Artificial intelligence, which is a kind of computer technology, is being used a lot in medicine to help diagnose illnesses.
A team of researchers from Austria and Australia, led by a skin doctor named Harald Kittler from Medical University of Vienna. Looked into how helpful AI is for diagnosing and treating skin problems with dark spots in real-life medical situations.
AI in Skin Cancer Diagnosis
In a study published, they tested two different computer programs on smartphones that use AI to see how well they can diagnose skin issues compared to human doctors. They found that the AI program is pretty good at diagnosing problems. Doctors were much better at deciding how to treat them.
The research team conducted a study in two types of skin cancer centers, one at the University Department of Dermatology in Vienna. The other at the Sydney Melanoma Diagnostic Center in Australia. They did this study to see how well an AI application works in real medical situations.
They tested it in two different situations:
One for finding skin changes that might be cancer, and the other for people with lots of moles. They compared the AI application’s performance to both experienced doctors and less experienced ones.
In the first situation, they looked at 172 suspicious dark spots (and found that 84 of them were cancerous) in 124 patients. In the second situation, they checked 5,696 dark spots (and discovered that 18 were cancerous) in 66 patients.
They used two different AI system programs on smartphones:
One was a new 7-class AI algorithm, and the other was an ISIC algorithm that had been used in previous studies.
In the second situation, the 7-class AI algorithm was just as good as the experienced doctors and better than the less experienced ones. However, the ISIC algorithm didn’t perform as well as the experts but did better than the less experienced users.
Taking a Close Look at AI Decisions
When it comes to deciding how to treat patients, the 7-class algorithm was not as good as the experts but was better than the less experienced users. The results suggest that using an AI-supported smartphone app for diagnosing skin cancer is as good as experienced doctors in a real medical setting when it comes to making accurate diagnoses.
However, doctors are better than AI when it comes to deciding on treatments.
Dr. Kittler mentioned, “The AI application tends to remove more benign lesions in the treatment recommendation than experts would. If you take this into account, the AI application can certainly be used. It should also be borne in mind that if it is used uncritically, too many false-positive findings would have to be clarified.”
The integration of AI technology in medical diagnostics, particularly for skin issues, has shown promise. Researchers, led by Dr. Harald Kittler, compared AI’s diagnostic capabilities to those of doctors. The results indicate that AI can excel in diagnosis but lags behind in treatment decisions. Careful, critical usage is crucial to avoid unnecessary interventions.