Skin disease diagnosis using deep neural network and large language model

The pretrained Large Language Models (LLMs) have shown remarkable performance in various fields. A potential application is to assist medical diagnosis given proper descriptions of the symptom. To assess the practical applicability of LLMs in healthcare, this thesis explores the realm of skin disea...

Full description

Saved in:
Bibliographic Details
Main Author: Xia, Deneng
Other Authors: Owen Noel Newton Fernando
Format: Thesis-Master by Research
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/172895
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-172895
record_format dspace
spelling sg-ntu-dr.10356-1728952024-01-04T06:32:51Z Skin disease diagnosis using deep neural network and large language model Xia, Deneng Owen Noel Newton Fernando School of Computer Science and Engineering OFernando@ntu.edu.sg Engineering::Computer science and engineering::Computer applications::Life and medical sciences The pretrained Large Language Models (LLMs) have shown remarkable performance in various fields. A potential application is to assist medical diagnosis given proper descriptions of the symptom. To assess the practical applicability of LLMs in healthcare, this thesis explores the realm of skin disease detection, a classic case within AI on medicine. Traditional AI-based methods for skin disease diagnosis rely on image classification models driven by deep networks like ResNet, VGG, DenseNet, etc., which often lack mechanistic understanding and offer single-dimensional functionality. Due to the inherent flexibility of LLMs despite their current instability in reasoning, we decided to combine the strengths of both paradigms. In this thesis, we introduce an interactive chat-based skin disease diagnosis system utilizing a multimodal large language model called VisualGLM and image classification models which are trained on the HAM10000 dataset achieved a validation accuracy of 93\%. This system engages with users in a dialogic manner, explaining the rationale behind the diagnosis while allowing users to contribute additional context during the chat session to enhance the automated diagnostic process. Our work stands as an exploration on the practical application of LLMs in healthcare, demonstrating the untapped potential of LLMs in this crucial field. Master of Engineering 2023-12-28T12:00:27Z 2023-12-28T12:00:27Z 2023 Thesis-Master by Research Xia, D. (2023). Skin disease diagnosis using deep neural network and large language model. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/172895 https://hdl.handle.net/10356/172895 10.32657/10356/172895 en This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Computer science and engineering::Computer applications::Life and medical sciences
spellingShingle Engineering::Computer science and engineering::Computer applications::Life and medical sciences
Xia, Deneng
Skin disease diagnosis using deep neural network and large language model
description The pretrained Large Language Models (LLMs) have shown remarkable performance in various fields. A potential application is to assist medical diagnosis given proper descriptions of the symptom. To assess the practical applicability of LLMs in healthcare, this thesis explores the realm of skin disease detection, a classic case within AI on medicine. Traditional AI-based methods for skin disease diagnosis rely on image classification models driven by deep networks like ResNet, VGG, DenseNet, etc., which often lack mechanistic understanding and offer single-dimensional functionality. Due to the inherent flexibility of LLMs despite their current instability in reasoning, we decided to combine the strengths of both paradigms. In this thesis, we introduce an interactive chat-based skin disease diagnosis system utilizing a multimodal large language model called VisualGLM and image classification models which are trained on the HAM10000 dataset achieved a validation accuracy of 93\%. This system engages with users in a dialogic manner, explaining the rationale behind the diagnosis while allowing users to contribute additional context during the chat session to enhance the automated diagnostic process. Our work stands as an exploration on the practical application of LLMs in healthcare, demonstrating the untapped potential of LLMs in this crucial field.
author2 Owen Noel Newton Fernando
author_facet Owen Noel Newton Fernando
Xia, Deneng
format Thesis-Master by Research
author Xia, Deneng
author_sort Xia, Deneng
title Skin disease diagnosis using deep neural network and large language model
title_short Skin disease diagnosis using deep neural network and large language model
title_full Skin disease diagnosis using deep neural network and large language model
title_fullStr Skin disease diagnosis using deep neural network and large language model
title_full_unstemmed Skin disease diagnosis using deep neural network and large language model
title_sort skin disease diagnosis using deep neural network and large language model
publisher Nanyang Technological University
publishDate 2023
url https://hdl.handle.net/10356/172895
_version_ 1787590741846917120