Using advanced artificial intelligence algorithms, scientists are hoping to identify brain wave patterns associated with the risk of dementia. Photo / 123RF
Using advanced artificial intelligence algorithms, scientists are hoping to identify brain wave patterns associated with the risk of dementia. Photo / 123RF
Using advanced artificial intelligence algorithms, scientists are hoping to identify brain wave patterns associated with the risk of dementia.
Imagine a sleek, portable home device that resembles a headband or cap, embedded with tiny electrodes. Placed on the head, these sensors detect subtle brain wave activity, behaving like a pulse-detecting smartwatch, a blood pressure wrist cuff or a heart rate monitor.
But this tool isn’t checking your heartbeat. Using advanced artificial intelligence algorithms to analyse data in real-time, a device like this could look for signs of Alzheimer’s disease years before symptoms become apparent. Such a monitor is not yet available, but AI could make it a reality.
“The readout could be as simple as a traffic light system - green for healthy activity, yellow for something to keep an eye on and red for when it’s time to consult a health care professional,” said David T. Jones, who directs the Neurology AI Programme at the Mayo Clinic. “You would be able to monitor your brain health the same way you now can monitor your heart rate and blood pressure. We’re not there yet, but that is the future.”
It could be a decade or longer before such technology is in widespread use, but the science is “moving quickly”, said Jones.
Mayo’s brain waves research is just one way scientists are working to harness the power of artificial intelligence to pinpoint early indicators of cognitive impairment. Scientists are using AI to study blood biomarkers - several are linked to Alzheimer’s disease. And AI is helping them search for data that can connect dementia to such chronic health conditions as inflammation, certain vision problems, high cholesterol, hypertension, diabetes and osteoporosis.
AI makes these efforts possible because it can analyse massive amounts of complicated data from electronic patient health records with enormous speed, and often with the ability to detect nuances imperceptible to humans.
“We want to find ways to detect dementia as early as possible,” said Jennie Larkin, deputy director of the Division of Neuroscience at the National Institute on Ageing. “AI is primarily helpful in understanding and managing big data too large or complex for traditional analyses. Its potential is to be an incredible assistant in helping us understand rich medical data and identify possibilities we never could unassisted.”
AI already is in use in other healthcare settings, including screening mammograms, and researchers are excited about its potential contributions to brain health.
“AI should accelerate our ability to predict an increase in risk for chronic diseases,” said Judy Potashkin, professor and discipline chairwoman of cellular and molecular pharmacology at the Chicago Medical School Centre for Neurodegenerative Disease and Therapeutics.
Alzheimer’s disease is the most common form of dementia, afflicting an estimated 5.8 million Americans older than 65 in 2020, according to the Centres for Disease Control and Prevention. The number is expected to nearly triple to 14 million by 2060. The disease is marked by progressive memory loss, personality changes, and ultimately the inability to perform routine daily tasks, such as bathing, dressing and paying bills.
Some people are nervous about the growing use of AI, fearing it will replace the work of humans. But experts insist it only will enhance it.
“AI is high-powered and has many databases to search, and can do so with incredible speed,” said Arthur Caplan, professor of bioethics at NYU Langone Health. “Humans get tired. AI does not.”
The number of New Zealanders with dementia is expected to double by 2040. Photo / 123RF
AI also has the potential to bridge the gap in expertise between seasoned clinicians and less experienced providers. For example, AI could recognise subtle signs, such as changes in a patient’s voice, that could help diagnose neurological disorders such as Parkinson’s disease, Alzheimer’s or amyotrophic lateral sclerosis (ALS). “Much of what experts do involves recognising patterns from training and experience, something AI can help non-experts replicate,” said Jones.
In the brain waves research that Jones believes eventually could result in home-based monitors, Mayo scientists used AI to scan electroencephalograms (EEGs) for abnormal patterns that are characteristic of patients with cognitive problems such as Alzheimer’s disease.
They studied data from more than 11,000 patients who received EEGs at the Mayo Clinic, identifying specific differences, including changes in brain waves in the front and back of the brain.
“Humans cannot see them, but machines can,” Jones said. The hope is that some day clinicians will use AI to catch these patterns early, before memory problems become apparent.
A team at Massachusetts General Hospital used AI and magnetic resonance imaging (MRI) to develop an algorithm to detect Alzheimer’s. They trained the model using nearly 38,000 brain images from about 2300 patients with Alzheimer’s and about 8400 who didn’t have the disease.
They then tested the model across five datasets of images to see whether it could accurately identify Alzheimer’s. It did so with 90.2% accuracy, said Matthew Leming, a research fellow in radiology at the hospital’s Center for Systems Biology and one of the study authors.
One challenge in interpreting the MRI data for future research is “people only come in to get MRI scans when they have symptoms of something else”, which could confound the results. “If a person comes into a hospital for an MRI, it’s not usually because they are healthy,” he said.
At the University of California at San Francisco, researchers used AI to design an algorithm to determine whether having certain health conditions could predict who might develop the disease in the future. The conditions included hypertension, high cholesterol and vitamin D deficiency in both men and women, erectile dysfunction and an enlarged prostate in men, and osteoporosis in women.
They designed the model using a clinical database of more than five million people both with and without Alzheimer’s. In a separate group of non-Alzheimer’s patients, the algorithm predicted with 72% accuracy those who would eventually receive an Alzheimer’s diagnosis within seven years.
The research raises the hopeful prospect preventing and treating these conditions might help protect against eventual dementia, said Alice Tang, one of the study authors.
The association of these conditions to Alzheimer’s “was stronger than that among people who did not have any of these other health issues”, said Tang, a bioengineer and medical student. However, she said it’s important to remember “not everyone who has Alzheimer’s has these conditions and not everyone who has these conditions will develop Alzheimer’s. It’s just a red flag. One predictive tool that needs further study”.
Some experts urge caution, emphasising much of the work with AI is still preliminary. “We don’t necessarily have enough data to see if any of these tools have been validated to predict someone’s risk,” said Rebecca Edelmayer, vice-president of scientific engagement for the Alzheimer’s Association.
“Overall, AI in this case is a good thing. But it carries a big ‘but'." Photo / 123RF
Today, Alzheimer’s and other forms of dementia usually are diagnosed only once symptoms appear. There are several drugs that might slow it down, although they don’t work for everyone, and their efficacy can wane over time.
The potential of AI to enable early diagnosis raises many of the same issues that greeted the early use of genetic testing.
“Overall, AI in this case is a good thing,” Caplan said. “But it carries a big ‘but,’” including the potential for health insurance and employer discrimination, he said. But the biggest questions, he added, are: Will people want to know? And if so, what will they do with that information?
“To be honest, I would do nothing,” said Joel Shurkin, a retired science writer from Baltimore whose wife, marine biologist Carol Howard, suffered from early-onset Alzheimer’s and died in 2019 at 70. “Except for a few meds, there is nothing to be done,” he said.
Kathleen, 76, from Bethesda, Maryland (using only her first name to protect her privacy), lost her 82-year-old husband in April to Alzheimer’s complications. His mother and older sister also had died of the disease, so the couple were not surprised when he was diagnosed in his mid-70s.
“We already were living with the risk and had our affairs in order,” she said. Knowing in advance “foretells a long, slow death with devastating psychological and financial consequences”, she said.
One of their daughters, now in her 40s, enrolled in research monitoring her brain health with the hope of catching it early. Kathleen believes AI research ultimately will make a dramatic difference in early diagnosis and treatment. “I think it will be miraculous,” she said.
Caplan said there are some advantages to knowing that dementia looms in your future.
“You can plan your life,” he said. “Take that vacation next year instead of waiting. Get your affairs in order. Discuss it so everybody will be ready, which is of great value to others.”
NIA’s Larkin noted finding the disease sooner “may provide opportunities for new treatments”.
“It’s very hopeful how much we are learning,” she said.
Caplan agrees. “By the time you are unable to speak and walk, it’s very hard to repair the brain,” he said. “Early detection raises the hope you will be able to try new interventions before the damage occurs. I’m not saying this will happen, but the potential of AI certainly opens the door.”
