Continuous, at home. Nothing to book, nothing to send away.
Reads flow, not a guess
Optical sensing against your own baseline — not a single snapshot compared to an average.
Responds automatically
Not just a number on a screen. A capped, low-current nudge, only when it's warranted.
Checked against sham
Some sessions carry no current at all — so a good week is measured, not assumed.
Your brain doesn't work in snapshots.
Marrow continuously tracks subtle physiological signals to understand how your state changes throughout the day.
01
Live, not a snapshot
Oxygenation, pulse and breathing update as you go, against your own baseline.
02
Weak spots shown too
One Core Score, with sub-scores underneath — including the low ones.
03
Every session logged
Before and after, with real and sham sessions kept apart.
9:41●●●
TODAY
Good afternoon
LIVE
72
CORE SCORE
↑ 4 from your baseline
Recovery is strong. Sleep readiness is holding you back.
HbO2 · PREFRONTAL72 bpm
Focus
78
Recovery
84
Load
42
Sleep readiness
48
LAST SESSION · REAL
61→74
Compared with sham sessions
Home
Trends
Sessions
Device
ILLUSTRATIVE DATA
What did that session actually change?
A closed-loop headband — that shows its working.
SESSION LOG
Every nudge, checked against your own baseline
Each session records the before and after. Some sessions carry no current at all, so you can see what really moved — and what was just expectation.
SIGNAL
READING
QUALITY
HbO2 · prefrontal
Rising
High
Heart rate variability
Stable
High
Breathing rate
14 / min
Moderate
Head motion
Detected
Filtered out
01 / SENSE■□□□
Optical readCONTINUOUS
NIRRed2 photodiodes
Read through the skin
Two wavelengths of light, continuously. Oxygenation, pulse and breathing come from the same signal.
02 / INTERPRET■■□□
HbO2 shiftKept
Pulse rippleKept
Head movementFiltered out
Motion isn't mistaken for you
An accelerometer flags head movement, so it never reaches your Core Score. Everything is read against your own baseline.
03 / RESPOND■■■□
THRESHOLD CROSSED
A brief, low-current session
< 2 mAIsolated circuit
One nudge, when warranted
A separate, isolated circuit delivers a capped session — only after a validated threshold is crossed.
04 / MEASURE■■■■
Before → after, logged
RealSham
Did it change anything?
Every session is compared with your baseline — and with sham sessions you can't tell apart.
The app
Your Core Score, at a glance.
One number, with the sub-scores underneath — including the weak ones.
Home
Core Score
72
Focus78
Recovery84
Load42
Sleep readiness48
↑ 4 from your baseline
Live
Monitoring
Load↑
Recovery↓
Head motionFiltered
State changing
Session
Before
61
After
74
Logged
Counted against your sham sessions before it's called a response.
Hardware
Built to sense. Built to respond.
OPTICAL
One chip, two LEDs, two photodiodes. Oxygenation, pulse and breathing come from the same signal.
MOTION
An accelerometer flags head movement before it can be mistaken for a physiological reading.
STIMULATION
Low-current tDCS, capped under 2mA by a hard resistor — regardless of any software fault.
ISOLATED
Separate batteries, no shared ground. Only a pulse of light crosses between the two sides.
SENSING DOMAIN
Always on. Reads you.
Optical sensor, accelerometer, main battery.
OPTOISOLATOR light only
STIMULATION DOMAIN
Only when warranted. Adjusts you.
Its own battery, constant-current source, two electrodes.
Two systems. One headband.
Honest about depth
Marrow reads your prefrontal cortex — not your feelings.
Light reaches about 1–2cm into tissue. That's the outer surface of the prefrontal cortex: focus, decisions, load. Deeper structures like the amygdala sit around 7.5cm away, out of reach of any light-based sensor.
What it reads
Prefrontal oxygenation (HbO2 / HbR)
Heart rate and HRV
Breathing rate
Head motion, to filter noise
Chip temperature, for calibration only
What it doesn't
Emotions or fear, directly
Deep brain structures
A diagnosis of any condition
Anything from a single snapshot
Validation
We want to know if it works.
YOUR BASELINE
Sessions are compared with your own history, not a generic population average.
CALIBRATION
A breath-hold test forces a predictable oxygenation change, confirming the sensor works on you before it's trusted.
SHAM
Some sessions deliver zero current, indistinguishable from real ones — standard practice in clinical tDCS research.
Signal
Source
Status
HbO2 / HbR
NIRS, 2 wavelengths
Continuous
Heart rate / HRV
Same optical chip
Continuous
Breathing rate
Derived, same signal
Continuous
Temperature
Chip sensor
Calibration only
Stimulation current
Isolated circuit
<2mA
Marrow is a wellness product currently being developed and validated. It is not FDA or CDSCO cleared.
Exploring
What we're exploring.
Focus & fatigue
Cognitive load
Sustained attention
Fatigue onset
Working memory load
Task-switching cost
Vigilance decline
Recovery windows
Stress & load
HRV shift detection
Stimulation response
Baseline drift
Strain spikes
Recovery half-life
Real vs. sham sessions
Session comparison
Post-exertion recovery
Oxygenation return
Time to baseline
Load tolerance
Recovery consistency
Exertion-to-rest ratio
Overtraining signals
Session-to-session trend
Evening wind-down
Pre-sleep breathing
Pulse deceleration
Stimulation windows
Evening HRV shift
Sleep-onset readiness
Nightly baseline reset
Areas we're investigating, not promises. Availability depends on validation holding up.
Closed-loop, with the receipts
Know your Core Score. See what moves it.
We're building Marrow. Join the early community and help shape what comes next.
No spam. Occasional notes on what we're learning. Privacy