Portable long-term EEG worn at home over days to weeks — catching infrequent spells that routine and even inpatient EEG can miss
Ambulatory EEG (aEEG) is designed around a simple insight: seizures and spells often will not obligingly occur during a 20-minute clinic recording, or even during a few controlled days in a hospital bed. By fitting a compact, battery-powered recorder to a standard array of scalp electrodes and sending the patient home, ambulatory EEG extends observation into the settings — bedroom, kitchen, commute, workplace — where events actually happen.
A technologist applies electrodes using the standard International 10–20 system, the same montage used for routine and inpatient EEG, so that traces remain comparable across settings:
• Scalp cup electrodes (Ag/AgCl) attached with conductive paste, secured under a mesh cap or with collodion for multi-day durability • Typical montage: 19–25 channels covering frontal, temporal, central, parietal, and occipital regions bilaterally • Electrode impedance checked (<5 kΩ target) before the patient leaves clinic • Reference and ground electrodes plus 1–2 ECG channels (to distinguish cardiac artifact from cerebral activity) • Wires bundled into a single cable running to a compact recorder, physically strain-relieved to survive days of normal movement
The recorder itself is a solid-state digital EEG amplifier no larger than a paperback book, worn on a belt clip, shoulder strap, or small backpack:
• Continuous 24-bit digitization at 200–500 Hz sample rate per channel, stored to removable flash memory or streamed via a companion phone app • Patients shower with waterproof cap covers in some systems, sleep normally, and are explicitly instructed to resume ordinary activity — walking, working, driving (per local regulations), household chores • This is the central methodological difference from inpatient EMU monitoring: instead of controlling the environment to provoke or capture an event (sleep deprivation, medication taper, video-monitored bed rest), ambulatory EEG accepts an uncontrolled environment in exchange for a much longer observation window • A written seizure diary and event-marker button are explained at fitting, since the patient becomes an active participant in data annotation, not just a passive recording subject.
The core trade-off of ambulatory EEG is captured in one sentence: trade video and environmental control for duration. A week of unsupervised home recording statistically outperforms a day or two of supervised, controlled inpatient recording whenever the events of interest are infrequent.
The single greatest advantage of ambulatory EEG over a routine outpatient EEG is duration. A routine EEG samples perhaps 20–40 minutes of brain activity — a vanishingly small fraction of a day. Ambulatory recorders instead capture continuous, gap-free EEG for 24 hours to two weeks, transforming the diagnostic question from "did we get lucky during a short snapshot?" to "how much cumulative time did we actually observe?"
Capture probability for an infrequent, randomly-timed clinical event follows an approximately Poisson process: if a patient experiences events at an average rate λ per day, the probability of observing at least one event during a monitoring window of T days is:
P(≥1 event) = 1 − e^(−λT)
This relationship explains why duration has an outsized effect on yield for rare events. A patient with spells occurring once every two months (λ ≈ 0.017/day) has only a ~0.3% chance of an event during a 20-minute routine EEG, a ~5% chance during a 3-day inpatient stay, but a ~13% chance during a 2-week ambulatory recording — and every additional day continues to compound that probability.
For more frequent events (several times per week), even a single day of continuous recording often suffices — which is why ambulatory EEG duration is typically individualized to the patient's reported event frequency rather than fixed.
Recording continuously for days generates far more data than any reviewer can watch in real time, so ambulatory EEG systems rely on:
• Gap-free multi-channel digital storage, typically 200–500 samples/second across 19–25 channels, accumulating several gigabytes per day • Onboard or offline compressed spectral array (CSA) and amplitude-integrated EEG (aEEG) trend displays that compress hours of raw signal into a single scrollable panel, letting reviewers scan days of data in minutes • Automated spike- and seizure-detection algorithms that flag candidate epochs for focused human review, dramatically reducing the review burden compared to reading every raw second • Time-synchronized channel for patient-triggered event markers and any actigraphy/accelerometer data, layered onto the same timeline as the EEG so movement and reported symptoms can be cross-referenced against brain activity
Unlike inpatient monitoring, where trained staff and continuous video document exactly what happens during a spell, ambulatory EEG depends on the patient (or a family member/caregiver) actively participating in data collection. Pressing an event marker button and writing a diary entry at the time of a spell creates the essential bridge between "something happened at 14:32" and "here is what the EEG was doing at 14:32."
At fitting, patients and household members are trained on two complementary tools:
• Event marker button: a physical or app-based button pressed at the moment a spell is recognized (by the patient or a witness), which writes a precise timestamp into the continuous EEG file — this is the single most important piece of metadata in the entire recording, because it tells reviewers exactly where to look in days of data • Paper or app-based diary: structured fields for time of onset, approximate duration, symptoms experienced (aura, altered awareness, motor activity, automatisms), activity at onset (sleeping, walking, stressed, post-meal), and recovery time — richer context than the marker alone can provide
Because the patient may be unaware they are having an event (as in absence seizures or nocturnal events), family members or bed partners are explicitly asked to press the marker on the patient's behalf whenever they witness a spell, and diaries often include a field for witness identity.
During review, every marked timestamp becomes an anchor point: the reviewer pulls up the EEG epoch surrounding that timestamp and asks three questions —
1. Is there an EEG correlate at all? (electrographic seizure, epileptiform discharge, or no change) 2. Does the EEG change align in time and duration with the reported symptoms? 3. Is the EEG pattern consistent with an epileptic mechanism, or does it look like a non-epileptic event (a normal or artifact-obscured trace during a clinically convincing "spell" raises suspicion for a psychogenic non-epileptic event)?
This correlation step is where ambulatory EEG earns its clinical value beyond passive spike-counting: a marked event with no EEG change is itself a critical diagnostic finding, often redirecting management entirely.
A frequent and clinically important finding is the "negative capture": the patient marks a clear clinical spell, but the simultaneous EEG shows no epileptiform change. This does not mean nothing happened — it may point toward a non-epileptic paroxysmal event (syncope, non-epileptic attack, movement disorder) and can be just as diagnostically decisive as capturing a true seizure.
A home environment, unlike a supervised inpatient bed, generates far more movement, muscle, chewing, electrode-pop, and environmental artifact. The reviewing neurophysiologist must sift through days of trace to distinguish genuine epileptiform discharges and electrographic seizures from the much larger volume of artifact that free-living recording inevitably produces.
Compared to a patient lying still in a monitored inpatient bed, an ambulant patient going about daily life introduces artifact from many additional sources:
• Myogenic (EMG) artifact: chewing, talking, facial expression, and general muscle tension produce high-frequency activity that can obscure or mimic fast epileptiform spikes • Movement/mechanical artifact: walking, exercise, and postural change cause electrode-cable motion and transient baseline shifts • Electrode-related artifact: sweat-induced impedance drift, partial electrode detachment over multiple days, and "electrode pop" (sudden brief high-amplitude deflection) • Environmental/electrical artifact: 50/60 Hz mains interference near appliances, electromagnetic interference from phones or electronics • Physiological non-cerebral artifact: eye blink and eye-movement (EOG), cardiac (ECG) artifact, and pulse artifact from electrodes overlying scalp vessels
Because the recording is unsupervised, none of this can be prevented in real time the way an inpatient technologist might reposition a lead — it must instead be recognized and excluded during offline review.
Modern ambulatory EEG review combines software and human expertise in a staged pipeline:
1. Automated spike/seizure detection algorithms scan the full multi-day file and flag candidate epochs based on waveform morphology, frequency content, and field distribution — these detectors are tuned for high sensitivity, which means they also flag a large number of artifactual look-alikes 2. A technologist performs a first-pass review of every flagged epoch, using the synchronized event-marker and diary timeline to prioritize clinically relevant windows, and discards epochs with clear artifact signatures (out-of-phase channels, non-physiological frequency content, correlation with the accelerometer/EMG channel) 3. A neurologist/epileptologist reviews the remaining candidate epochs plus all patient-marked events, applying standard criteria for epileptiform discharges (spike or sharp wave with after-going slow wave, physiological field, reproducible morphology) versus artifact 4. Only after this adjudication is a discharge or electrographic seizure formally reported — a step that, in ambulatory recordings, consumes substantially more reviewer time per recorded day than inpatient review because of the higher artifact burden
Because muscle and movement artifact in the beta/gamma range can superficially resemble fast epileptiform spikes, experienced reviewers rely heavily on field distribution (a true cerebral discharge respects an anatomically plausible electrode topography) and correlation with the simultaneous ECG/EMG or accelerometer channel to reject artifact confidently rather than over-calling epileptiform activity.
Ambulatory EEG earns its place in the diagnostic pathway because, for infrequent spells, a longer real-world observation window usually beats a shorter but more heavily instrumented inpatient admission. But that advantage comes with real trade-offs: no synchronized video to visually characterize an event, no controlled provocation (sleep deprivation, medication withdrawal), and a less predictable, less supervised recording environment.
Ambulatory EEG occupies a middle position between two other tools:
• Routine outpatient EEG (20–40 min, ± sleep deprivation): fastest and cheapest, but samples only a tiny fraction of a day — well suited to frequent interictal abnormalities, poorly suited to capturing an actual clinical event • Inpatient video-EEG monitoring (EMU): the gold standard for characterizing an event, because continuous synchronized video lets the team see exactly what the patient did during a spell (automatisms, tonic-clonic pattern, eye deviation) alongside the simultaneous EEG — but admissions are typically only a few days and are logistically and financially costly, so the same duration limitation as routine EEG applies, just less severely • Ambulatory EEG: sacrifices video and environmental control in exchange for the ability to extend recording to a week or two at comparatively low cost and burden, which is often decisive when spells occur less than once every few days
The single largest limitation of standard ambulatory EEG is the absence of time-synchronized video. When an electrographic seizure or an epileptiform discharge is captured, the team knows what the brain was doing but not what the patient was doing — was there loss of awareness, automatisms, a focal motor pattern, or a convincing non-epileptic behavior?
This matters clinically in two directions: • A captured electrographic event without video cannot always be confidently classified by seizure semiology, which can affect syndrome classification and surgical planning • A marked clinical spell with a clean EEG trace is suggestive of a non-epileptic event, but without video the team cannot describe what was observed to confirm a psychogenic or physiological (e.g., syncopal) mechanism
Some newer ambulatory systems add a smartphone or wearable camera to partially close this gap, but picture quality, patient compliance with keeping the camera in frame, and battery/storage constraints make this a partial rather than complete substitute for inpatient video-EEG.
In practice, the choice between routine EEG, ambulatory EEG, and inpatient video-EEG is driven primarily by expected event frequency and the clinical question being asked:
• Frequent events (several per week) or a need to characterize semiology in detail → inpatient video-EEG, since a short admission is enough to capture events with video • Infrequent events (roughly weekly to monthly) where an interictal or electrographic correlate would answer the clinical question → ambulatory EEG, extended long enough (using the Poisson-style capture-probability logic from Stage 2) to give a reasonable chance of capture • Very infrequent events (less than monthly) → either a longer ambulatory recording (up to ~2 weeks), or in select cases an implantable/subscalp long-term monitor capable of recording for months, since even a 2-week ambulatory study may fall short
The practical rule of thumb clinicians apply: ambulatory EEG trades the rich contextual detail of inpatient video-EEG for statistical reach — a longer window multiplies the odds of capturing something, even though what is captured will be interpreted with a real-world, artifact-prone trace and no picture of what the patient was doing.