Artificial Intelligence Use in Acne Diagnosis and Management-A Scoping Review.

Area of Science:

• Dermatology and Artificial Intelligence (AI)
• Medical Informatics
• Health Equity

Background:

• AI offers potential for early acne diagnosis and treatment.
• Bias in AI model training poses challenges to health equity in clinical practice.
• Existing research needs updated overview on AI applications in acne care.

Purpose of the Study:

• To review AI-based tools for acne diagnosis and management.
• To assess the performance of AI tools in acne care.
• To evaluate the reporting of skin diversity in AI model training data.

Main Methods:

• Systematic literature search of PubMed, Cochrane, and Scopus databases.
• Keywords included: acne, artificial intelligence, machine learning, deep learning, large language model, ChatGPT.
• Analysis of 105 research articles focusing on AI applications in acne.

Main Results:

• Most studies (96.2%) focused on acne diagnosis; few on management (9.5%).
• Image-based models, particularly deep learning (76.2%), were predominant.
• Ensemble models showed the highest accuracy (89.7%), followed by deep learning (88.5%).
• Only 13% of studies reported patient skin color data, with few including diverse skin tones.

Conclusions:

• AI, especially ensemble models, can aid remote acne diagnosis and management.
• Significant gaps exist in reporting skin diversity in AI training datasets.
• Developing guidelines for diverse data representation is essential for healthcare social justice.