Impact of real-world confounders on the accuracy of an AI model to support read out of skin prick automated test results.

Skin prick testing (SPT) is the gold standard for diagnosing allergic sensitization in individuals with a suspected airborne allergy (1). Skin tests are used as first option in 90% of individuals suffering from respiratory allergies and almost two-third of all types of allergies (2). However, its accuracy highly depends on the operator, causing variability during pricking and readout. S.P.A.T. or skin prick automated test standardises the SPT procedure and has been clinically validated (3).