Who we serve · Research institutions
Instrument-grade data, and the raw signal underneath it.
A device that only gives you its conclusions is not an instrument, it is an opinion. Every Nexaform device will hand you the signal it started from, along with everything you need to explain where it came from.
What study teams run into
Usually somewhere between the grant application and the second reviewer.
Devices that only report conclusions
A number arrives with no way to see what produced it. When a result is questioned, the instrument is a black box and the answer is a shrug.
Timestamps that do not line up
Two devices, two clocks, and a drift of a few hundred milliseconds that nobody notices until an analysis depends on the ordering of two events.
Export formats built for dashboards
A CSV of five-minute averages is fine for a clinic view and useless for a study that needs the waveform.
Data handling that does not fit consent
Retention, deletion, and export rules written for a clinical product, applied to a study whose ethics approval says something different.
What we bring
The features on this list exist because researchers asked for them, occasionally quite firmly.
Study mode firmware
A device configuration intended for research rather than routine care, enabled per device group.
- Full-rate raw signal retained on device
- On-device processing can be disabled entirely
- Fixed firmware version pinned for the study duration
- No silent updates mid-study
Raw signal access
Everything the front end recorded, not a summary of it.
- 512 Hz 16-bit ECG, 128 Hz accelerometry
- Unfiltered and filtered streams both available
- Signal-quality flags preserved, not applied
- Gaps reported explicitly, never interpolated silently
Time synchronisation
Because a study with two data sources is really a study about clocks.
- PTP synchronisation to ±1 ms across a site
- Per-recording synchronisation quality attached
- Monotonic device clock separate from wall clock
- Drift measured and recorded, not assumed
Open APIs and bulk export
Get the data out in the shape your analysis already expects.
- REST and bulk export APIs, documented and versioned
- EDF+, Parquet, CSV, and raw binary with a reader
- Scheduled export to your own storage
- No per-record or per-query charge
Consent-aware handling
Retention and deletion that can be told what the ethics approval says.
- Per-participant retention rules
- Withdrawal removes data across every store, provably
- Study-scoped access, separate from clinical access
- Audit log suitable for a monitoring visit
Reproducibility
Every record can explain the instrument that made it.
- Device serial and firmware build hash on every record
- Calibration record and last calibration date attached
- Front-end gain and filter settings recorded
- Configuration changes versioned and timestamped
What the data actually looks like
The details that decide whether a dataset is usable for a given analysis.
| Capability | What we support today |
|---|---|
| ECG | 512 Hz, 16-bit, single lead, filtered and unfiltered streams |
| Accelerometry | 128 Hz, 3-axis, ±8 g |
| Temperature | 1 Hz skin contact, 0.1 °C resolution |
| Export formats | EDF+, Parquet, CSV, raw binary with an open reader |
| API | REST, versioned, plus a bulk export endpoint for whole studies |
| Time sync | PTP to ±1 ms; per-recording quality metric attached |
| Provenance | Serial, firmware hash, calibration record, front-end configuration |
| Retention | Per-study, per-participant; withdrawal propagates and is auditable |
| License on our own published data | CC BY 4.0, no citation of Nexaform required |
We asked people whether their heart data could be given away, and a surprising number said yes, provided we said plainly who would get it and what they could do with it.
Dr. Naomi Achterberg
Cardiovascular research group, Cascade University
Products that show up in this work
The same four families, put together differently depending on what the deployment has to survive.
Writing a protocol?
It is much cheaper to find out what the device can record before the protocol is written than after. We would rather have that conversation early, including the parts where the answer is no.