Safety
SUDEP happens after convulsive seizures nobody saw. Nocturnal supervision is one of the few known protective factors — and it is exactly what a device could provide.
The problem
Why this project
Epilepsy in numbers, the gaps in care in Canada and British Columbia, and what a seizure detector that works at home would change.
Epilepsy in numbers
Every figure below is from the cited source; nothing is rounded up.
Global burden
50 millionpeople worldwide have epilepsy — one of the most common neurological diseasesWHO fact sheet, 7 Feb 2024
≈ 80 %of them live in low- and middle-income countries; three quarters in low-income countries get no treatmentWHO 2024
up to 3×higher risk of premature death than the general populationWHO 2024
up to 70 %could live seizure-free if properly diagnosed and treated — diagnosis and monitoring are the bottleneck, not the drugsWHO 2024
≈ 1 in 3newly diagnosed patients develop drug-resistant epilepsy — the group that needs monitoring and surgeryCouper et al., CJNS 2024
Canadian access
≈ 300,000Canadians live with epilepsy (about 1 %); about 14,000 new diagnoses a year, 38 a dayEpilepsy Canada / provincial associations
> 50,000people in British ColumbiaBC Epilepsy Society, cited by SFU, Sept 2025
7 of 13provinces and territories have no epilepsy monitoring unit (EMU) beds at allCouper et al., CJNS 2024
0.83EMU beds per 1,000 people with refractory epilepsy in BC — the lowest of any province that has beds (Nova Scotia: 2.5)Couper et al., CJNS 2024
Unwitnessed seizures
≈ 1 in 1,000adults with epilepsy die of SUDEP each year (sudden unexpected death in epilepsy); most deaths follow an unwitnessed convulsive seizure, usually at nightAAN practice guideline, Harden et al. 2017; Langan 2005
> 50 %of seizures are missing from patients' own diaries — almost all of those during sleep — yet medication is adjusted on those diariesHoppe et al., Arch Neurol 2007
≈ 57 %of seizures are not recognised even by a family member or carer in the same roomEpilepsia 2025, witness-recognition study
Four gaps, one cause
Safety, treatment, access and trust all fail at the same point: nobody is counting the seizures.
Treatment
Neurologists adjust medication on diaries that miss more than half of all seizures. A wrong count means a wrong dose, for years.
Access
The only reliable count comes from video-EEG in an epilepsy monitoring unit. BC has 0.83 beds per 1,000 refractory patients; seven provinces and territories have none. Patients wait months to over a year, and sometimes the seizure does not happen during the stay.
Trust
Consumer seizure watches sense movement, so they catch convulsions and miss the other 94 % of focal seizures, and their false alarms lead families to switch them off.
ICTRA's question follows from the gaps: can two behind-the-ear EEG electrodes, a chest ECG and an arm EMG, read by one model, count seizures at home with clinically useful sensitivity and fewer than one false alarm a day — on patients the model has never seen?
In the news and the literature
Screenshots of the actual pages, captured 17–18 October 2026, each linked to its source.
The global baseline: 50 million people, up to 70 % treatable, risk of early death up to three times higher. who.int ↗
"Epilepsy affects more than 50,000 people in B.C." Pre-surgical video-EEG "requires patients to stay in the hospital for extended periods … and sometimes seizures don't occur in the hospital environment." sfu.ca ↗
Seven of 13 provinces and territories have no EMU beds; BC has 0.83 beds per 1,000 refractory patients, the lowest where beds exist. cambridge.org ↗
Consumer devices track movement and heart rate: useful for convulsive seizures, blind to non-motor ones. epilepsy.com ↗
125 patients, 11,626 hours, 883 seizures, behind-the-ear EEG + ECG + EMG + movement, released openly (CC0). The first dataset that makes a project like this possible. nature.com ↗
Behind-the-ear EEG + ECG: 0.73 sensitivity but 149 false alarms a day on its own; 0.31 after a human reviewer. The published state of the art for this device class — our reference point. pmc.ncbi.nlm.nih.gov ↗
28 algorithms, blind-tested on 65 unseen patients: the best reached F1 32 %; self-reported scores of 80–90 % did not survive. Why ICTRA locks its test set. arxiv.org ↗
What ICTRA would change
For a family
A device that costs a few hundred dollars and sits behind the ear, on the chest and on the shoulder wakes someone about half a minute into a convulsive seizure — and does not cry wolf ten times a night. A tonic-clonic seizure usually lasts one to three minutes, so half a minute still leaves time to reach the person. Today that witness is a hospital bed or a watch that only knows convulsions.
For a neurologist
A seizure count backed by signals instead of memory, with an honest statement of what the device can and cannot see by seizure type — the difference between a monitoring tool and a gadget.
For the field
Measure the brain–heart–muscle timing first, then design the fusion; let a multi-signal model teach a single-sensor device; test on locked and external patients and publish every number, including the failures. See dead ends and where ICTRA stands.
ICTRA is a student research prototype. It is not a medical device and must not replace prescribed monitoring or guide medication.
Sources for this page
- World Health Organization. Epilepsy fact sheet, 7 February 2024. who.int
- Couper RG, Antaya T, Nguyen DK, Burneo JG. Disparities in drug-resistant epilepsy care in Canada. Can J Neurol Sci 2025;52(1):42–43 (online 11 April 2024). doi:10.1017/cjn.2024.57
- Simon Fraser University. Innovating epilepsy care: SFU study explores advanced brain imaging for B.C. patients. Media release, 22 September 2025 (BC Epilepsy Society figure).
- Epilepsy Canada and provincial associations: ~300,000 Canadians; ~14,000 new diagnoses per year.
- Harden C et al. Practice guideline summary: Sudden unexpected death in epilepsy incidence rates and risk factors. Neurology 2017;88:1674–1680 (≈ 1.2 per 1,000 adult patient-years). Langan Y et al. Neurology 2005 (case-control, SUDEP and supervision).
- Hoppe C, Poepel A, Elger CE. Epilepsy: accuracy of patient seizure counts. Arch Neurol 2007;64:1595–1599.
- Witness recognition of focal and generalized seizures. Epilepsia 2025. doi:10.1111/epi.18499
- Epilepsy Foundation. An update on SmartWatch. New therapies pipeline.
- Bhagubai M et al. SeizeIT2. Sci Data 2025;12:1228. Swinnen L et al. The SeizeIT2 study. Epilepsia Open 2026;11(3):883–894. Dan J et al. SzCORE challenge. arXiv:2505.18191.