Can recording fit into life at home?
Ear-EEG in a selected Alzheimer's population.
Read the home-recording insight →Research insights
New ways to detect seizure-related activity in dementia
Researchers are testing longer recordings, small ear electrodes, sensors beneath the scalp, and computer analysis. This page explains what three leading studies found, how their methods might help dementia research, and what must be proven before they can improve care.
Three studies, three questions
Each new method addresses a different problem.
A recording may be too short, deep activity may be difficult to see at the scalp, or the equipment may be difficult to use in everyday life. The percentages from these studies cannot be used to choose a “best” EEG.
Can software reveal a hidden signal?
Computer-assisted detection studied in temporal lobe epilepsy, not dementia.
Read the signal-detection insight →Can we observe beyond a short visit?
Implanted subscalp recording studied in adults with epilepsy.
Read the longer-term monitoring insight →Keep the finding clear. A brief seizure-related electrical discharge is different from a recorded seizure. An event missing from a diary may still have caused a subtle change that no one recognized. Start with the EEG and event distinctions.
Insight 01 | Alzheimer's disease study
Small ear electrodes allowed repeated two-day recordings.
The study shows what researchers can learn from repeated recordings and why comfort, comparison groups, and usable recording time affect the result. [1]
What was tested?
The study included 24 people with mild to moderate Alzheimer's disease and 15 people without dementia. Customized earpieces recorded EEG for two days at a time. People with Alzheimer's could have up to three sessions, while the comparison group had one.
People with Alzheimer's needed a care partner who could help with the equipment. Nursing-home residents and daily hearing-aid users were excluded, so the group did not represent everyone receiving dementia care.
What did the study find?
During the first recording, brief discharges appeared in 18 of 24 people with Alzheimer's disease and 7 of 15 people without dementia. The difference in those proportions was not statistically clear, although discharges occurred more often in the Alzheimer's group.
Across the repeat sessions, discharges were found at least once in 22 of 24 people with Alzheimer's disease. Activity varied substantially between sessions. These percentages describe brief discharges, not seizures without outward signs.
What limits the interpretation?
The ear electrodes and the longer recording time changed together. The study did not compare ear EEG with an equally long scalp EEG, and the ear findings were not checked against scalp recordings taken at the same time.
During the first session, people with Alzheimer's disease averaged 34.6 recorded hours, but only 23.7 hours remained usable after signal cleaning. Discomfort was the main reason eight people left the study. A wearable device can still be uncomfortable or difficult to use.
What remains unproven?
The study did not establish that detecting or treating these discharges improves cognition or daily function. Frequent findings in controls also mean that the mere presence of a discharge is not a stand-alone marker of Alzheimer's disease.
Validation needs to include the people excluded here, the assistance required, and an appropriate comparison recording. A separate small DLB study adds relevant early evidence, not proof across all dementia types.
Study provenance and commercial context
Longitudinal observational study published in Neurobiology of Disease in 2023. The full paper was reviewed from Seagull's source collection. T&W Engineering provided funding alongside foundations and research funding; company-affiliated authors contributed. The authors reported that funders did not determine data collection, analysis, interpretation, writing, or publication. Full citation.
Insight 02 | Epilepsy study
Computer analysis may find signs of deep activity in scalp EEG.
What was tested?
HEAnet learned from scalp recordings taken at the same time as electrodes closer to a deep memory area in 51 people with temporal lobe epilepsy. Separate groups with and without epilepsy were also studied. The target was brief discharges from the hippocampus, not whole seizures. [2]
What did it find?
The software found signal patterns that experts could not easily see. At a stricter setting during sleep, about 90% of the software's detections matched the reference markings. However, the software found only 13% of all marked events. A highly reliable positive result did not mean the system found most of the activity.
What could it mean for dementia?
An existing scalp recording might contain more information than standard review can reveal. Researchers must test the method in each dementia type and measure false alarms, missed activity, and whether the result improves care.
What remains unproven?
The external datasets lacked simultaneous deep recordings. This study did not demonstrate dementia diagnostic accuracy or clinical benefit. Public and nonprofit funding and author commercial relationships were disclosed.
Seagull's perspective: Better computer analysis may reveal difficult signals in scalp EEG. Current evidence has not shown that software can recover every seizure that scalp electrodes fail to show.
Insight 03 | Epilepsy study
A sensor beneath the scalp can record for months.
What was tested?
The UMPIRE study placed recorders beneath both sides of the scalp in 26 adults with epilepsy and followed them for six months. [3]
What did it find?
Among eight people who had seizures visible on regular scalp EEG, all 25 of those seizures also appeared in the beneath-scalp recording. Fourteen were missing from the participants' seizure diaries. Reviewers knew when the scalp-recorded seizures occurred, so this was not a blind search for every seizure.
What could it mean for dementia?
Months of recording could provide a more complete event history. After two device failures were excluded, usable data were available about 53% of the time on average. One failure involved cognitive difficulty using the device. Placing a sensor beneath the scalp did not guarantee continuous usable information.
What remains unproven?
Dementia-specific performance, benefit, and detection of scalp-invisible seizures were not established. Scalp pain or headaches occurred; no serious event was attributed to the device or implantation. Epiminder sponsored the study; author financial relationships were disclosed.
Seagull's perspective: A longer recording window and the help required to maintain it must be evaluated together.
The dementia opportunity
Better detection should lead to an answer that improves care.
Future research must connect the electrical finding with the person, the dementia type, and an outcome that matters. The questions below describe what researchers still need to prove.
Study the right population
Which dementia type, disease stage, and coexisting conditions were included? Can the method be used by people with language difficulties, movement symptoms, hearing aids, or substantial care needs? Compare the current subtype-specific EEG evidence.
Ask what the finding explains
Does recorded activity coincide with a change in response, language, memory, or sleep? Does it add information beyond the clinical history? A brief discharge, a seizure, and dementia-related background slowing need different interpretations.
Measure the difference in care
Does the result change an assessment or management decision? Are function, comfort, safety, or quality of life improved? Include false alarms, missed events, caregiver work, and access in the evaluation, not only the number of detected abnormalities.
Two reasons the opportunity deserves serious study
Direct evidence that some activity is missed
Deep recordings in five selected Alzheimer's cases found at least 95% of brief discharges were not visible on scalp EEG. Two seizures in one participant also lacked a visible scalp correlate. This establishes a blind spot in those cases, not a population-wide miss rate. [4]
Read the detection evidence in context →The next question is benefit
In the 34-participant LEV-AD trial, levetiracetam did not improve the primary cognitive outcome overall. Exploratory findings in participants with epileptiform activity supported further study. A way to detect activity is not, by itself, evidence that suppressing it improves dementia. [5]
Read the evidence on cognition and care →Sources and editorial scope
See the research behind each insight.
The first three sources support the featured technology discussions. The final two provide dementia context. This is a focused review of selected studies rather than a complete technology directory.
Musaeus et al. (2023). Detection of subclinical epileptiform discharges in Alzheimer's disease using long-term outpatient EEG monitoring. Neurobiology of Disease.
Full paper reviewed. Direct, selected Alzheimer's sample with controls; exploratory discharge detection, not a test of improved care.
Abou Jaoude et al. (2022). Noninvasive Detection of Hippocampal Epileptiform Activity on Scalp Electroencephalogram. JAMA Neurology.
Published methods, results, limitations, and disclosures reviewed. Epilepsy research.
Halliday et al. (2025). The UMPIRE study: A first-in-human multicenter trial of bilateral subscalp monitoring for epileptic seizure detection. Epilepsia.
Published methods, results, and disclosures reviewed. First-in-human safety and feasibility.
Devulder et al. (2025; online 2024). Epileptic activity on foramen ovale electrodes is associated with sleep and tau pathology in Alzheimer's disease. Brain.
Five selected cases with simultaneous deep and scalp recordings. Provides direct evidence of a detection limitation.
Vossel et al. (2021). Effect of Levetiracetam on Cognition in Patients With Alzheimer Disease With and Without Epileptiform Activity: A Randomized Clinical Trial. JAMA Neurology.
Treatment trial, not a device study. Distinguishes a negative primary outcome from exploratory subgroup findings.
How to use these insights
Prepared by Seagull Health. Study details reviewed . All featured papers fall within the preceding ten years. Their role here is to illuminate a research question, not to represent every newer publication in the field.
This page does not establish device availability, regulatory status, suitability for an individual, or a recommendation for testing or treatment. Read the evidence standards or suggest a correction or a study.
Help strengthen the atlas
Review the evidence. Bring the next question.
Explore Seagull's invitation for independent clinical review and focused research collaboration on recognition, dementia-specific detection, and outcomes that matter to people.