AI Model Accurately Scores Psoriasis Severity from Clinical Images

KEY TAKEAWAYS

• The YOLOv8 deep learning model successfully classifies psoriasis lesion severity based on clinical image features, a new study showed.

• Stratified k-fold cross-validation ensured accurate performance across diverse datasets.

• AI-driven PASI scoring could reduce subjectivity and variability in clinical assessments, the authors wrote.

A new study demonstrates the potential of artificial intelligence (AI) to improve the consistency and objectivity of psoriasis severity assessments based on 2D clinical images.

Researchers for the study used the YOLOv8 deep learning model to classify psoriatic lesions according to erythema, thickness, and scaling. Their model categorized lesion severity [none (0), mild (1), moderate (2), severe (3), or very severe (4)]. The team conducted training using three image datasets in a cloud-based Google Colab environment and used stratified k-fold cross-validation to ensure robustness.

According to their results, the YOLOv8 model demonstrated high accuracy in classifying lesion severity, with confusion matrices confirming performance consistency across datasets. The integration of stratified cross-validation further enhanced reliability, they wrote.

“This study represents a significant advancement in the application of AI to the automated classification of lesion severity based on erythema, thickness, and scaling—key subcomponents of PASI,” the authors said in their abstract.

Source: Chou C, et al. Dermatology. 2025. Doi:10.1159/000549640