Publication Date: 2024/02/12
Abstract: This study investigates the vital role of highlevel language models (LLMs) in advancing the medical field and underscores the need for incorporating these models into a biomedical chatbot framework. The application of large language models (LLM) in medicine is both hopeful and concerning, as it can provide answers with some degree of autonomy. The main objective is to enhance the availability of crucial medical information and streamline the extraction of relevant data. By utilizing cutting-edge LLMs such as GPT 3.5 Turbo, PaLM2, Llama2 and Mistral 7B, biomedical chatbots are emerging as a robust platform that enables access to essential medical information. This study showcases the efficacy of integrating GPT 3.5 Turbo, PaLM2, and Gemini Pro into a biomedical chatbot framework, exhibiting their accuracy and response time. The GPT 3.5 Turbo displayed high accuracy and a swift response time of 2 seconds, making the team favour PaLM2 as the second choice, but the difference was not significant and they were mostly impressed by the performance. Moreover, performance tests reveal that PaLM2 has an average response time of around 6.50 seconds, while Gemini Pro has an average response time of around 8.68 seconds.
Keywords: Large Language Models, Chatbot, Natural Language Processing, Prompts, Accuracy.
DOI: https://doi.org/10.5281/zenodo.10649891
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT24JAN1870.pdf
REFERENCES