The Science Behind Sleepal

Sleepal Reads Breath

Wearables track motion. Sleepal tracks breathing.

Sleepal reads the signals your body actually sleeps by - your breath and your heartbeat - not the wrist motion a watch or ring relies on.

Watches + Rings

Motion-based

Wrist or finger micro-movement + PPG heart rate indirectly signal lying still at sleep.

Signal source
Wrist/finger motion
Body contact
Must wear to bed
Charging
Daily or weekly
Sleep staging accuracy (vs hospital PSG) 55-79% Varies by device and sleep stage

Sleepal AI Lamp

Breathing-based

60GHz mmWave radar reads chest movement at direct signal-the same one doctors trust most.

Signal source
Chest breathing pattern
Body contact
None. Just sleep.
Charging
Always on (plugged in)
Sleep staging accuracy (vs hospital PSG) 80% Trained on 2,000+ clinical PSG nights

Accuracy, Proven

Why a wearable can't see what matters

Most trackers guess your sleep from how much your wrist moves. But Sleep's real signals aren't in your hands — they're in your breath and heartbeat. Sleepal reads those directly, without touching you. Wrist movement is a proxy. Breath and heartbeat are the source.

The Breakthrough of Sleep clinical paper preview

Sleepal’s clinical study achieved Kappa scores[1] exceeding top-tier wearables, as demonstrated in our peer-reviewed study accepted for publication.

Sleepal
0.695
Apple Watch
0.680
Oura Ring
0.650
Research link: doi.org/10.1088/1361-6579/aeb104 * Cohen's Kappa coefficient is the widely accepted accuracy metric in sleep staging research. It measures the agreement between a model's predictions and ground truth, adjusting for chance agreement that inflates simple accuracy. A score of 0.61-0.80 indicates substantial agreement, 0.81-1.00 indicates almost perfect agreement.

Validated by globally recognized sleep expert

Most trackers guess your sleep from how much your wrist moves. But Sleep's real signals aren't in your hands — they're in your breath and heartbeat. Sleepal reads those directly, without touching you. Wrist movement is a proxy. Breath and heartbeat are the source.

Prof. Dr. Thomas Penzel

Prof. Dr. Thomas Penzel

IEEE Fellow

President of World Sleep Society

Scientific director of the Interdisciplinary Center for Sleep Medicine at Charite - Universitatsmedizin Berlin (Germany)

Dr. Thomas Penzel co-authored the paper and comments as this

This study demonstrates the strong potential of a large and innovative AI model for advancing contactless sleep monitoring. It suggests a promising avenue for future research and applications in sleep and home health, and I look forward to our continued academic collaboration in this field.

How Sleepal Works

How Sleepal sees your sleep

Inside every Sleepal AI Lamp is a 60 GHz millimeter-wave radar — resolution down to 0.1 mm. Precise enough to sense your chest rise and fall with each breath, even with the heavy blanket. Part of the signal reaches deeper, picking up your heartbeat.

Three-up feature row:

  • Contactless, all night. Nothing to wear, charge, forget, or lose.
  • No camera, ever. Radar and thermal sensing only.
  • Safe nightly. Low-power mmWave radar, FCC-compliant.
Sleepal bedside radar sensing while a person sleeps

The Sleep AI

Raw breathing and heart-rate signals are only the start. Reading them as accurate sleep stages takes a purpose-built model: a deep neural network with nearly 100 million parameters, trained on 2,000+ nights of clinical polysomnography (PSG) from leading hospitals.

PSG is the gold standard for sleep staging — the physician-scored overnight test sleep medicine treats as ground truth. By recording with Sleepal and PSG on the same sleepers at the same time, we taught the model the breathing and heart-rate signatures behind each stage: Awake, Core, REM, and Deep.

Stat strip:

  • 2,000+ clinical PSG nights in training
  • ~100M parameters in the sleep model
  • 4 stages scored nightly: Awake · Core · REM · Deep
Sleep AI model architecture diagram

Beyond Sleep Stages

the whole night, the whole room

Stages are just the start. Sleepal uses a suite of sensors to capture everything that shapes a night — and reads them together.

Posture. A thermal-array sensor tracks how you lie — back, side, or front — the only consumer sleep tracker that quantifies position this way. Temperature & humidity. Whether your room is helping or hurting your sleep.

Ambient light. Stray light that pulls you out of deep sleep, made visible. Sound. Snoring and noise, flagged as factors in your night. Multi-person separation. Share a bed? Sleepal locks onto the sleeper closest to it and filters out a partner's signals — your report stays yours.

By measuring the body and the room, Sleepal connects cause to effect — not just how you slept, but why.

Sleepal sensors map showing acoustic, body rhythm, and thermal awareness