Across 101 patients on a neurology clinic dataset, our system identified neurodegenerative impairment with 92% accuracy. The underlying physiology pipeline was tested on 118 people, published in IEEE, and now has formalized clinical collaborations underway at Harvard Medical / MGB and UCI Health.
Output
neurodegenerative impairment
92%
accuracy
101
patients
Tested across six gait conditions on neurology clinic data.
92%
accuracy
Detecting neurodegenerative impairment, healthy versus impaired.
118
people
IEEE-published study on WiFi vital-sign sensing, 0.14 BPM error.
9
patents
One U.S. provisional filed; eight more drafted, ready to file.
MGB
+ UCI
clinical studies
Formalized clinical work underway with Harvard Medical / MGB and UCI Health.
Manifesto
Neurodegenerative decline does not arrive all at once.
It leaks out slowly: a shorter stride, a slower walk, fragmented sleep, time spent motionless, a fall that should have been noticed earlier.
While clinics see snapshots, the home sees time.
Every breath, heartbeat, step, fall, and shift in posture changes the wireless channel around us.
PulseFi turns those changes into passive health data without needing cameras or wearables.
A neurologist sees a patient under artificial conditions, for minutes at a time. Decline happens between visits.
The people who most need monitoring are often the least able to charge, wear, trust, and remember another device.
They may see motion, but they turn the home into a surveillance product. PulseFi starts with the wireless field instead.
What we proved
We started with heart rate because it would prove that WiFi had the capabilities we needed and provided clinically useful data. If we could recover cardiac micro-motion from $10 commodity devices, the signatures the body was leaving behind contained vast amounts of unfound data. Everything else followed.
We then used that same architecture for breathing rate and apnea, and it is what will drive the falls, gait impairment, and pauses ahead.
Simultaneously, we want to verify this technology at a clinical level — which is why we have formalized studies underway with Harvard Medical / MGB for post-cardiac ICU monitoring, and with UCI Health for long-term psychiatric patient monitoring.
Clinical wedge
The first sign of decline should not be a fall.
The system
Commodity WiFi signals move through a room and are modified by the body. This includes breathing, cardiac micro-motion, posture, walking, falls, and far more. The home is already full of usable signal.
Signal processing turns the raw channel-state information into proxies for cardiac, respiratory, motion, and gait — all sourced from a single-antenna $10 ESP32.
Lightweight ML infers heart rate, breathing, apnea, falls, gait impairment, and changes from a person's own baseline. Deployed to a near-clinical resolution, not just consumer grade.
The goal is for neurodegenerative decline to become continuous functional monitoring — noticing slow decline weeks before a crisis would force the family to.
Roadmap
Now
Heart rate, breathing rate, apnea
Next
Falls, sleep staging, gait impairment
Then
Recovery, time-on-ground, baseline drift
Long-term
Continuous functional decline monitoring
Why neurodegenerative decline
Parkinson's, ALS, MS, dementia-related decline, and other neurological conditions affect movement before a person becomes acutely unsafe. Stride length shortens, cadence slows, asymmetry increases, and sleep quality drops. Yet nobody measures these changes.
Neurologists have known this for decades. The reason early signs are still missed isn't medical knowledge. It's that there is no ambient, continuous, no-effort way to watch how a person actually walks across their own living room over six months.
PulseFi makes the home itself a longitudinal sensor. The goal is to give clinicians a useful signal layer they have never had access to.
Design constraints
What $100K does
01
File the eight pending provisional patents protecting the gait pipeline and the downstream applications it enables.
02
Run the formalized clinical studies with Harvard Medical / MGB and UCI Health, and buy nine months of full-time work to do it.
03
Run our own independent 100+ patient cohort to better understand the neurological design space and improve multi-class disease ID.
04
Build the hardware and software container that makes the platform deployable in the home.
Built by
Pranay Kocheta — primary inventor on PulseFi's nine provisional patents, IEEE-published author on the underlying WiFi vital-sign sensing work.
Nayan Bhatia — PhD candidate in wireless sensing at UC Santa Cruz, working on the channel-side foundations of the platform. 6+ publications in wireless sensing.