A deep-learning model developed by using over 1,300 dermatoscopic images of scalp psoriasis and seborrheic dermatitis from over 600 patients exhibited good sensitivity (96.1%), specificity (88.2%), and area under the curve (0.922) in terms of determining which condition was which, according to a paper published by Yu et al in Frontiers in Medicine. The deep-learning model also outperformed all five dermatologists (who had various levels of experience) participating in the study in terms of diagnosing the conditions; health-care providers with less experience in diagnosis, such as graduate students and general practitioners, used the deep-learning model as an aid in diagnosis and improved their rates of correct diagnoses when incorporating the model.
May 24, 2023