Research & Evidence

Built on science.
Designed for life.

Nura doesn't compare your parent to a population average. It learns their normal — across sleep, movement, heart rate, heart rate variability, skin temperature and blood oxygen — and tells you when something has changed enough to matter.

53
peer-reviewed
studies referenced
7
signals Nura learns
and follows
24/7
passive
sensing
The hardware

Multiple sensors.
One picture that's actually yours.

Nura's sensors were chosen by working through the published research on what a wrist can reliably read — and what it can't. Each is a high-precision, research-validated component. No single sensor tells you much on its own; the value is in reading several together, over time, against one person's own baseline. From these same sensors, Nura derives a wider set of signals without adding hardware.

💓
Heart rate sensor

The foundational sensor. Uses light to read blood flow at the wrist, following heart rate and heart rate variability across the day and night. A single reading tells you almost nothing. Weeks of readings tell you what's usual for this person — and make a small, sustained shift visible.

🩺
Blood oxygen sensor

Reads blood oxygen saturation (SpO₂) at the wrist and learns each person's typical range. Nura reports changes relative to that personal range rather than against a fixed threshold.

🏃
Motion sensor

Follows movement in every direction. Builds a picture of daily activity, sleep timing, and the shape of a normal day — how much someone moves, when, and at what pace. For older adults the useful signal is rarely a step count; it's whether this week looks like the weeks before it.

🛡️
Fall detection

Recognises the distinctive motion signature of a fall — a sudden high-G impact, preceded by free fall or rapid rotation, followed by stillness. Machine learning confirms it's a real event rather than someone sitting down heavily. When Nura sees one, the family is alerted immediately, with location.

🌡️
Skin temperature

High-precision skin temperature, followed continuously. Nura learns each person's own temperature rhythm across days — including how it moves with their body clock — and notices sustained departures from it.

🔬
Multi-signal reading
The Nura approach

No single sensor gives you the whole picture, and any one of them alone produces noise. Nura's core technical work is in reading several signals together, over time, against one person's baseline — which is what the research consistently shows produces the fewest false alarms. Nura is built to say something only when several signals agree and the change persists.

Derived signals — inferred from the sensors above, no dedicated sensor required
Stress signals
Derived · no dedicated sensor

Nura has no stress sensor. It reads stress-related patterns from heart rate and heart rate variability — the beat-to-beat changes research ties to the body's stress response — and follows them over time, which is what separates a hard afternoon from a hard month.

💧
Hydration signals
Derived · no dedicated sensor

No wrist device can measure hydration directly, and Nura doesn't claim to. Nura reads hydration risk by looking at rising resting heart rate, falling heart rate variability and skin temperature patterns together. Thirst becomes less reliable with age, which is why a passive signal is useful here.

The evidence

What the research says
about what a wrist can read.

For each signal Nura follows, there's a body of peer-reviewed work establishing that it's measurable at the wrist and meaningful over time in older adults. That research is why we built Nura the way we did.

What these studies are, and aren't

The research below establishes what wrist-worn sensors can measure in principle, in the populations those studies examined. It is not a measurement of Nura's performance. Our own validation work is underway, and we'll publish it when we have it.

Why a baseline, and not a threshold

The most consistent finding across wrist-wearable research is that measurable physiological change — resting heart rate, sleep, activity — appears before a person feels different. The body shifts before anyone would think to look. That's why Nura is built around each person's own baseline rather than a set of thresholds: a threshold only fires once something is already obviously wrong, and by then the family usually knows. A baseline can see a change that's unremarkable for everyone else but unusual for this person.

Research context
Multiple large wearable cohorts have shown that resting heart rate, sleep and activity signals shift measurably in advance of subjective symptom onset. See the publications below.
Fall detection

The most thoroughly studied application in wearable health technology. A 6-axis motion sensor captures the signature of a fall — a high-G impact spike, preceded by free fall or rapid rotation, followed by inactivity — and machine learning distinguishes a real fall from someone dropping into a chair. Reviews of the literature find that validated systems consistently use accelerometer and gyroscope signals with machine-learning classification, which is the approach Nura takes.

Research context
JAMDA (2024), scoping review of 73 fall-detection studies. JMIR (2024), deep-learning framework for fall detection.
Sleep

Heart rate and motion together support detailed sleep tracking — timing, quality, stages, regularity and nighttime movement. Nura's screenless, all-day design helps here for an unglamorous reason: people actually keep a bracelet on at night, where a device that needs nightly charging comes off.

Research context
A sleep-staging algorithm validated on 1,522 nights of recordings from 1,430 participants using only heart rate and motion (ScienceDirect, 2024). A 35-article review found heart-rate-plus-motion outperforms motion alone for sleep staging (npj Digital Medicine, 2024).
Hydration risk

No wrist sensor reads hydration directly. But changes in hydration produce measurable changes elsewhere in the body, and those are readable at the wrist: resting heart rate, heart rate variability, skin temperature and activity. Nura reads them together and notices when the combined picture shifts.

Research context
A 240-participant field evaluation established that hydration status produces measurable physiological change at the wrist (PLOS ONE, 2022).
Heart rate variability and stress

The body's stress response changes how the heart beats — faster, with less variation between beats. Nura's optical sensor reads those beat-to-beat intervals, which gives a non-invasive window into how someone's body is handling the day. Over weeks, it separates a bad afternoon from sustained strain.

Research context
HRV is well established in the literature as a physiological marker in older adults (Frontiers in Cardiovascular Medicine, 2025). Machine-learning approaches to reading stress from wrist PPG signals (PMC, 2025).
Temperature patterns

Nura follows wrist skin temperature continuously and builds a personal baseline over the first 7–14 days of wear. Rather than checking against a fixed number, it notices when someone's own pattern shifts and stays shifted — including changes in the daily rhythm itself.

Research context
Continuous skin-temperature measurement has been studied in older adult populations, including work showing that residents in care settings can apply a wearable temperature sensor without assistance (PMC, 2021).
Activity and movement patterns

Wrist accelerometers give validated activity classification, but for older adults the useful information is in the pattern rather than the count. A week of unusually quiet days says more than any single day's steps. Nura follows activity level, routine consistency and movement trends.

Research context
A 747-participant study drawn from the National Health and Aging Trends Study showed wrist accelerometer data supports machine-learning assessment of physical function in older adults (BMC Geriatrics, 2024).
The horizon

From awareness to
anticipation.

Everything above is what Nura does today, as a wellness product. Below is what we're researching — and being direct about it: these are not features Nura offers. Each would require FDA clearance before we could put it in front of a family, and pursuing that clearance is a deliberate part of our roadmap rather than something we've done.

The reason this matters for the people wearing Nura today: our hardware is already capable of supporting this work. Cleared capability, when it arrives, arrives as a software update — not a new bracelet.

Research
Fall risk

Fall detection reacts after a fall. Understanding fall risk means following walking patterns, activity, sleep and physical trends over time to see when someone is more vulnerable. There's a substantial research literature here. Turning it into something we tell a family requires clearance.

Research
Health trajectory

Research on multi-sensor wearables shows that changes across several signals — heart rate, heart rate variability, activity, sleep, temperature — often move together in the days before a significant health event. Reading that pattern reliably enough to act on it is an active research question, and one we're working on with our own population.

Core R&D
Building on data from older adults

Most published wearable research uses younger, healthier participants. Older bodies behave differently, and sensors behave differently on them. Building and validating on data specifically from adults over 70 is our central research priority — and the reason our own validation work matters more to us than anyone else's published numbers.

The publications

The research that shaped
our design.

The science behind Nura draws on peer-reviewed work across clinical medicine, medical engineering and aging research. These are among the findings that informed how we built it.

📖
npj Digital Medicine, 2024
Review of 35 articles and 62 wearable configurations: optical heart rate plus accelerometer outperforms accelerometer alone for sleep staging.
📖
ScienceDirect, 2024
Sleep-staging algorithm validated on 1,522 nights from 1,430 participants using heart rate and motion signals only.
📖
JAMDA, 2024
Scoping review of 73 fall-detection studies in older adults; validated systems predominantly use accelerometer and gyroscope signals with machine-learning classification.
📖
JMIR, 2024
Deep-learning framework for fall detection; model development and study design.
📖
BMC Geriatrics, 2024
747 participants from the National Health and Aging Trends Study: wrist accelerometer data supports machine-learning assessment of physical function in older adults.
📖
Frontiers in Cardiovascular Medicine, 2025
Heart rate variability as a physiological marker in elderly populations.
📖
Nature Scientific Reports, 2021
Accuracy of heart rate variability estimated from reflective wrist PPG in elderly vascular patients.
📖
PLOS ONE, 2022
240-participant field evaluation: hydration status produces measurable physiological change at the wrist.
📖
Frontiers in Digital Health, 2022
Review of wearable sensor systems for fall risk assessment.
📖
Apple Machine Learning Research, 2024
Foundation models for wearable biosignals; PPG and accelerometer models fine-tuned across multiple tasks.
📖
PMC, 2021
Wearable temperature sensors applied without assistance by residents in care facilities.
How we think about this

Science as the
foundation, not the sales pitch.

1
All the sensors together, not any one alone

The research is consistent: several signals read together outperform any single signal, and produce far fewer false alarms. We built that in as a design principle rather than adding it later. It's also why Nura stays quiet unless a change shows up across more than one signal and persists.

2
Your parent's normal, not a population average

"Normal" varies enormously between people, and published averages are mostly drawn from younger, healthier bodies. Nura learns each person's own patterns over the first 7–14 days before it says anything at all.

3
Signals, not diagnoses

We say "hydration risk," not "dehydration." "Sleep signals," not "insomnia." "A change in their temperature pattern," not "a fever." This isn't legal caution dressed up as principle — it's a more accurate description of what a wrist can actually know, and we'd rather tell you what we see than tell you what it means.

4
Nothing that pretends to be a doctor

Nura will tell you that something about your mother's week looks different from her usual. It will not tell you why, and it will not tell you what to do about it. That's a conversation for her and her doctor — and Nura's job is to make sure that conversation happens sooner than it otherwise would.

5
Data from older adults as the priority

Sensors behave differently on older skin and older vessels. Building and validating on data from the people who actually wear Nura is our core R&D commitment, and it's the work we intend to publish.

The science is here.
Nura is here.

Nura is in pre-order. Be among the first families to see what patient, honest, passive wellness awareness actually feels like.

Reserve your spot Back to Nura
References

Selected peer-reviewed citations.

A selection of the peer-reviewed research that informs Nura's design. Full bibliography available on request.

[1]Nature Scientific Reports (2021). Accuracy of heart rate variability estimated with reflective wrist-PPG in elderly vascular patients. nature.com ↗
[2]PMC (2025). PPG-based HRV analysis and machine learning for real-time stress quantification. pmc.ncbi.nlm.nih.gov ↗
[3]PLOS ONE (2022). Wearable hydration field study: real-world physiological change measured across 240 participants. plosone.org ↗
[4]JMIR (2024). An effective deep learning framework for fall detection: model development and study design. jmir.org ↗
[5]JAMDA (2024). Emerging digital technologies used for fall detection in older adults in aged care: a scoping review of 73 studies. jamda.com ↗
[6]npj Digital Medicine (2024). Evaluating reliability in wearable devices for sleep staging — 35-article scoping review. nature.com ↗
[7]ScienceDirect (2024). Sleep staging algorithm based on smartwatch sensors: validated on 1,522 nights from 1,430 participants. sciencedirect.com ↗
[8]BMC Geriatrics / PMC (2024). Predicting physical functioning status in older adults: insights from wrist accelerometer sensors using 747 participants. pmc.ncbi.nlm.nih.gov ↗
[9]Frontiers in Digital Health (2022). Wearable sensor systems for fall risk assessment: a review. frontiersin.org ↗