Refractory Epilepsy
Your patient's seizure diary says '3 this week.' It doesn't say which one happened in the shower.
Refractory epilepsy affects one-third of all epilepsy patients — people who've tried 2+ medications and still seize. They manage 3-5 drugs simultaneously, carry rescue medications everywhere, and organize their entire lives around seizure risk. Yet the data that reaches their neurologist is a paper diary count and a 15-minute visit every 3-6 months. The daily granularity needed to evaluate novel therapies, detect cluster patterns, and prevent SUDEP goes uncaptured. Forma captures this signal.
Forecast
Based on prior Topics logging
Topics
How it works
15 seconds of speech. 25 data points.
Forma turns each patient into a continuous data stream. By voice or text, patients log seizure events with type, duration, triggers, and post-ictal state, medication doses and side effects across 3-5 drugs simultaneously, rescue medication use with response tracking, VNS/RNS device interactions, injury events, and quality-of-life impacts in under a minute. No forms, no recall bias — and voice is essential because some patients have post-ictal confusion that makes typing impossible. The result is a structured, longitudinal dataset that pairs seizure frequency with medication timing gaps, cluster patterns with trigger profiles, and device-assisted aborted seizures with progression trajectories — visible to the neurologist before the next visit.
“”
Forecast
Based on prior Topics logging
Topics
Structured, coded, longitudinal
- Timing
- Seizure time:6:00 AMSeizure duration:45 secRecovery time:20 minMissed dose time:Previous evening
- Location
- Injury location:Tongue
- Severity
- Injury:Tongue bitePost-ictal symptom:Confusion
- Actions
- Adherence:Missed doseMedication:Keppra
- Environmental
- Barometric pressure:29.82 inHg (falling)Moon phase:Waxing gibbous (87%)
From data to insights
What the data reveals
Daily voice logs compound into actionable insights — for patients managing their condition and for researchers running studies.
For Patients
What's triggering my seizure clusters?
Your clusters follow a pattern: 3 of 4 happened within 4 hours of a missed or late Brivaracetam dose. The fourth was catamenial — day 2 of your cycle. No clusters occurred during weeks with perfect medication timing. Setting a second alarm may prevent most of these.
Cluster triggers
Is the VNS actually helping?
You swiped the magnet 11 times this month. It fully aborted 5 seizures, shortened 4, and had no effect on 2. Success rate is highest when you swipe during the aura (80%) vs. after progression starts (25%). Your abort rate is improving — it was 27% three months ago.
VNS abort rate
Are my new medications working?
Since starting Cenobamate 6 weeks ago, your monthly seizure count dropped from 14 to 6 — a 57% reduction. Focal aware seizures dropped most. The dizziness from titration peaked at week 2 and is now minimal. Clobazam is helping with clusters but causing morning sedation.
Monthly seizures
Summarize this month for Dr. Park.
Total seizures: 6 (down from 14 baseline). One cluster day 7 requiring rescue diazepam — linked to missed Brivaracetam. VNS abort success: 45%. Catamenial seizure day 14 confirms hormonal pattern. Longest seizure-free streak: 5 days. Cenobamate dizziness resolving. Quality of life: went out alone for first time in weeks.
For Researchers
How does polypharmacy burden correlate with outcomes?
Patients on 4+ concurrent AEDs report side effect burden scores 2.8x higher than 2-drug regimens, but seizure reduction plateaus after the 3rd drug (median 12% additional reduction). Voice-logged side effects identify the culprit drug in 74% of cases within 7 days of a dose change.
Drug count vs. burden
Trend
What predicts rescue medication need?
Missed AED doses precede 62% of rescue medication events (median latency: 3.4 hours). Catamenial timing accounts for 18% of female rescue events. Sleep deprivation >2 hours below baseline predicts next-day cluster risk (OR 2.9). Daily logging captures these patterns 3 months before paper diaries.
Rescue predictors
Trend
What is the real-world VNS efficacy?
Across 31 VNS patients, magnet-swipe abort rate averages 38% (range 12–71%). Aura-timed activations succeed 3.2x more than post-progression swipes. Patients who log daily show 22% higher abort rates at 6 months — likely from improved aura recognition and faster response.
Abort success trend
Trend
How does seizure control affect quality of life?
Each additional seizure-free day per month correlates with a 0.8-point improvement in QoL composite (p<0.001). But the relationship is non-linear: the first seizure-free week drives 60% of QoL gains. Driving eligibility (typically 6-month seizure-free) is the strongest single QoL predictor (β=0.47).
Seizure-free vs. QoL
Trend
Two focal aware seizures today — one at my desk, one making dinner. Déjà vu and stomach rising, about a minute each. Cenobamate titration to 200mg started this morning.
Dizzy and exhausted from the Cenobamate increase. Slept 12 hours and still dragging. No seizures though.
Missed my morning Brivaracetam — was rushing for a work call. Had a focal impaired awareness 3 hours later. Lost about 5 minutes, confused after.
Felt one building and swiped the VNS magnet immediately. Tingling started but it never went to a full seizure. First successful abort in weeks.
Cluster — three focal seizures in 2 hours. Took rectal diazepam after the second. Stopped the cluster but wiped out the rest of the day. Scared.
Telehealth with Dr. Park. Cluster concerned her — adding Clobazam 5mg at night as bridge while Cenobamate titrates. Now on 4 medications.
Clobazam knocked me out. Slept through my alarm, brain like mud. But no seizures for 3 days. First 3-day streak in months.
Period started. Catamenial pattern — always worse around my cycle. One focal aware at bedtime. Used VNS magnet, didn't fully abort but shortened it.
5 seizure-free days. Cenobamate dizziness finally fading. Walked to the coffee shop — first time I've gone alone in weeks. Felt almost normal.
Had a focal impaired awareness while on a video call with my mom. She saw the whole thing. I don't remember 10 minutes. Breaks the streak but it's still better.
Follow-up with Dr. Park. She pulled up my logs — said the data shows 60% seizure reduction since Cenobamate started. The cluster-to-medication gap on day 4 confirms the Brivaracetam timing matters. Keeping current regimen.
Key insight
Over three weeks, Jessica's daily logs reveal what a seizure diary count cannot: a missed Brivaracetam dose on day 4 that triggered a seizure within 3 hours, a cluster on day 7 that required rescue medication, and a catamenial pattern on day 14 that confirms hormonal triggers. The Cenobamate titration response is visible in the data — from 2 seizures/day at baseline to a 5-day seizure-free streak by day 17, with a 60% overall reduction her neurologist could quantify before the next scheduled visit. The VNS abort on day 5 provides device efficacy data no diary captures.
See Refractory Epilepsy configured for your study
See Forma configured for Refractory Epilepsy