In a new frontier for deep brain stimulation, researchers used AI to develop individualised algorithms, which helped a skateboarder and other patients with Parkinson’s disease.
When Shawn Connolly was diagnosed with Parkinson’s disease nine years
Shawn Connolly outside the San Francisco Skate Club. After he developed Parkinson’s, his balance and alignment seemed off and his right hand started cupping. “I was noticing that things weren’t really working right,” he recalled. Photo / Jason Henry, The New York Times
In a new frontier for deep brain stimulation, researchers used AI to develop individualised algorithms, which helped a skateboarder and other patients with Parkinson’s disease.
When Shawn Connolly was diagnosed with Parkinson’s disease nine years ago, he was a 39-year-old daredevil on a skateboard, flipping and leaping from walls, benches and dumpsters through the streets of San Francisco. He appeared in videos and magazines, and had sponsorships from skateboard makers and shops.
But gradually, he began to notice that “things weren’t really working right” with his body. He found that his right hand was cupping, and he began cradling his arm to hold it in place. His balance and alignment started to seem off.
Over time, he developed a common Parkinson’s pattern, fluctuating between periods of rapid involuntary movements like “I’ve got ants in my pants” and periods of calcified slowness when, he said, “I could barely move”.
A couple of years ago, Connolly volunteered for an experiment that summoned his daring and determination in a different way. He became a participant in a study exploring an innovative approach to deep brain stimulation.
In the study, which was published Monday in the journal Nature Medicine, researchers transformed deep brain stimulation – an established treatment for Parkinson’s – into a personalised therapy that tailored the amount of electrical stimulation to each patient’s individual symptoms.
The researchers found that for Connolly and the three other participants, the individualised approach, called adaptive deep brain stimulation, cut in half the time they experienced their most bothersome symptom.
Connolly, now 48 and still skateboarding as much as his symptoms allow, said he noticed the difference “instantly”. He said the personalisation gave him longer stretches of “feeling good and having that get-up-and-go”.
The study also found that in most cases, patients’ perceived quality of life improved. “That’s very important,” said Dr Sameer Sheth, a professor of neurosurgery at Baylor College of Medicine who was not involved in the research.
Although the study was small, it represents the strides being made in using brain implants and artificial intelligence to personalise treatment for neurological and psychiatric disorders – essentially developing pacemakers for the brain.
Recent experiments have begun individualising brain stimulation for depression, obsessive compulsive disorder and chronic pain. While much more research is needed – along with ways to make the approaches pragmatic and affordable – several experts predicted that some versions of brain pacemakers could be available within five or 10 years.
“I do think this points the way that indeed personalized, individualized stimulation is the wave of the future,” said Dr. Jaimie Henderson, a professor of neurosurgery at Stanford University who was not involved in the study.
Deep brain stimulation has been used for years for Parkinson’s. It’s typically introduced when patients have achieved what benefit they can from medications containing levodopa, a drug that combats the deficit of the hormone dopamine that drives Parkinson’s disease.
Under conventional deep brain stimulation, patients receive a constant level of electrical pulses. While it helps most patients, many eventually reach a plateau or, because the therapy does not adjust to a patient’s experience, the stimulation may be too much or too little and lead to drastic swings between periods of rigidity and unbridled motion.
“It’s not like we’ve like maximised, optimised, finalised our abilities to treat Parkinson’s patients,” Sheth said.
In recent years, neuroscientists have identified brain signals that correspond to phases of stiffness, called bradykinesia, and phases of uncontrolled movement, called dyskinesia. In the new study, researchers used methods derived from AI to devise a personalised algorithm for each patient and a way to detect and respond to brain activity as the patient’s symptoms fluctuated.
“The brain changes in its needs moment to moment, hour to hour, week to week,” said Dr Philip Starr, a professor of neurological surgery at the University of California, San Francisco, and a senior author of the study who has worked on deep brain stimulation for decades. “So it’s been a dream to make these stimulators self-regulating.”
Dr Simon Little, an associate professor of neurology at UCSF who led the study along with Starr, said that electrodes implanted in patients’ brains recorded signals from populations of neurons, not individual brain cells.
“When neurons start synchronising together, when they start all firing at the same time, they’re a bit like a crowd at a sports game,” Little said, adding that in Parkinson’s and some other disorders, neurons become excessively synchronised, with different patterns for different symptoms. “Some synchronisation is good, but if you get over-synchronised, if your whole network starts chanting at the same time, then that’s an unhealthy state because the brain is not really doing much information processing.”
The individualised system in the study reads signals from two separate brain areas and delivers more electrical stimulation when patients enter periods of stiffness and less when they begin phases of involuntary movement – more stimulation when their levodopa medication is wearing off and less when the drug is active.
The study’s participants were men in their 40s through 60s who had been diagnosed with Parkinson’s at least six years earlier. First, electrodes were implanted in their brains and they received conventional deep brain stimulation for months, until they had benefitted as much as they could from that.
Then researchers began developing personalised algorithms for the patients, primarily focusing on the symptom each patient found most bothersome. For three patients, including Connolly, that symptom was stiffness. For the fourth patient, it was involuntary spasms called dystonia.
Starr said the first patient’s algorithm took two years to devise, partly because stimulation itself would change brain signatures so they had to be repeatedly reassessed. But by the fourth patient, it took only two weeks.
To test if individualised stimulation improved patients’ experience, they were encouraged “to live as normal a life as they could,” Little said. “That could include skating, travelling, exercise.”
At intervals of two to seven days over a period of two months, the system would change to deliver either conventional stimulation or the personalised version. Neither the patients nor most of the research team knew which type of stimulation was being delivered when.
“Most prior studies were done in the laboratory, so that means they’re not very practical,” Henderson said. “This was in actual real-world conditions, and that’s extremely impressive because that’s a high bar. That’s hard to do.”
Patients completed questionnaires daily, and wearable monitors tracked how their movements changed. Little said three patients, including Connolly, had guessed when they were receiving adaptive stimulation “because of the symptom improvement”.
Most patients went from experiencing their worst symptoms for about 25% of the day to about 12% of the day, Starr said. Adaptive stimulation also decreased those symptoms’ severity. And, importantly, it did not worsen – and in some cases, it improved – the “opposite” symptoms. (The “opposite” of stiffness was uncontrolled movement, for example.)
One patient reported less time enduring a third symptom: gait disturbance. Starr said another patient “made a big deal about the fact that his day just started on time” instead of spending hours feeling too stiff to begin his morning activities. Three patients reported improved quality of life in categories like mobility, pain and ability to do usual activities. “The fourth patient is kind of a happy person” who rated his quality of life high to begin with, Starr said.
Connolly said he had volunteered for the study partly because his wife, Thuy Nguyen, who died of cancer in 2020, had urged him to try deep brain stimulation. “When my wife was sick in the hospital, I would take my skateboard and my cane, and during some parts of the day I could skate around, and other parts I’d have to use my cane,” he recalled.
During the study, Connolly, who runs a skateboard programme for kids that he and his wife founded, arranged to take a break midtrial because he wanted to avoid the experimental switching of stimulation while running a summer skateboard camp.
He could recognise if he had been switched to conventional or personalised stimulation because, he said, “I’d either feel good or I’d feel sluggish, almost immediately”.
He said the individualised algorithm had also improved his sleep, which can become disrupted in patients with Parkinson’s disease.
One wrinkle with adaptive stimulation, Starr said, is that algorithms will require frequent adjusting because as patients “progress in their Parkinson’s, they change in medications, they change in their activities”.
After the study, three patients continued with adaptive stimulation, though one had to pause for adjustments, Starr said. Connolly also plans to restart the personalised stimulation in the fall, after skateboard camp.
This article originally appeared in The New York Times.
Written by: Pam Belluck
Photographs by: Jason Henry
©2024 THE NEW YORK TIMES
Harris makes a surprise early appearance on the night of Biden’s speech.