Parkinson’s disease is the fastest growing neurodegenerative disease, currently affecting more than 10 million people worldwide, yet clinicians still face enormous challenges in tracking its severity and progression.
Clinicians typically evaluate patients by testing their motor skills and cognitive function during outpatient visits. These semi-subjective measures are often influenced by external factors—perhaps a patient feels tired after a long drive to the hospital. More than 40% of people with Parkinson’s have never been treated by a neurologist or a Parkinson’s specialist, often because they live too far from the city center or have difficulty getting around.
To address these issues, researchers at MIT and elsewhere have demonstrated a home device that can monitor a patient’s movement and gait speed, which can be used to assess Parkinson’s disease severity, disease progression, and a patient’s response to medication .
About the size of a Wi-Fi router, the device passively collects data using radio signals that bounce off a patient’s body as they move around the home. Patients do not need to wear gadgets or change their behavior. (For example, a recent study showed that this type of device could be used to detect Parkinson’s disease from a person’s breathing patterns while they sleep.)
The researchers used the devices to conduct two studies involving 50 participants. They showed that by using machine learning algorithms to analyze the vast amount of data they collected (more than 200,000 gait speed measurements), clinicians could track Parkinson’s disease progression more effectively than regular clinical assessments.
“By having a device in the home that can monitor the patient and tell the doctor remotely how the disease is progressing and how the patient is responding to their medication, they can look after the patient even if they can’t come to the hospital. Clinics — now they have real, reliable information — This actually goes a long way towards improving equity and access,” said senior author Dina Katabi, Professor Thuan and Nicole Pham in the Department of Electrical Engineering and Computer Science (EECS), and the Computer Science and Artificial Intelligence Laboratory (CSAIL) and Hemp Principal Investigator at the Jameel Clinic at the Provincial Institute of Technology.
Co-lead authors are EECS graduate students Yingcheng Liu and Guo Zhang.The study was published in Science Translational Medicine.
The work took advantage of a wireless device previously developed in Katabi’s lab that analyzes radio signals bouncing off the human body. The signals it transmits use a fraction of the power of a Wi-Fi router – these ultra-low-power signals won’t interfere with other wireless devices in your home. When radio signals pass through walls and other solid objects, they are reflected by the human body due to the water in our bodies.
This creates a “body radar” that can track the movements of people in the room. Radio waves always travel at the same speed, so the length of time it takes for the signal to reflect back to the device indicates how a person is moving.
The device contains a machine-learning classifier that can pick out precise radio signals reflected off a patient, even when other people are walking around the room. Sophisticated algorithms use this movement data to calculate gait speed — how fast a person can walk.
Because the device runs in the background, running all day long, a huge amount of data can be collected every day. The researchers wanted to see if they could apply machine learning to these datasets to gain insights into the disease over time.
They collected 50 participants, 34 of whom had Parkinson’s disease, and conducted two observational studies of home gait measurements. One study lasted two months, the other two years. From these studies, the researchers collected more than 200,000 individual measurements, which they averaged to remove variability due to equipment condition or other factors. (For example, a device might turn off unexpectedly during cleaning, or a patient might walk more slowly while on the phone.)
Using statistical methods to analyze the data, they found that home gait speed can be used to effectively track the progression and severity of Parkinson’s disease. For example, they found that gait speed declined almost twice as fast in Parkinson’s patients than in non-Parkinsonian patients.
“Continuously monitoring patients as they move around the room allows us to get a really good measure of their gait speed. With so much data, we’re able to aggregate that allows us to see very small differences,” Zhang said.
better, faster results
Digging into these changes provides some key insights. For example, researchers could see intraday fluctuations in patients’ gait speed that corresponded to how they responded to the drug — gait speed might increase after taking the drug, and then start to decline after a period of time.
“This does make it possible for us to objectively measure how your mobility responds to the drug. Previously, this was nearly impossible because the drug effect could only be measured by having patients keep a diary,” Liu said.
Clinicians can use this data to adjust drug dosages more efficiently and accurately. This is especially important because many drugs used to treat disease symptoms can cause serious side effects if patients take too much.
After studying 50 people for a year, the researchers were able to demonstrate statistically significant results regarding the progression of Parkinson’s disease. By contrast, an oft-cited study by the Michael J. Fox Foundation involved more than 500 people and monitored them for more than five years, Katabi said.
“For pharmaceutical companies or biotech companies trying to develop drugs for this disease, this could greatly reduce the burden and cost and speed up the development of new treatments,” she added.
Katabi attributes much of the study’s success to a dedicated team of scientists and clinicians who worked together to overcome the many difficulties that arose along the way. For one thing, they started this research before the Covid-19 pandemic, so engineers initially went into people’s homes to set up the devices. When this was no longer possible, they developed a way to deploy the device remotely and created a user-friendly app for participants and clinicians.
Through the research process, they learned to automate processes and reduce workload, especially for participants and clinical teams.
This knowledge will prove useful when they hope to deploy the device in home studies of other neurological disorders such as Alzheimer’s, ALS and Huntington’s. They also want to explore how these methods can be used, in conjunction with other work in the Katabi lab showing that Parkinson’s can be diagnosed by monitoring breathing to collect a global set of markers that can diagnose the disease early, which can then be used to track and treat it.
This work was supported in part by the National Institutes of Health and the Michael J. Fox Foundation.