Seminars

PRESENTER INFORMATION

ABSTRACT: 

Under-resourced populations with chronic inflammatory skin disease, such as psoriasis, experience challenges in access to specialty services, often presenting with more severe skin disease but less likely to receive outpatient dermatologic care. Clinicians are less accurate in diagnosing skin conditions on skin of color. Without accurate assessment of skin disease burden, it is difficult to start adequate treatments and follow response. Furthermore, without accurate documentation of disease severity, advanced treatments that require insurance prior authorization are not accessible to patients. A critical unmet need exists in accurately establishing skin disease burden, especially for patients with richly pigmented skin. Artificial intelligence (AI)-assisted models could address this need to improve the accuracy of skin assessments and knowing which patients need earlier access to dermatology care. However, the performance of AI models depends on the representation and quality of the training data. AI-based models that have been developed to assess psoriasis severity are typically based on proprietary datasets that were rated to have moderate to high levels of bias. Our work aims to develop more accurate and clinically reliable AI-assisted tools for assessing skin rashes across the spectrum of skin tones to improve patient outcomes.