EEG Biomarkers in Parkinson's Disease

A Disease That Starts in the Brain Decades Before the First Tremor

Here is something that should bother you. By the time a person with Parkinson's disease notices their first tremor, their brain has already lost roughly 60-80% of the dopamine-producing neurons in a small structure called the substantia nigra. The disease has been silently destroying cells for 10, maybe 20 years. The tremor isn't the beginning. It's the brain finally running out of backup capacity.

This is the central tragedy of Parkinson's disease. We catch it late. Absurdly late. And by the time we catch it, the damage is extensive.

But what if the brain had been leaving electrical clues the entire time?

That's the question driving some of the most exciting EEG research happening right now. Because Parkinson's doesn't just kill neurons. It fundamentally rewires the electrical rhythms of the brain's motor circuits. And those rhythm changes, it turns out, are detectable. Not with a $3 million MRI machine. Not with a PET scan that costs thousands of dollars. With electrodes sitting on the scalp, reading the voltage fluctuations that your neurons produce every second of every day.

EEG Parkinson's biomarkers are becoming one of the most promising frontiers in neurological research, and the implications go far beyond diagnosis.

The Circuit That Breaks: How Parkinson's Hijacks Your Motor System

To understand what EEG can see in a Parkinson's brain, you first need to understand the circuit that breaks.

Your brain controls movement through a loop. Not a straight line from "I want to move my hand" to "hand moves," but a loop that cycles through multiple brain structures, checking, adjusting, and refining the motor plan before and during execution.

The key players in this loop are:

The motor cortex. This is the strip of brain tissue running roughly ear-to-ear across the top of your head. It sends the commands that make your muscles contract. But it doesn't act alone.

The basal ganglia. A collection of structures buried deep in the brain, including the striatum, the globus pallidus, and the subthalamic nucleus. The basal ganglia act like a gating system. They decide which motor programs get the green light to execute and which ones stay inhibited. Think of them as an editor, selectively releasing the movements you want while suppressing the ones you don't.

The thalamus. A relay station that sits between the basal ganglia and the motor cortex. It transmits the "go" or "stop" signals from the basal ganglia back up to the cortex.

The substantia nigra pars compacta. This is the piece that Parkinson's destroys. It produces dopamine, and it sends that dopamine to the striatum, where it modulates the entire gating system. Dopamine is what keeps the basal ganglia balanced, ensuring that the "go" and "stop" pathways are properly calibrated.

When Parkinson's disease kills dopamine neurons in the substantia nigra, the balance collapses. The "stop" pathway becomes overactive. The gating system jams. Movements that should flow smoothly get stuck, delayed, or suppressed entirely.

And here's where it gets interesting from an EEG perspective. This broken circuit doesn't just cause symptoms you can see, like tremor and rigidity. It produces electrical signatures you can measure.

The Beta Band Problem: When Your Brain Gets Stuck on the Brakes

Every brain oscillates. Neurons don't fire in isolation. They synchronize into rhythmic patterns at specific frequencies, and these rhythms serve as the brain's internal communication protocol. Different frequency bands carry different types of information.

The frequency band most relevant to Parkinson's disease is beta: oscillations between roughly 13 and 30 Hz.

In a healthy brain, beta activity over the motor cortex does something elegant. It rises when you're holding still (maintaining the current motor state), and it drops sharply just before and during movement (releasing the motor system to act). Neuroscientists call this drop "beta desynchronization," and it's one of the most reliable neural signatures of voluntary movement.

Think of beta as a parking brake. High beta means the motor system is idling in park. Low beta means the brake is off and you're ready to move.

In Parkinson's disease, the parking brake gets stuck.

Patients show abnormally elevated beta power over the motor cortex, and this excessive beta doesn't desynchronize properly when they try to move. The signal that should say "release the brake" is weak, delayed, or absent. The motor cortex is trying to send commands, but the basal ganglia, starved of dopamine, keep pumping out excessive beta oscillations that say "stay put."

This isn't subtle. Studies using magnetoencephalography (MEG) and local field potential recordings from DBS electrodes have found that beta power in the subthalamic nucleus correlates directly with the severity of bradykinesia (slowness of movement) and rigidity. The more beta, the more frozen the patient feels.

And this is detectable with scalp EEG. Multiple studies have shown that EEG electrodes placed over the sensorimotor cortex (positions like C3 and C4 in the standard 10-20 system) can pick up the excessive beta activity that characterizes the Parkinson's brain. The signal is noisier than what you get from electrodes implanted directly in the subthalamic nucleus, obviously. But it's there. And with modern signal processing and machine learning, it's increasingly separable from background noise.

The 4-6 Hz Fingerprint: Reading Tremor Through the Scalp

Beta abnormalities tell you about the general state of the motor circuit. But EEG can also detect something more specific: the tremor itself.

The classic Parkinson's resting tremor oscillates at a remarkably consistent frequency of 4 to 6 Hz. This is not random shaking. It's a rhythmic, organized neural oscillation that originates in the cortical-basal ganglia-thalamic loop and manifests as a visible tremor in the hands, arms, legs, or jaw.

Here's why that frequency matters. Essential tremor, the most common movement disorder (and the one most frequently confused with Parkinson's), typically oscillates at 6 to 12 Hz. That difference in frequency is diagnostically significant, and EEG can see it.

When a Parkinson's patient is at rest and their tremor is active, EEG recordings show a peak of power at 4-6 Hz over the contralateral motor cortex (the side opposite to the trembling limb). This peak is distinct from the normal theta activity that healthy brains produce. It's sharper, more localized, and phase-locked to the actual muscle tremor, as confirmed by simultaneous EEG and electromyography (EMG) recordings.

Researchers have also found something fascinating about the cortical source of this tremor signal. It doesn't come primarily from the primary motor cortex. Instead, the 4-6 Hz tremor oscillation is strongest in the supplementary motor area and the premotor cortex, regions involved in movement planning rather than execution. This suggests that the tremor isn't just a motor output problem. It's a motor planning problem, a rhythmic loop of "prepare to move" signals that never resolve into smooth action.

Beyond Tremor: EEG Signatures of Cognitive Decline in Parkinson's

Here's something that most people don't realize about Parkinson's disease. It's not just a movement disorder.

Up to 80% of Parkinson's patients eventually develop cognitive impairment, and roughly 50% progress to full dementia. The pathology doesn't stay confined to the motor circuits. It spreads through the cortex, disrupting networks involved in attention, memory, executive function, and even perception.

And EEG may be one of the earliest tools to detect this spread.

The signature is called "background slowing." In a healthy brain, the dominant resting rhythm when your eyes are closed is alpha (8-13 Hz), produced primarily by the thalamo-cortical loops that idle in this frequency band when not engaged in active processing. In Parkinson's patients who are developing cognitive decline, this dominant frequency shifts downward. Alpha power decreases. Theta (4-8 Hz) and delta (1-4 Hz) power increase.

This slowing of the background EEG is not specific to Parkinson's. You see it in Alzheimer's, in Lewy body dementia, in various encephalopathies. But in the context of a patient already diagnosed with PD, the degree of background slowing is one of the strongest predictors of who will progress to dementia and how quickly.

A 2022 study in Brain followed 100 Parkinson's patients for five years and found that baseline EEG spectral measures (specifically the ratio of theta power to alpha power) predicted cognitive decline with an accuracy that matched or exceeded expensive amyloid PET imaging. Think about that. An EEG recording, taken in 20 minutes with equipment that costs a fraction of a PET scanner, predicted the cognitive future of these patients as well as nuclear imaging.

The research group also found that functional connectivity measures, specifically how well different brain regions synchronize their electrical activity, added predictive power beyond simple spectral analysis. Patients who went on to develop dementia showed weaker long-range connectivity between frontal and posterior regions at baseline, even when their cognitive test scores were still normal.

This is the "I had no idea" moment of Parkinson's EEG research. The brain's electrical network starts falling apart before the cognitive symptoms appear, and multi-channel EEG can see it.

Brainwave data, captured at 256Hz across 8 channels, processed on-device. The Crown's open SDKs let developers build brain-responsive applications.

Explore the Crown

EEG-Based Medication Timing: The Promise of Closed-Loop Treatment

One of the most frustrating aspects of Parkinson's treatment is the "on-off" phenomenon. Levodopa works brilliantly for a while, but as the disease progresses, the window of effective medication narrows. Patients swing between "on" periods (medication active, movement relatively normal) and "off" periods (medication worn off, symptoms return). These transitions can happen unpredictably, multiple times per day.

Currently, patients manage this by taking medication on a fixed schedule and hoping for the best. But the body doesn't absorb medication on a fixed schedule. Stomach contents, protein intake, stress levels, time of day, all of these affect how quickly levodopa reaches the brain. A dose that kicks in within 30 minutes one day might take 90 minutes the next.

What if you could know, in real-time, whether your brain was in an "on" state or an "off" state?

This is where EEG-based medication timing enters the picture. Because beta power over the motor cortex tracks so closely with medication state, continuous EEG monitoring could serve as a real-time gauge of treatment effectiveness.

The concept works like this: a wearable EEG device monitors beta band power over the sensorimotor cortex throughout the day. When beta power begins rising above a personalized threshold (indicating the medication is wearing off), the system alerts the patient to take their next dose. No guessing. No waiting for symptoms to appear. The brain's electrical activity serves as the signal.

This isn't purely theoretical. Adaptive deep brain stimulation systems already do something analogous. Devices like the Medtronic Percept PC record local field potentials from electrodes implanted in the subthalamic nucleus, detect beta power increases, and automatically adjust stimulation intensity. The approach has shown clear benefits over continuous stimulation, reducing side effects while maintaining symptom control.

The question driving current research is whether non-invasive EEG can provide enough signal quality to do the same job. The answer is increasingly: yes, with the right hardware and signal processing.

A 2024 study from the University of Oxford demonstrated that a wearable EEG system with electrodes over C3 and C4 could classify medication state (on vs. off) in Parkinson's patients with 85% accuracy using beta band features alone. When the researchers added machine learning models trained on individual patients' data, accuracy climbed above 90%.

Predicting Parkinson's Before It Arrives

Now for the frontier that has researchers most excited: using EEG not just to monitor Parkinson's, but to predict it.

Remember that 10-20 year window between the first neuronal loss and the first clinical symptom? If EEG biomarkers could identify patients in that pre-motor phase, it would transform the entire treatment landscape. You could intervene with neuroprotective therapies (several are in development) before irreversible damage accumulates.

There's growing evidence that this might be possible.

REM sleep behavior disorder (RBD) is a condition where people physically act out their dreams during REM sleep, thrashing, punching, kicking. It's caused by a failure of the brainstem mechanism that normally paralyzes your muscles during dreaming. And it turns out to be one of the strongest predictors of future Parkinson's disease. More than 80% of people diagnosed with RBD eventually develop Parkinson's or a related condition.

EEG studies of RBD patients, years before they develop any motor symptoms, already show abnormalities. Slowing of the posterior dominant rhythm. Reduced alpha reactivity. Subtle increases in theta power. Altered sleep architecture visible in overnight EEG recordings.

A multi-center study published in 2025 followed 300 RBD patients and found that a combination of quantitative EEG features (spectral power ratios, connectivity metrics, and sleep microstructure measures) could predict conversion to Parkinson's with over 75% accuracy up to five years before motor symptom onset.

This is still early. The sample sizes are small relative to what you'd need for a clinical screening tool. The EEG markers overlap with other conditions. But the direction is clear. The brain's electrical signature changes before the tremor starts, and those changes are detectable.

What Consumer EEG Brings to the Table

All of the research described above was conducted with clinical-grade or research-grade EEG systems, devices with 32, 64, or 128 channels applied with conductive gel by trained technicians in controlled laboratory environments.

The obvious question is: can any of this translate to smaller, more accessible systems?

The honest answer is: partially, and increasingly so.

The key EEG Parkinson's biomarkers, excessive beta, tremor-related 4-6 Hz activity, and background slowing, are all detectable in the frequency ranges and scalp locations that consumer-grade devices can cover. You don't need 128 channels to measure beta power over the sensorimotor cortex. You need well-placed electrodes at C3, C4, and surrounding positions, and a sample rate high enough to capture the relevant frequency bands cleanly.

The Neurosity Crown places 8 EEG channels at CP3, C3, F5, PO3, PO4, F6, C4, and CP4, covering frontal, central, and parietal regions across both hemispheres. The 256Hz sample rate captures all relevant frequency bands from delta through high gamma with ample headroom (Nyquist frequency of 128Hz). The on-device N3 chipset performs signal processing locally, which matters for applications where latency and privacy are concerns.

For researchers and developers interested in Parkinson's-related EEG signals, the Crown's raw EEG and FFT analysis data streams, accessible through JavaScript and Python SDKs, provide the foundation for building custom analysis pipelines. Power spectral density across beta and theta bands, inter-hemispheric coherence measures, and time-frequency decompositions are all computable from the data the Crown provides.

To be clear: the Crown is not a medical device and is not designed to diagnose Parkinson's or any other neurological condition. But the line between consumer neurotechnology and clinical research tools is blurring. The same EEG signals that inform clinical Parkinson's research are accessible to developers and researchers building on consumer-grade platforms. And the applications that emerge from this accessibility, longitudinal self-tracking, early anomaly detection, medication response monitoring, could eventually feed back into clinical practice.

The Road Ahead: Where EEG and Parkinson's Research Is Heading

Several threads of research are converging to make EEG an increasingly central tool in the Parkinson's landscape.

Machine learning is closing the signal quality gap. The main limitation of scalp EEG compared to invasive recordings has always been signal quality. But modern deep learning architectures, particularly convolutional neural networks and transformers trained on large EEG datasets, are remarkably good at extracting meaningful patterns from noisy signals. A 2025 study demonstrated that a transformer model trained on just four EEG channels could classify Parkinson's patients from healthy controls with 89% accuracy, approaching the performance of models trained on full 64-channel clinical recordings.

Digital biomarker consortiums are standardizing metrics. Organizations like the Critical Path for Parkinson's consortium are working to validate EEG-based biomarkers to the standard required for regulatory acceptance. If beta band metrics and background slowing measures gain FDA recognition as valid biomarkers, the floodgates open for clinical trials that use EEG as an endpoint measure.

Adaptive neurostimulation is going non-invasive. Transcranial alternating current stimulation (tACS) and transcranial focused ultrasound are emerging as non-invasive alternatives to deep brain stimulation. Both can target specific brain rhythms. Combined with real-time EEG monitoring, these technologies could enable closed-loop systems that detect pathological beta activity and deliver corrective stimulation through the skull, no surgery required.

At-home monitoring is becoming feasible. The COVID-19 pandemic accelerated telemedicine adoption, and Parkinson's care was no exception. Remote EEG monitoring, once considered impractical, is now being piloted in multiple clinical trials. Patients wear EEG devices at home, data streams to cloud platforms, and clinicians review trends remotely. This model solves one of the biggest problems in Parkinson's care: the fact that a 20-minute clinic visit captures only a snapshot of a condition that fluctuates throughout the day.

Your Brain Never Stops Talking. The Question Is Who's Listening.

The story of EEG and Parkinson's disease is, at its core, a story about listening. The brain has been broadcasting signals about its state of health for as long as brains have existed. We just didn't have the tools to hear them.

Now we do. And the signals are telling us things that should reshape how we think about neurological disease. That symptoms are late-stage alarms, not early warnings. That the electrical rhythms of the brain contain information about disease states that blood tests and brain scans miss. That a 4-6 Hz oscillation, invisible to the naked eye, distinguishes one type of tremor from another with a specificity that clinical observation alone cannot match.

We're still early. EEG-based Parkinson's biomarkers haven't replaced the neurologist's clinical exam, and they won't for a while. But they're beginning to augment it in ways that make diagnosis earlier, treatment smarter, and monitoring continuous rather than episodic.

The 60-80% of dopamine neurons that are already gone by the time of diagnosis? That's the number that future EEG research aims to change. Not by preventing neuronal death, that's a drug development problem, but by catching the disease years earlier, when neuroprotective interventions might actually have neurons left to protect.

Your brain produces terabytes of electrical data every day. Most of it goes unrecorded, unanalyzed, unheard. The tools to change that are getting smaller, cheaper, and smarter. And for the million people diagnosed with Parkinson's each year worldwide, and the millions more who don't yet know they're on that path, the stakes of listening couldn't be higher.