N1-Headache (formerly Curelator Headache) is one of the most research-grounded apps in the migraine space. Vincent Martin, Anne MacGregor, and other recognised migraine researchers have used it in clinical work, and the Curelator team has published more than 40 peer-reviewed papers on N-of-1 trigger analysis methodology.
Hermly is a different kind of tool. This page is an honest comparison — not “Hermly wins” marketing — laid out so you can choose the right one for what you’re actually trying to do.
What N1-Headache is
N1-Headache is built around N-of-1 trigger discovery: identify your personal triggers from your own data, using multivariate statistical analysis. The app asks you to log 70+ daily factors across mood, sleep, lifestyle, diet, cycle, medications, and environment. After enough data accumulates (typically 90 days), the analytics surface per-trigger odds ratios — “based on your data, your migraine probability is 2.3× higher on days you had X”.
The methodology comes from the published Peris-MacGregor-Donoghue line on N-of-1 headache analytics. It treats each user as a single-person cohort and computes statistical associations between candidate triggers and attack frequency.
Strengths:
- Deep clinical lineage. Methodology published in Cephalalgia, Headache, and similar peer-reviewed venues.
- Per-user trigger discovery is genuinely useful — if you have reliable triggers, this app will surface them.
- Strong reputation in the headache-specialist community.
Trade-offs:
- Logging burden is heavy. 70+ daily factors for 90 days is real effort.
- Server-side analytics means your data is uploaded for processing.
- Output is descriptive (what triggers you have), not predictive (what’s likely tomorrow).
What Hermly is
Hermly is a 24-hour forecast app. The output is a 0–100 risk number for today, computed from 26 features pulled mostly from sensors (HealthKit + WeatherKit + cycle data) and your own attack history. The optional self-report is a single 1-tap stress prompt, not a 70-item daily survey.
The core architecture difference: Hermly’s prediction runs on your iPhone. The cohort model ships with the app, the personalised layer trains on-device from your own labelled days, and raw HealthKit values never leave the device. Compare with N1-Headache’s server-side analytics.
Strengths:
- Very low daily friction. Sensors do most of the work.
- On-device privacy — raw health data stays on your iPhone.
- iOS-native surfaces: widgets, Lock Screen, Standby Mode, Live Activity, Apple Watch, Action Button.
- Methodology transparent — see /how-it-works.
Trade-offs:
- The output is one number plus three factors — less detail than a per-trigger odds-ratio report.
- iOS-only. No Android.
- Newer app; smaller user base than N1-Headache.
Where they solve different problems
| Question | Better answered by |
|---|---|
| ”What are my personal triggers?” | N1-Headache |
| ”What’s my risk today?” | Hermly |
| ”Should I take medication earlier?” | Neither — your doctor |
| ”Will pressure-drop weeks be worse for me?” | Both, different ways |
| ”Can I see this at a glance without opening an app?” | Hermly (widgets, Watch) |
| “Can I show a clinician detailed per-trigger associations?” | N1-Headache |
The honest framing: discovery and forecasting are different problems. The same person at different stages of their migraine journey might want different tools.
Specific feature comparison
| Feature | N1-Headache | Hermly |
|---|---|---|
| Daily logging burden | High (~70 factors) | Low (sensors + optional 1-tap stress) |
| Personalised analytics | Yes, server-side | Yes, on-device |
| Output | Trigger associations | 24h risk forecast + factor breakdown |
| Methodology | N-of-1, peer-reviewed | XGBoost cohort + Bayesian per-user head |
| Apple Watch | No | Yes, dedicated app + complications |
| Lock Screen / Dynamic Island | No | Yes, four Live Activity regions |
| Standby Mode | No | Yes, dedicated widget |
| On-device privacy | Server-side analytics | All raw data stays on iPhone |
| Doctor report | Yes, association-focused | Yes, calendar + factor-focused |
| Platform | iOS, Android | iOS only |
| Pricing | Subscription | Free + Pro $49.99/yr |
Privacy: the architectural difference
N1-Headache’s analytics requires uploading your daily logs to its servers for processing. The trade-off is real: server-side analytics enables more complex statistical methods, but your data leaves your device.
Hermly’s prediction model runs on-device. The cohort weights ship with the app, the per-user head trains locally, and inference happens in milliseconds on your iPhone. Our servers see only subscription state (anonymous Apple transaction ID), never your attack records or HealthKit data.
Both architectures are defensible engineering choices. The question is which one matches your priorities.
Who should pick what
Use N1-Headache if you: want the deepest per-user trigger analysis available, are willing to log many factors daily, and prioritise rigorous statistical methodology over forecast outputs.
Use Hermly if you: want low-friction sensor-driven forecasting, value on-device privacy, are on iOS (especially with Apple Watch), or want native widget/Live Activity integration.
Use both if: you want both discovery and forecasting and don’t mind running two apps. They don’t directly conflict — Hermly reads HealthKit; N1-Headache reads its own logs.
Neither will diagnose, treat, or prevent a migraine. Both are informational tools that make your patterns more legible.