Search

Relation extraction from the scientific literature to comply with a domain ontology is a well-known problem in natural language processing and is particularly critical in precision medicine. The advent of large language models (LLMs) has paved the way for the development of new effective approaches to this problem, but the extracted relations can be affected by issues such as hallucination, which must be minimized. In this paper, we present the initial design and preliminary experimental validation of SPIREX, an extension of the SPIRES-based system for the extraction of RDF triples from scientific literature involving RNA molecules. Our system exploits schema constraints in the formulations of LLM prompts along with our RNA-based KG, RNA-KG, for evaluating the plausibility of the extracted triples. RNA-KG contains more than 9M edges representing different kinds of relationships in which RNA molecules can be involved. Initial experimental results on a controlled data set are quite encouraging.

Biomedical research and clinical practice are in the midst of a transition toward significantly increased use of artificial intelligence (AI) and machine learning (ML) methods. These advances promise to enable qualitatively deeper insight into complex challenges formerly beyond the reach of analytic methods and human intuition while placing increased demands on ethical and explainable artificial intelligence (XAI), given the opaque nature of many deep learning methods. The U.S. National Institutes of Health (NIH) has initiated a significant research and development program, Bridge2AI, aimed at producing new "flagship" datasets designed to support AI/ML analysis of complex biomedical challenges, elucidate best practices, develop tools and standards in AI/ML data science, and disseminate these datasets, tools, and methods broadly to the biomedical community. An essential set of concepts to be developed and disseminated in this program along with the data and tools produced are criteria for AI-readiness of data, including critical considerations for XAI and ethical, legal, and social implications (ELSI) of AI technologies. NIH Bridge to Artificial Intelligence (Bridge2AI) Standards Working Group members prepared this article to present methods for assessing the AI-readiness of biomedical data and the data standards perspectives and criteria we have developed throughout this program. While the field is rapidly evolving, these criteria are foundational for scientific rigor and the ethical design and application of biomedical AI methods.

Phenotypic data are critical for understanding biological mechanisms and consequences of genomic variation, and are pivotal for clinical use cases such as disease diagnostics and treatment development. For over a century, vast quantities of phenotype data have been collected in many different contexts covering a variety of organisms. The emerging field of phenomics focuses on integrating and interpreting these data to inform biological hypotheses. A major impediment in phenomics is the wide range of distinct and disconnected approaches to recording the observable characteristics of an organism. Phenotype data are collected and curated using free text, single terms or combinations of terms, using multiple vocabularies, terminologies, or ontologies. Integrating these heterogeneous and often siloed data enables the application of biological knowledge both within and across species. Existing integration efforts are typically limited to mappings between pairs of terminologies; a generic knowledge representation that captures the full range of cross-species phenomics data is much needed. We have developed the Unified Phenotype Ontology (uPheno) framework, a community effort to provide an integration layer over domain-specific phenotype ontologies, as a single, unified, logical representation. uPheno comprises (1) a system for consistent computational definition of phenotype terms using ontology design patterns, maintained as a community library; (2) a hierarchical vocabulary of species-neutral phenotype terms under which their species-specific counterparts are grouped; and (3) mapping tables between species-specific ontologies. This harmonized representation supports use cases such as cross-species integration of genotype-phenotype associations from different organisms and cross-species informed variant prioritization., Competing Interests: 7.Conflict of Interest None declared.

Objective: Clinical deep phenotyping and phenotype annotation play a critical role in both the diagnosis of patients with rare disorders as well as in building computationally-tractable knowledge in the rare disorders field. These processes rely on using ontology concepts, often from the Human Phenotype Ontology, in conjunction with a phenotype concept recognition task (supported usually by machine learning methods) to curate patient profiles or existing scientific literature. With the significant shift in the use of large language models (LLMs) for most NLP tasks, we examine the performance of the latest Generative Pre-trained Transformer (GPT) models underpinning ChatGPT as a foundation for the tasks of clinical phenotyping and phenotype annotation., Materials and Methods: The experimental setup of the study included seven prompts of various levels of specificity, two GPT models (gpt-3.5-turbo and gpt-4.0) and two established gold standard corpora for phenotype recognition, one consisting of publication abstracts and the other clinical observations., Results: The best run, using in-context learning, achieved 0.58 document-level F1 score on publication abstracts and 0.75 document-level F1 score on clinical observations, as well as a mention-level F1 score of 0.7, which surpasses the current best in class tool. Without in-context learning, however, performance is significantly below the existing approaches., Conclusion: Our experiments show that gpt-4.0 surpasses the state of the art performance if the task is constrained to a subset of the target ontology where there is prior knowledge of the terms that are expected to be matched. While the results are promising, the non-deterministic nature of the outcomes, the high cost and the lack of concordance between different runs using the same prompt and input make the use of these LLMs challenging for this particular task., (© 2024. Crown.)