How a Saudi university is using AI to change the diagnosis and treatment of skin diseases
RIYADH: To improve the efficiency and effectiveness of dermatological care, experts at King Abdullah University of Science and Technology in Saudi Arabia have developed a groundbreaking new diagnostic system called SkinGPT-4 that harnesses the power of artificial intelligence.
Xin Gao, professor of computer science, co-chair of the Center of Excellence on Smart Health and chair of the Bioinformatics Platform at KAUST, is leading the research. He says the goal of SkinGPT-4 is to detect, diagnose and find appropriate treatments for skin diseases.
The technology was developed in collaboration with Juexiao Zhou, a PhD student at KAUST who is also the first author of SkinGPT-4. Gao says it could be potentially life-saving for patients, especially those in rural areas where there is often a shortage of trained dermatologists.
“These specific challenges in dermatology led to the development of SkinGPT-4,” Gao told Arab News. “The variability of skin appearance and the need for expertise to properly identify and treat these conditions highlighted the need for an advanced, AI-driven solution.”
The team recognized the need for such a solution after recognizing the limitations of traditional diagnostic methods and the potential of AI, particularly large language models (LLMs) such as the eponymous ChatGPT, in improving the accuracy and efficiency of dermatological diagnoses.
“With SkinGPT-4, users can upload their own skin photos for diagnosis, and SkinGPT-4 can independently determine the characteristics and categories of skin diseases, perform analysis, provide treatment recommendations, and enable interactive diagnosis,” said Gao.
Gao and the KAUST team behind the AI dermatologist SkinGPT-4. (Supplied)
SkinGPT-4 diagnoses diseases with distinct visual features such as acne, rosacea, melanoma, psoriasis, basal cell carcinoma, eczema and many more.
Gao said the development of SkinGPT-4 began with data collection and preprocessing, followed by model training and validation. “The team collected a large dataset of dermatological images and patient records to train the AI model,” he said.
“One of the biggest challenges was integrating different data types, including images and text, which required collaboration between computer scientists and dermatologists. The multidisciplinary team worked together to ensure that the AI could effectively interpret and analyze images of skin diseases.”
DO YOU HAVEKNOWLEDGE?
• SkinGPT-4 diagnoses diseases with distinct visual features, such as melanoma, psoriasis and eczema.
• It uses a combination of computer vision algorithms, large language models and natural language processing.
• The technology could help doctors and patients in rural areas where there is often a shortage of trained dermatologists.
SkinGPT-4 uses a combination of computer vision algorithms, LLMs and natural language processing (NLP) that enables programs to understand human languages.
“The model processes dermatological images using a Vision Transformer (ViT) to detect patterns and features that indicate different skin diseases,” said Gao.
“The ViT is matched with an LLM called Llama-2-13b-chat on our dataset with a customized two-stage training strategy. In this way, the LLM Llama-2-13b-chat can understand the images of skin diseases and enable conversational diagnosis with the patient in natural language.”
This section contains relevant reference points, placed in the (opinion field)
SkinGPT-4 could be particularly useful in diagnosing rare skin diseases that general practitioners may not easily recognize.
“For a patient with an unusual rash, SkinGPT-4, which has been trained on a large number of dermatological images, including rare diseases, can provide a rapid and accurate diagnosis,” Gao said.
“In addition, when treating chronic skin diseases such as psoriasis, SkinGPT-4 can monitor the course and response to treatment, provide ongoing support and adjust treatment plans as needed.”
Researchers hope SkinGPT-4 will bring about a breakthrough in remote or underserved areas with a shortage of dermatologists.
“For example, in a rural community where the nearest dermatologist is hundreds of miles away, a patient comes to us with a suspicious lesion that could be a rare form of skin cancer,” Gao said.
Properly trained, AI can be a huge help to doctors. (Shutterstock image)
“With SkinGPT-4, a local healthcare provider can take a high-resolution image of the lesion and enter the patient’s medical history into the system. SkinGPT-4 analyzes the image and patient information and quickly provides a preliminary diagnosis and recommendations for further action.”
And as SkinGPT-4 continues to evolve, Gao says the system will learn from its own mistakes through continuous learning and feedback mechanisms.
“By analyzing misdiagnoses and incorporating corrections, the system can refine its algorithms and improve its accuracy over time,” he said. “This iterative learning process ensures that SkinGPT-4 continues to evolve and adapt to new data and emerging trends in dermatology.”
However, Gao emphasizes that SkinGPT-4 is not intended to replace dermatologists entirely. Rather, the program is designed as an evolving and optimizing tool that acts as an assistant in facilitating communication between patients and doctors.
“Our aim with SkinGPT-4 is to give patients more information about skin diseases and at the same time offer doctors valuable help in the diagnostic process.”
