Diagnosis by artificial intelligence
Artificial intelligence (AI) can detect skin cancer more accurately than experienced dermatologists, new research shows. Scientists in the United States, Germany, and France taught a machine learning system known as a deep-learning convolutional neural network (CNN) to identify skin cancer, by showing it more than 100,000 images of malignant melanomas and harmless moles. When pitted against the skills and experience of 58 dermatologists from 17 different countries, the machine came out on top. The CNN correctly identified 95 percent of melanomas out of a total of 300 images. In a similar test, the doctors picked up 87 percent of melanomas. The humans’ accuracy improved slightly once they were given additional photos and patient information, but even doctors with many years of experience were still bested by the CNN, which missed fewer melanomas and misdiagnosed fewer benign moles, reports ScienceDaily.com. It’s still too early to use AI in standard clinical practice, doctors who reviewed the study said, but sooner rather than later, machines will be making diagnoses.