The accuracy of the preliminary diagnosis of malignant melanocytic skin tumors using the artificial intelligence program “Melanoma Check”

Abstract

Objective. To investigate the accuracy of preliminary diagnosis of malignant melanocytic skin tumors at the initial medical appointment using the artificial intelligence (AI) computer program (smartphone application) “Melanoma Check”. Material and methods. The effectiveness of the AI program “Melanoma Check” for visual identification of pigmented malignant and benign skin tumors, intended for physicians providing primary health care, was assessed in 133 patients aged 19 to 90 years who had various melanocytic skin tumors. The conclusion of the program was compared with the preliminary and final clinical diagnosis. Results. In the diagnosis of various melanocytic skin tumors, the program “Melanoma Check” showed high diagnostic accuracy. The efficiency of the AI program “Melanoma Check” (diagnostic accuracy 89.5%, sensitivity 92.5%, specificity 88.2%) is almost comparable to the efficiency of experienced dermatovenereologists (diagnostic accuracy 95.5%, sensitivity 95.0%, specificity 100%). Conclusion. The AI program for the smartphone “Melanoma Check” demonstrates comparable indicators of the effectiveness of the diagnosis of malignant melanocytic skin tumors compared with the results of clinical and dermatoscopic conclusions of experienced dermatovenerologists and the results of morphological studies. The use of this program makes it possible to detect skin melanoma early during initial admission by general practitioners and internists.

Ähnliche Arbeiten