At the University of Missouri, a research team led by Kamlendra Singh tested artificial intelligence as a decision-support tool to detect melanoma from images of suspicious skin abnormalities. The effort is intended to speed identification of cases that may require closer medical attention and earlier treatment, not to replace doctors.
Singh and colleagues trained and evaluated AI models using a database of 400,000 images collected with three-dimensional total body photography, a high-resolution method that produces a digital map of the whole skin surface and helps researchers analyse subtle visual details across the body. The team compared three existing AI models and reported that each model reached up to 88% accuracy on its own.
When the researchers combined the three models, performance improved, with accuracy exceeding 92%. They note that training on larger, more diverse datasets — including images that represent different skin tones, lighting conditions and camera angles — should further improve predictions. Singh cautioned that clinical use will take time, and he said better explanations of how AI reaches conclusions will help health professionals trust and adopt the technology.
The study appears in Biosensors and Bioelectronics: X. Source: University of Missouri.
Difficult words
- decision-support tool — software that helps make clinical decisions
- melanoma — a dangerous skin cancer that can spread
- database — organized collection of stored digital information
- three-dimensional — showing depth as well as height and width
- high-resolution — image quality with very clear, fine visual detail
- accuracy — the degree to which results are correct
- dataset — a group of related data used for analysisdatasets
Tip: hover, focus or tap highlighted words in the article to see quick definitions while you read or listen.
Discussion questions
- What advantages and risks do you see in using AI as a decision-support tool for detecting melanoma? Give reasons.
- How important is it to include different skin tones, lighting conditions and camera angles in training datasets? Explain with examples.
- Do you think combining multiple AI models is a good strategy for medical diagnosis? Why or why not?
Related articles
Cyber risks to US infrastructure as strikes on Iran continue
As US strikes on Iran continue, experts warn of possible retaliatory cyberattacks on essential US infrastructure. Alex K. Jones reviewed the risks, naming water systems, power grids and the possible future effect of quantum computing.
Plague and the role of human fleas in Madagascar
The plague is still endemic in some countries, including Madagascar. Researchers found that the human flea Pulex irritans contributes to outbreaks and that household conditions and insecticide use influence flea levels and control efforts.
