Title:	LLMs Assisted Neurological Disease Diagnosis - Continuation
DNr:	Berzelius-2023-338
Project Type:	LiU Berzelius
Principal Investigator:	Danica Kragic Jensfelt <dani@kth.se>
Affiliation:	Kungliga Tekniska högskolan
Duration:	2023-12-29 – 2024-07-01
Classification:	10201
Keywords:

Abstract

Our project aims to harness the capabilities of large pretrained language models to assist in the diagnosis of neural system diseases. We innovative approach combines fine-tuning techniques and customized prompts to enhance the precision and efficiency of diagnostic processes during medical consultations. The goal of our project is to deploy and fine-tune LLAMA on a server utilizes the strengths of state-of-the-art language models, empowering healthcare professionals to improve their diagnostic accuracy and decision-making. By utilizing pretrained language models, our objective is to aid medical practitioners in recognizing and diagnosing neural system diseases more effectively, leading to better patient outcomes. We plan to train the language model on several comprehensive dataset of medical QA, enabling it to grasp the subtleties of neural system diseases and the diagnostic clues that are critical in this domain. Additionally, we will create specialized prompts that guide the model to generate relevant diagnostic insights based on patient information and clinical data. In summary, our project leverages the power of large pretrained language models to enhance the diagnosis of neural system diseases. Through the combination of fine-tuning and customized prompts, we aspire to provide medical practitioners with a more accurate and efficient diagnostic tool, benefitting both patients and healthcare professionals. Our approach exemplifies the potential of artificial intelligence in transforming medical diagnostics and enhancing patient care, thereby advancing the field of healthcare and natural language processing.