New AI can diagnose Alzheimer’s 10 years before human doctors

Saturday, September 23, 2017 by

Researchers from Italy have developed a machine-learning algorithm that can detect signs of Alzheimer’s disease almost 10 years before human doctors can.

According to the DailyMail.co.uk, the team from the University of Bari Aldo Moro trained their artificial intelligence (AI) by using 67 MRI scans: 29 came from a healthy control group, while 38 were from patients with Alzheimer’s disease. Each scan was divided into smaller regions, which were then analyzed by the AI for neuronal activity. Through this, they were able to teach the AI to tell the difference between healthy brains and those who were at risk of the disease. Moreover, they found that the efficacy of the algorithm peaked at the analyses of regions in the 2,250 to 3,200 cubic mm range — which, coincidentally, are the sizes of the hippocampus and amygdala, the structures of the brain linked to Alzheimer’s disease.

The researchers then tested the AI on the brain scans of 148 subjects, 52 of whom were healthy, 48 who had Alzheimer’s disease, and 48 with mild cognitive impairments. The algorithm successfully differentiated the healthy brains from those with Alzheimer’s disease with 86 percent accuracy; moreover, it was able to single out the brains with mild cognitive impairments 84 percent of the time.

Study co-author Marianna La Rocca said of current detection procedures: “Nowadays, cerebrospinal fluid analyses and brain imaging using radioactive tracers can tell us to what extent the brain is covered with plaques and tangles, and are able to predict relatively accurately who is at high risk of developing Alzheimer’s 10 years later. However, these methods are very invasive, expensive and only available at highly specialized centers.”

In addition to being accurate, their AI is simpler, more affordable, and noninvasive. They further wrote in their paper: “Also, the method is very suitable to application to longitudinal studies, ideally in association with functional imaging, to improve our understanding of the different patterns of neurodegeneration in different diseases.”

Although promising, there are still many questions surrounding the AI, the most pressing of which is whether or not it can be protected from abuse. By this, we mean if it can be programmed in such a way that it won’t be taken advantage of by outside sources to hand out misdiagnoses and put innocent, healthy people through the wringer the same way the cancer industry does.

In fact, a 2014 study unearthed a few shocking and unsettling statistics that should definitely give you pause for thought. After poring through medical records and doctor’s clinics visits, the researchers discovered that, in the U.S. alone, at least one in 20 adults are misdiagnosed by their doctors annually. Hike it up to the national level and that number swells to 12 million people yearly. That’s 12 million people who are either diagnosed to be clean of particular diseases, or diagnosed with afflictions that they didn’t even have in the first place.

And those solidly belonging to the latter are those who have to put down their own money or take out loans for treatments they never needed. More than that, these people will be scared out of their minds, wondering if every medical procedure they’re undergoing and medication they’re taking is going to help “cure” them. And that’s not even going into the mountains of paperwork or managed-care issues. (Related: Misdiagnosis can kill: Young woman suffers from agonizing pain after 7 years of unnecessary chemotherapy.)

Will the AI be able to avoid a problem like this? Will it be protected against manipulation to diagnose everyone with Alzheimer’s disease?

Visit HealthCoverage.news for more stories that are similar to this one.

Sources include:

DailyMail.co.uk
Arxiv.org
LiveScience.com



Comments

comments powered by Disqus