Artificial Intelligence for early image diagnosis of SJS/TEN
Published: September 18, 202
Stevens–Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN) is a life-threatening cutaneous adverse drug reaction (cADR). However, it is difficult to distinguish SJS/TEN from non-severe cADRS, especially in the early stages of the diseases.
In an original article in The Journal of Allergy and Clinical Immunology: In Practice, Fujimoto et al. trained a deep convolutional neural network (DCNN) using a dataset of 26,661 individual lesion images obtained from 123 patients diagnosed with SJS/TEN or non-severe cADRs. The DCNN’s classification accuracy was compared to that of 10 board-certified dermatologists and 24 trainee dermatologists.
The DCNN achieved 84.6% sensitivity, whereas the sensitivities of the board-certified dermatologists and the trainee dermatologists were 31.3 % and 27.8%, respectively. The negative predictive value was 94.6% for the DCNN, 68.1 % for the board-certified dermatologists, and 67.4% for the trainee dermatologists. The area under the receiver operating characteristic curve of the DCNN for SJS/TEN diagnosis was 0.873, which was significantly higher than that of all the board-certified dermatologists and trainee dermatologists.
In conclusion, the DCNN demonstrated superior performance in screening for SJS/TEN compared with dermatologists. Along with clinical information, the DCNN may, as a diagnostic test, provide useful supportive data for a comprehensive diagnosis.
The Journal of Allergy and Clinical Immunology: In Practice is an official journal of the AAAAI, focusing on practical information for the practicing clinician.