National Cancer Institute thesaurus (NCIt). However, the methods can be applied to other pairs of ontologies with hierarchical relationships from the UMLS.