New AI Tool Predicts Alzheimer’s With Higher Accuracy Than Clinical Tests : ScienceAlert

While we remain skeptical of artificial intelligence’s storytelling and filmmaking abilities, it is proving to have genuinely useful applications in science. As a new study shows, AI can even perform better than clinical tests at predicting the progress of Alzheimer’s disease.

That could mean people showing signs of the early stages of dementia can be better informed about the risk of the condition progressing, and treatments and precautions can be put in place sooner, if necessary.

The research was led by a team from the University of Cambridge in the UK, and used a machine learning approach to train AI algorithms on cognitive ability tests and brain scans from 410 individuals – finding patterns matching cognition with levels of gray matter, which helps the brain process information.

“We’ve created a tool which, despite using only data from cognitive tests and MRI scans, is much more sensitive than current approaches at predicting whether someone will progress from mild symptoms to Alzheimer’s – and if so, whether this progress will be fast or slow,” says senior author and cognitive computational neuroscientist Zoe Kourtzi, from the University of Cambridge.

When the AI was tested on 1,486 cases outside of the training data, it was able to correctly identify people who would go on to develop Alzheimer’s within three years 82 percent of the time, and those who wouldn’t develop the disease 81 percent of the time.

That’s about three times better than current clinical assessments and could make a huge difference in terms of Alzheimer’s diagnosis. The AI was also able to identify how quickly the dementia would progress in many cases.

This could help doctors identify those who would be most eligible for new treatments, as well as enable further study of Alzheimer’s in its earliest stages, which is vital to finally figuring out how the disease gets started.

“AI models are only as good as the data they are trained on,” says Kourtzi. “To make sure ours has the potential to be adopted in a healthcare setting, we trained and tested it on routinely-collected data not just from research cohorts, but from patients in actual memory clinics. This shows it will be generalizable to a real-world setting.”

The new approach has a lot in its favor: It’s relatively cheap, for example, and doesn’t require any invasive procedures in terms of tissue or blood collection. With healthcare resources being continually stretched, that’s important.

And highlighting those at a low risk of developing Alzheimer’s is useful too – offering peace of mind for those who are experiencing memory problems as they get older, and are worried that dementia might be setting in.

“The fact that we might be able to reduce this uncertainty with information we already have is exciting and is likely to become even more important as new treatments emerge,” says psychiatrist Ben Underwood, from the University of Cambridge.

The research has been published in eClinicalMedicine.