Already, engineers and researchers are trying to develop various forms of technology — from iPhone apps to watches — to detect Parkinson’s disease early in patients, which is difficult for clinicians to do.
“For diseases like Parkinson’s… one of the biggest challenges is that we need to get to it [it] “It’s very early, before the damage often occurs in the brain,” said Dina Katabi, study author and professor of electrical engineering and computer science at MIT. “So being able to detect Parkinson’s disease early is essential.”
However, medical ethicists said, the algorithm underscores a broader concern in health care: Technological advances are being used to bolster claims that computers should fuel more medical decision-making without strong evidence to back it up. They said the algorithms could be useful in detecting Parkinson’s disease, but they urge more testing because they worry the technology could create false positive diagnoses.
“If you read about AI, there is a huge amount of overselling … that AI will solve huge amounts of practical problems,” said Torbjørn Gundersen, who researches the use of algorithms in medicine at Oslo Metropolitan University in Norway. “It hasn’t really been proven yet.”
Parkinson’s disease is a neurological disease that reduces the amount of dopamine neurons that are released in the part of the brain that controls movement. As it progresses, people can experience tremor, stiff limbs, and general slowness. according to Parkinson’s Foundationwith nearly 10 million people living with it globally.
James Beck, chief scientific officer of the Parkinson’s Foundation, said that despite the disease’s prevalence, clinicians do not have a widely accepted way to detect Parkinson’s disease in patients. This often results in doctors misdiagnosing the disease or discovering it too late in its progression, when tremors may already be evident.
“It’s really tough,” Beck said. “There is no blood test. No brain scan. There is no objective way to say whether or not someone has Parkinson’s disease. It requires a skilled physician.”
Katabi and Yuzhe Yang, a researcher at the Massachusetts Institute of Technology and lead author of the study, set out by trying to solve this problem using machine learning. They trained the algorithms on sleep data collected from more than 7,600 people, nearly 750 of whom had Parkinson’s disease.
To collect the data, the researchers developed a tool — which looks like a small box — that can be placed in the study participants’ room and wirelessly collect breathing patterns from people while they sleep. Some data were also excluded from existing data sets collected at academic sleep centers.
The data was used to train a neural network that ended up predicting with high accuracy whether or not a person had Parkinson’s disease. It was 90 percent accurate based on data from one night’s sleep. The model improved to 95 percent accuracy when analyzing 12 nights of breathing patterns. The neural network can also track how severe a patient’s Parkinson’s disease is.
Al Ketbi said the AI model could provide a range of benefits. She said drug companies trying to create drugs to treat and treat Parkinson’s disease could use the tool to better track disease severity in patients enrolled in their clinical trials, speeding up the drug creation process. People who live in remote places, far from neurologists, can have a way to detect and track disease without having to take long trips.
“Most people with Parkinson’s disease tend to live far from these medical centers,” she said. “So they end up not receiving proper treatment and care from an expert.”
Katabi added that the tool, called the Emerald device, is being used by large pharmaceutical and biotech companies working on Parkinson’s treatments, but declined to name the companies, citing confidentiality agreements.
The AI tool is simply one of the many ways scientists are racing to better detect and track Parkinson’s disease, said Beck, of the Parkinson’s Foundation. He said these tools should not replace a doctor’s diagnosis, but should be used as part of a broader strategy that helps doctors identify disease early.
“This should not supplement or replace a clinical diagnosis,” he said. “You should help with that… so we can come up with it [a test] This is based on a slightly more biological basis. “
Gundersen, of Oslo Metropolitan University, said that while the research study is promising, more needs to be done. He said that there are many studies emerging that show that AI has an advantage over humans in performing certain medical tasks, such as diagnosing diseases, but he noted that there are fewer studies showing whether these algorithms improve health outcomes when used in a clinical setting.
Written approved. “We need more data,” she said. “We’re just getting started with these results, and we need more evidence.”
Ethically speaking, Gundersen added, AI algorithms in healthcare pose a bigger problem: Who is to blame if computers misdiagnose?
He said, “If we believe that holding people accountable is something of value in society, then the use of artificial intelligence [would] challenge this.”