OrphaGPT: An Adapted Large Language Model for Orphan Diseases Classification

Kushal Pokhrel; Cesar Sanin; Md Rafiqul Islam; Edward Szczerbicki

doi:10.1007/978-981-97-4982-9_16

OrphaGPT: An Adapted Large Language Model for Orphan Diseases Classification

Abstrakt

Orphan diseases (OD) represent a category of rare conditions that affect only a relatively small number of individuals. These conditions are often neglected in research due to the challenges posed by their scarcity, making medical advancements difficult. Then, the ever-evolving medical research and diagnosis landscape calls for more attention and innovative approaches to address the complex challenges of rare diseases and OD. Pre-trained LLMs are a crucial component of contemporary artificial intelligence (AI), contributing to significant advancements in the performance of complex AI tasks. In this research, we aim to introduce a novel model that leverages the capabilities of a fine-tuned GPT-3.5 Turbo model with reasonable accuracy. We design a comprehensive, customized user interface named OrphaGPT, an interactive GPT chat that allows users to engage in deeper conversations about ODs. Our model achieves an 80% accuracy rate, attained through an exploration of Natural Language Processing (NLP), and domain-specific fine-tuning and fine-prompting. Our findings provide valuable insights into the new perspectives of prompting as a way of fine-tuning LLMs while customizing them to specialised domains. This showcases the potential for adaptive generative AI to play a pivotal role in the specific field of OD. The implications of this research extend to medical practitioners, researchers, and the OD community, offering new interactive ways to understand, identify, and diagnose such complex diseases through the customized advanced language model. The successful customization of LLMs into specific fields signifies an advancement of AI, contextualising dialogues and presenting implications for future advances.