Rabits for Clinicians — Effort, measured. Not inferred.

The gap

The nights your patient won't wear a Holter.

A PSG gives you one night under observation. A wrist wearable gives you thirty nights of an inference. Rabits gives you thirty nights of a measurement, from a module on the nightstand that the patient never has to remember.

Question

Wearable / single-night PSG

Rabits

Is the event central or obstructive?

Wearable: cannot distinguish. PSG: yes, with belts.

Effort measured directly — absence of effort is central

Will the patient comply for 30 nights?

Adherence falls sharply after the first week

Nothing to wear, nothing to charge, nothing to remember

Is the HRV real or sensor noise?

Reported as-is, noise floor unstated

Jitter variance subtracted; suppressed when noise dominates

Can I get the raw signal?

Vendor-locked summary metrics only

Full CSV export, no portal, no account

Direct

Effort, not a proxy

Chest-wall displacement is the measurement. Effort cessation ≥ 10 s is a central apnea — unambiguous, with no oxygen dip to reason backwards from.

De-biased

The noise floor is subtracted

Beat-timing jitter adds in variance to every interval statistic. Rabits removes σ² from SDNN and 2σ² from RMSSD before reporting, and reports nothing once noise exceeds half the variance.

Paperless

The night arrives as a file

Every session exports as a timestamped CSV straight from the patient's phone. No printouts, no scanning, no vendor portal to log into.

Data & export

From the bedroom to your inbox.

The patient records the night and taps share. The file lands wherever you already work — email, secure message, EHR attachment, or straight into R or Python. No paper, no portal, no proprietary viewer.

Patient recordsOne tap. The module runs overnight unattended.
App writes a CSVTimestamped, on the device. Nothing is uploaded anywhere.
Patient shares itNative iOS share sheet — email, Files, secure messaging, whatever you both already use.
You open it anywherePlain CSV. Excel, R, Python, MATLAB. No licence, no login, no lock-in.

CSV

rabits_2026-07-14_0243.csv

72 KB

t_s	hr	rr	effort	gate
0.000	62.4	14.1	0.812	open
0.040	62.4	14.1	0.809	open
0.080	62.3	14.0	0.804	open
0.120	—	—	0.061	motion
0.160	—	—	0.058	motion
0.200	61.9	13.8	0.798	open

The validity gate is a column, not a filter. You see exactly which samples were suppressed and why — nothing is silently interpolated.

iOS app

What the patient sees. What you get back.

Built in the visual language of Apple Health, so the patient already knows how to read it — and every figure carries the caveat it earned.

ConnectBLE pairing, no account

DaytimeResting HR and respiration trend

WaveformsCarotid pulse and respiration traces

SleepStaging, REI, event log

Hardware

One module. Nightstand to desk.

24 GHz CW radar — sub-millimetre chest-wall displacement
0.3–0.8 m — aimed at the sternum, through clothing and bedding
BLE to iPhone — all processing on device, nothing leaves the patient's phone
No camera, no microphone — radar sees motion, never identity
Carotid contact mode — arterial pulse morphology, not just a beat count

The boundary

What we don't claim.

You will ask this in the first five minutes, so it is on the page instead of in a footnote.

Measured

Respiratory effort, from chest-wall displacement, in real time
Central apnea — effort cessation ≥ 10 s, unambiguous
Heart rate, respiration rate, breath-interval variability
RSA regressed on true respiratory phase, not estimated from the tachogram

Not claimed

Airflow — obstructive events are inferred from arousal. A surrogate, sensitivity unknown until PSG.
SpO₂ — no optical path. Output is REI, effort-based; it is not an AHI.
Hypopnea per AASM — requires a ≥3% desaturation we cannot see.
Bed-partner rejection — CW has no range resolution. One body in the beam.
Diagnosis — Rabits is a screening and monitoring tool. Any flag requires ECG or PSG confirmation.

On validation: beat-timing jitter σ is currently reported as a conservative upper bound derived from the tachogram's second difference, and every HRV figure is flagged provisional until σ is calibrated by Bland–Altman against a reference ECG. AF thresholds are literature-derived and have not been calibrated against ECG on this device. We would rather tell you that than let you find it out.

Effort, measured.Not inferred.