Psoriasis is a chronic inflammatory skin condition affecting millions of people globally, with prevalence varying significantly between countries. Conventional treatments, including topical agents, phototherapy, and systemic medications, often fail to account for individual variability, leading to suboptimal outcomes and potential adverse effects. Artificial intelligence (AI) has emerged as a promising approach to enhance precision and personalization in psoriasis management, potentially transforming diagnostic accuracy and treatment selection. This review examines the integration of AI across multiple domains of psoriasis treatment: (1) machine learning algorithms for phototherapy outcome prediction, (2) deep learning techniques for lesion segmentation and severity assessment, (3) AI-enhanced remote photographic monitoring systems, and (4) predictive modeling for response to systemic therapies and biologics. The analysis encompasses various AI methodologies, including random forest classifiers, convolutional neural networks, multiscale superpixel clustering, and gradient-boosted decision trees applied to clinical datasets, imaging analysis, and multi-omic patient data. AI-driven models demonstrate significant clinical utility with phototherapy outcome prediction, achieving high sensitivity (>84%) and accuracy (75-85%). Automated lesion segmentation reaches 86.99%-pixel accuracy, while remote AI assessments strongly correlate with clinical evaluations (Intraclass Correlation Coefficient ICC = 0.78-0.99). Notably, predictive models can forecast biologic therapy responses with > 95% accuracy within 2-4 weeks of treatment initiation, substantially reducing evaluation timelines from the conventional 12-week assessment period. AI technologies offer transformative potential in psoriasis management by enabling precise diagnosis, outcome prediction, and personalized therapy selection. Current implementations show promising results across diverse clinical applications, from phototherapy optimization to biologic response prediction. While challenges in dataset diversity, standardization, and validation remain, these represent opportunities for further advancement toward precision medicine in dermatology.
Prithiviraj et al. (Mon,) studied this question.