In medical diagnostics, artificial intelligence (AI) is already widely employed. In a realistic clinical environment, a research team studied how it benefits the diagnosis and management of pigmented skin lesions. The team evaluated the accuracy of two separate algorithms in smartphone applications in diagnosis and therapy suggestions to that of doctors in a study. The results suggest that the AI application works well in diagnosing in general. Doctors, on the other hand, were superior when it came to treatment judgments.
The AI application was evaluated in two skin cancer hospitals under realistic clinical situations by the study team. The prospective study had two scenarios, with Artificial Intelligence applied in scenario A for alterations suggestive of skin cancer and in scenario B for individuals with many moles. In both situations, the AI-assisted application was compared to medical professionals and less experienced clinicians.
The study analyzed 172 pigmented lesions in 124 patients in two scenarios, using a 7-class AI algorithm and an ISIC algorithm. The 7-class algorithm outperformed less experienced clinicians, while the ISIC algorithm performed better than inexperienced users.
In terms of treatment decisions, the 7-class algorithm performed much worse than specialists but performed better than inexperienced users. The findings imply that an AI-assisted smartphone application for skin cancer diagnostics performs as well as specialists in a real-world clinical setting. However, when it came to treatment judgments, the professionals outperformed the AI. In the therapy recommendation, the AI application removes more benign lesions than specialists would. Taking this into consideration, the AI application can undoubtedly be deployed. It should also be noted that if it is utilized indiscriminately, too many false-positive results must be clarified.
Related Content: Plasmonic Plastic For Optical Sensor Production