Residents of Oklahoma who are interested in the growing role that artificial intelligence plays in the medical field will want to know about a new international study published in the Annals of Oncology. Researchers tested a form of deep learning called convolutional neural networks (CNN) and found that AI can diagnose skin cancer more precisely than experienced dermatologists.
The test began with researchers showing the CNN more than 100,000 images (captured through dermoscopy) of benign and malignant skin cancers and moles together with their diagnoses. This trained the network to distinguish between the two types. Afterward, a new set of 300 images followed by 100 pictures were shown. The first set tested the network's diagnostic ability, and the second tested both the CNN and a team of 58 doctors from 17 different countries.
The CNN missed fewer cases of skin cancer than the team of doctors. The latter correctly diagnosed 86.6 percent of melanomas, the most fatal form of skin cancer, while the network correctly diagnosed 95 percent. With the CNN, fewer benign moles were misdiagnosed as melanoma.
The network continued to outperform the doctors even when they were given more clinical information and images. The results of the study present no cause for alarm, though. Researchers assert that the CNN will help reduce misdiagnoses but will not replace humans in this critical line of work.
These findings are good news because misdiagnoses are an all-too-common form of medical malpractice. When victims suffer from the worsening of a condition and believe that it could have been prevented, they can speak with a lawyer who focuses on malpractice claims. An attorney could ask to meet with the local medical board. He or she may even bring in third-party medical experts to determine the extent of the victim's injuries. A lawyer could then negotiate for a settlement covering physical and emotional trauma, lost wages and more.