Salú Encyclopedia
Salú Encyclopedia · Understand

Your body isn't complicated. It's just been explained badly.

Salú is a health library for people who were never handed the manual. Every topic is built the same way — what it is, how it works, what you can do with it — so learning about your own body feels less like homework and more like curiosity.

Start with sleep Browse the encyclopedia
health in your palm
Who we are

Salú comes from salud. We kept the accent because that's how the word is actually spoken — quick, warm, out loud, between people who know each other.

Most health information arrives in one of two registers: a research paper you can't parse, or a wellness post that flatters you into buying something. Neither one leaves you knowing more about your own body.

We're building the third option — a proper encyclopedia. Plain language, real physiology, no shame, no fear, no miracle mornings. Every entry answers three questions in the same order, so once you've read one page you already know how to read all of them.

Salú isn't a tracker and it isn't a coach. It's the friend who happens to have read the literature, sitting across the table, explaining it clearly and then letting you get on with your day.

yes

Cited physiology, written for a reader with no background

yes

Cultural range in food, bodies, households and routines

yes

Honest uncertainty — we mark how strong the evidence is

yes

Small actions you can actually take today

no

Streaks, badges, or notifications designed to pull you back

no

Before-and-after bodies, or food framed as sin

no

Fear as a motivator, or diagnosis dressed up as education

no

Supplements to sell you at the bottom of the page

The library

Ten shelves. One shared grammar.

Colour isn't decoration here — it's the filing system. Each area of health keeps its hue across the site, the diagrams, the printed page and the carousels, so you always know which shelf you're standing in.

How every entry is built

Know, understand, apply — in that order, every time.

Three stages, marked by ridge-count rather than numbers: one line, three lines, a line that opens. You learn the rhythm on your first page and never have to relearn it.

Know

What is it?

A single, plain-language definition — the sentence you'd want if someone asked you at dinner. Set large, in serif, with nothing else competing for the space.

Understand

How does it work?

The mechanism, in running prose at reading size, with one diagram that carries the idea. Terms are underlined where they're defined, never assumed.

Apply

What do I do with it?

Two to five specific actions on tinted ground, each tagged with when it belongs in a day. Tickable, but never scored — no streak is waiting for you tomorrow.

Knowledge connections

Nothing in the body happens in one chapter.

Sleep is also a hormone story, a light story, a meal-timing story.

Sleep Circadian rhythm Hormones Stress Nutrition Movement Light SIX THREADS OUT OF ONE TOPIC · THE SAME MAP CLOSES EVERY ENTRY
Related-topic cards sit under every section Explore next names why the jump is worth making Pathways string entries into a reading order
Where the science lands

Everything here is aimed at an ordinary Tuesday.

The illustrations are drawn in the same contour lines as the mark, and the people in them vary — different bodies, hair, dress and households. None of them is the default person.

A meal you actually make

Nutrition entries end in the kitchen, not in a supplement aisle.

A walk in the first hour

Light, movement and air, doing three jobs at once before the day starts.

A slower breath

One of the few voluntary levers on a system you don't otherwise control.

A table with people at it

One of the most consistent findings in population health, and the least marketable.

The Salú system

Understand it. Cook it. Live it.

Three products, one visual grammar. You should be able to tell which one you're reading from a single slide, at arm's length, with the logo cropped off.

Understand

Encyclopedia

The science of the body, one question at a time. Colour-coded by system, cross-linked, cited.

Live · 82 entries Nourish

Kitchen

A cookbook that explains itself — ingredients, nutrients, physiology and the recipe on one page.

In development Live

Blueprint

Turning what you've learned into routines that survive a real week — meals, movement, recovery.

Planned
First chapter · Sleep & Recovery

Ten entries, read in order, and sleep stops being mysterious.

Sleep & Recovery is the prototype chapter — the one every other shelf is built against.

The Sleep chapter

Start at why we sleep and finish at a wind-down routine you'll actually keep. About 40 minutes of reading, or one entry a night for ten nights.

Open entry 01
Encyclopedia

A library you can walk into from any door.

Browse by the system it belongs to, the pillar of life it touches, or just type the word that's been bothering you.

All topics 82 entries
If you wear a tracker

Your numbers

HRV, sleep scores, recovery scores. What each one actually measures, what it can't tell you, and where the confident graphic outruns the measurement underneath it.
Filed by system

Supplements

A supplement isn't a body system, so it doesn't get a shelf. Each one is filed under the system it acts on — and gathered here, with every claim on the label rated separately.
Featured learning path

Sleep, from first principles

Ten entries, in order

A path is just a reading order with a reason. This one moves from the question everyone asks to the routine most people are actually looking for.

Open the shelf

—

—

Entry in production

This page is drafted and awaiting citation review — the layout below is the one every entry uses, so you can see exactly where the writing will land.

KnowOne-sentence definition
UnderstandMechanism, in prose
ScienceOne ridge diagram
ApplyTwo to five actions
See a finished entry
Salú Kitchen · Nourish

The cookbook is the encyclopedia, with a pan in it.

Same grid, same type, same ridge diagrams — green takes the lead instead of blue, and the page runs recipe-first. Every ingredient links back to the entry that explains it.

Salú Kitchen · Fibre

Black bean & charred corn bowl

01 Recipe 25 min · serves 2
02 Ingredients 9 items
03 Why these ingredients? links to 4 entries
04 Nutrient spotlight fibre · folate
05 The recipe 6 steps
06 Salú tip one line
07 Explore the science gut–brain axis →
Telling the three apart

At a glance, on a phone, with the logo cropped off: the header bar's colour, the eyebrow's verb, and the shape of the first block do all the work.

Encyclopedia

Eyebrow reads UNDERSTAND. Opens on a question. Serif definition first.

Kitchen

Eyebrow reads NOURISH. Opens on a dish. Numbered flow down the page.

Blueprint

Eyebrow reads LIVE. Opens on a week. Horizontal timeline, not a column.

Salú Blueprint · Live

Knowledge that survives a real week.

Blueprint is the third pillar and the last one we'll build, because it only works once the first two are dense enough to draw from. It takes what you understood in the Encyclopedia and what you cooked in the Kitchen, and lays them across a week you'd actually live — meals, movement, light, recovery, and the space between them.

Visually it turns the column on its side: the same colours and type, arranged as a horizontal week rather than a vertical article.

Brief interaction→Useful information→Real-world action→Leave the phone
Shelf

—

—

—entries in this shelf
—to read the whole thing
—rated strong evidence
Start here

—

—
The shelf, in reading order
What this shelf touches

Nothing in the body stays on one shelf. These are the other areas this one keeps running into.

Editorial standards

How we decide what to tell you.

Health writing gets untrustworthy in a specific way: a single promising study becomes a headline, the headline becomes a rule, and nobody records how thin the original evidence was. Salú puts that record on the page.

Three ridges, and what each one means

Strong & consistent

Broad scientific consensus, supported by several independent lines of research over time. Something would have to go badly wrong in the field for this to reverse.

e.g. sleep supports memory consolidation

Well supported, open questions

The direction is clear and repeatedly observed, but the size of the effect, the mechanism, or who it applies to is still being worked out. We say which part is unsettled.

e.g. the gut–brain axis

Early or contested

Promising, preliminary, or actively disputed among researchers. We include it when the question is one people are already asking — and we say plainly that the answer isn't in yet.

e.g. the physiology of burnout

The rules we hold ourselves to

Every entry is dated and reviewed on a cycle

The review date sits at the top of the page, not in a footer. If an entry hasn't been looked at in over a year, that's visible to you before you read it.

We prefer reviews and consensus statements to single studies

One trial is a data point, not a finding. Where we cite a single study we say so, and we say what it did and didn't measure.

Diagrams are labelled schematic when they're schematic

An illustration of a mechanism is not a plot of measured data. Ours say which they are, on the figure, every time.

No numbers without a source, no source without a link

If we can't attribute a figure to something you could go and read, the figure comes out of the entry.

Nothing here is diagnosis, and no conclusion is for sale

Salú does not sell supplements or treatments, and takes no money from anyone who does. If Salú ever charges for something, it will be for a book, a tool or a licence — never for a finding. There is no commercial reason for us to overstate anything, which is the point.

What you read here is not a product

Salú does not sell, share or broker reader data — not to advertisers, not to wearable or supplement companies, not to anyone, ever. We count which entries get opened and which searches come back empty, because that is how we decide what to write next. We do not build profiles of you, we do not follow you around the internet, and what you look up here does not leave here.

Corrections are published, not quietly patched

When we get something wrong we say what changed and when. A library that edits its own history isn't one.

The short version

Salú is health education, not medical advice. It is not a diagnosis and it is not a substitute for a clinician who knows your history.

Get in touch

Salú is early. That's the interesting part.

The Encyclopedia is being written now, the Kitchen is in development, and the Blueprint is on the far side of both. If you want to read it, write for it, review it, or build it with us — this is the moment where that changes the shape of the thing.

Write to us

One inbox, read by a person. Tell us which of the reasons below you're writing about and it gets to the right place.

hello@salulibrary.org
ReadersTell us what confused you, or what you wish had an entry. Reader questions set the writing queue.
CliniciansReview an entry in your area. Reviewers are credited on the page unless they'd rather not be.
WritersIf you can explain physiology without either dumbing it down or showing off, we want to hear from you.
PartnersSchools, clinics and community health programmes that want the Encyclopedia in front of the people they serve.
FundersSalú is independent and takes no money from anyone selling supplements or treatments. We'll show you the plan and the numbers.

Where it goes from here

NOWSleep & Recovery, written in fullTen entries, cited and reviewed — the chapter every other shelf is built against.IN PROGRESS
NEXTNutrition and FoundationsThe two shelves everything else links back to, so cross-references stop dead-ending.QUEUED
THENSalú Kitchen, first twenty recipesThe cookbook that explains itself, wired back into the entries that explain the ingredients.IN DESIGN
LATERSalú Blueprint, and the companion appOnly once there's enough to draw from. Brief interaction, useful information, real-world action, leave the phone.PLANNED

Why do we sleep?

You spend about a third of your life unconscious. Evolution doesn't keep an expensive habit like that unless it's paying for something.

Evidence Strong & consistent
Reading time 6 min
Last reviewed Aug 2026
Chapter 01 of 10 · Sleep & Recovery

Know

Sleep is a reversible state your brain actively puts itself into — not the absence of being awake, but a different mode of running, with its own chemistry, its own electrical signature, and its own work to get through.

also filed under · rest · recovery · circadian biology

Understand

Two systems decide when you sleep, and they run independently. The first is sleep pressure: from the moment you wake, a by-product of brain activity called adenosine builds up, and the more of it there is, the heavier you feel. Sleep clears it. Caffeine doesn't — it just sits in adenosine's parking space for a few hours.

The second is your circadian rhythm, an internal clock running on roughly 24 hours and set mostly by light hitting your eyes. It doesn't measure how tired you are. It issues an alerting signal that rises and falls on schedule regardless, which is why you can be exhausted at 3pm and wide awake at midnight.

You fall asleep easily when both line up: high pressure, low alerting signal. Most sleep trouble is a disagreement between the two.

Once you're under, the night isn't uniform. It cycles roughly every 90 minutes through lighter stages, deep sleep and REM, with deep sleep front-loaded into the first half and REM stretching longer towards morning. That shape is why cutting the last two hours of a night doesn't just cost you sleep — it costs you disproportionately more REM.

The science

How to read this
  1. Two separate things decide when you sleep. They are not the same thing, and they do not run on the same schedule.
  2. Sleep pressure builds up every hour you are awake, like a debt. Sleeping is what pays it off.
  3. The alerting signal is your body clock telling you to stay awake. It goes by the time of day and ignores how long you have been up.
  4. Whichever line is on top is winning — and the further apart they are, the more decisively. The two bars mark the day's clearest example of each.
ASLEEP · PRESSURE DISCHARGES STRONGER 07:0012:0017:00 22:0003:0007:00 TIME OF DAY · SCHEMATIC, NOT MEASURED DATA Sleep pressure Alerting signal A B
Swipe the chart sideways to see the full 24 hours →
Fig. 01A — midday. The clock's alerting signal is on top, so you feel awake even though pressure has been building since breakfast. B — around 10pm. Pressure has reached its peak for the day and the alerting signal has dropped away underneath it. That vertical distance is the widest it gets, which is why sleep arrives with the least effort here. Known as the two-process model. Curves are illustrative, not measured values.

Why it matters

Sleep isn't downtime you're paying for with productivity — it's when a large amount of the body's maintenance is scheduled. Memory from the day gets consolidated. Immune signalling, appetite hormones and glucose handling all shift measurably after even a few short nights. And the effects are cumulative: you don't feel the deficit arrive, you feel the version of yourself that's been carrying it for a fortnight.

7–9hNightly range recommended for most adults by sleep-medicine consensus
~90 minLength of one full sleep cycle, repeated four to six times a night
1st halfWhere most deep sleep is concentrated — REM builds towards morning

Apply

Ticks are for your own place-keeping. Nothing is counted, nothing is stored, and there's no streak waiting for you tomorrow.

Myth / Reality

Myth

Some people simply don't need much sleep — they've trained themselves out of it.

Reality

Genuinely short sleepers are rare and it appears to be genetic, not trainable. What most people build is a tolerance for feeling tired — performance measures keep declining after the feeling of impairment plateaus.

Salú Snapshot

Salú Snapshot · Why do we sleep?
In one line

Sleep is active maintenance, scheduled by two systems that don't talk to each other.

The mechanism

Pressure builds while you're awake. The clock runs on light. Sleep arrives where they align.

The move

Fix the wake time and the morning light first — the evening tends to follow.

Every entry ends in a snapshot. It's also the card that becomes slide seven of the carousel and the boxed summary in print.

Bite-size tip

If you only change one thing this week, get outside within an hour of waking. It's the cheapest lever on the whole system.

Explore next

How we handle evidence

Every entry carries a three-ridge evidence mark. Three means broad scientific consensus across independent lines of research; two means well supported with open questions; one means early or contested, and we say so in the text. Entries are reviewed on a rolling basis and dated. Salú is health education, not medical advice — nothing here is a diagnosis, and it isn't a substitute for talking to a clinician who knows your history. Read the full editorial standards →

Supplement entry

Ashwagandha

The best-selling adaptogen in the world, carrying more claims than it can hold. Three of them stand up. One of the popular ones runs backwards.

Evidence Rated per claim, below
Reading time 7 min
Last reviewed Sep 2026
Shelf Stress & Mindfulness

What it is

Ashwagandha is the root of Withania somnifera, a shrub used in Ayurvedic medicine for a very long time and sold in Western capsules for about fifteen years.

also filed under · adaptogen · Indian ginseng · winter cherry

Before the claims

Nearly every trial worth reading used a concentrated root extract, standardised to its withanolide content — not raw powder, not the leaf, and not whatever is in a gummy. If a bottle doesn't say which extract and how much, the trials on this page don't describe what's in it.

The species name, somnifera, means sleep-inducing. That is a better clue to what this plant does than most of the words printed on the front of the box.

The claims, one at a time

One rating for a whole supplement would be a lie in both directions. Each claim gets its own.

It lowers cortisol Strong & consistent Yes — this is the finding that replicates. A 2024 systematic review pooling 15 randomised trials in 873 adults found a significant reduction in blood cortisol after eight weeks of supplementation compared with placebo. Of everything ashwagandha is sold for, this is the claim on the firmest ground.
It reduces stress and anxiety Supported, open questions Yes, on questionnaires. The same review found meaningful drops in standard anxiety and perceived-stress scores. An international taskforce went as far as a provisional recommendation for generalised anxiety disorder in 2022. The open questions are real, though: the trials are small, they run six to twelve weeks, they come overwhelmingly from one country, and a good share are funded by extract manufacturers. Note also that the same review found no improvement in quality of life — symptom scores moved, the rest of life didn't visibly follow.
It helps you sleep Small but real A little. A 2021 meta-analysis found a small but statistically significant improvement in sleep versus placebo, concentrated in people taking 600 mg a day for eight weeks or more, and in people who actually had insomnia rather than healthy sleepers. Small is doing real work in that sentence — this is a nudge, not a sleeping pill.
It spikes cortisol, so take it at night Backwards This one is the opposite of what the trials show. It is repeated constantly in supplement marketing and on social media, and we could not find a controlled trial behind it. The evidence points the other way: ashwagandha lowers cortisol.

Evening dosing is still reasonable — just for a different reason. The plant is mildly sedating, and the sleep trials that found a benefit dosed in the evening. So the advice survives; the explanation attached to it does not. Worth knowing which part you were relying on.
It raises testosterone and builds muscle Early or contested Thin. A handful of small trials, mostly in men doing resistance training, several of them run or funded by the companies selling the extract, with inconsistent results. Not nothing — but nowhere near the footing of the cortisol finding, and it is the claim doing most of the work in gym marketing.
It's safe to take indefinitely No data either way Unknown, and we won't pretend otherwise. Trials show it is generally well tolerated for up to about three months. Past that there is no good evidence in either direction, which is not the same as a clean bill of health. See what can go wrong — the liver signal matters here.

Who it's actually for

An adult under sustained, ordinary life stress who wants a modest and measurable reduction in it, and who is willing to give it eight weeks. That is the person the trials were run on, and roughly the effect they got.

It is not a treatment for an anxiety disorder, it is not a substitute for sleeping enough, and it will not out-run a situation that is genuinely too much. A supplement that lowers a stress hormone does not remove the thing producing the stress.

Dose and form

What the trials used
Dose

240–600 mg a day of root extract. Benefits showed up more consistently at the 500–600 mg end.

Form

Standardised root extract. The label should name the extract and the withanolide percentage. If it doesn't, you can't map it to any of this.

Timing

Evening, if sleep is what you're after — that's what the sleep trials did. The cortisol trials mostly dosed twice daily.

Give it eight weeks before deciding. Nearly every positive result in this literature took that long, and the ones measured at four weeks were weaker.

What can go wrong

Serious

Liver injury

The US National Institutes of Health's drug-injury database rates ashwagandha a likely cause of clinically apparent liver injury, with around 23 published cases. Onset is typically two to twelve weeks in, and usually shows up as jaundice and itching. Most cases settle within one to four months of stopping — but acute liver failure, transplant and death have all been recorded, particularly in people who already had liver disease. If your skin or eyes yellow, or you start itching without a rash, stop and get liver bloods.

Watch

Thyroid

It can raise thyroid hormone levels, and there is at least one published case of thyrotoxicosis. If you have a thyroid condition or take thyroid medication, this is a conversation with your doctor, not a self-experiment.

Avoid

Pregnancy, and hormone-sensitive prostate cancer

Experts advise against it in pregnancy over abortifacient potential, and in men with hormone-sensitive prostate cancer because of its effect on testosterone.

Note

You often can't be sure what's in the bottle

Investigations into the liver cases were repeatedly complicated by contaminated or mislabelled product. In a market with no pre-market approval, the gap between the label and the capsule is its own risk — separate from the plant.

Common

Mild and reversible

Nausea and drowsiness are the usual complaints, and they stop when you do.

The bottom line

In one paragraph

One of the few supplements with a replicated effect on something you can measure. It lowers cortisol, it takes the edge off stress scores, and it helps sleep a little. It does not spike cortisol. It has a real liver signal that most of the people selling it don't mention, no long-term safety data, and a claim on testosterone that is thinner than the marketing suggests. Eight weeks, a named root extract, and stop if you go yellow.

Sources

  1. Effects of Ashwagandha Supplements on Cortisol, Stress, and Anxiety Levels in Adults: A Systematic Review and Meta-Analysis. BJPsych Open, 2024. 15 randomised trials, 873 participants.
  2. Effect of Ashwagandha (Withania somnifera) extract on sleep: A systematic review and meta-analysis. PLOS ONE, 2021.
  3. Ashwagandha: Is it helpful for stress, anxiety, or sleep? NIH Office of Dietary Supplements, Health Professional Fact Sheet.
  4. Ashwagandha. LiverTox: Clinical and Research Information on Drug-Induced Liver Injury. National Institute of Diabetes and Digestive and Kidney Diseases.

Salú is health education, not medical advice. Nothing here is a diagnosis, and no conclusion on this page is for sale.

Explore next

Cross-cutting index

Supplements, rated one claim at a time.

A whole diet should cover most of what a body needs. It frequently doesn't — and people take these whether or not the evidence supports it, so the useful thing is to say plainly which parts hold up.

A supplement isn't a body system, so it doesn't get a shelf of its own. Each one is filed under the system it acts on and collected here, and each entry rates every claim on the label separately — because one rating for a whole compound is always wrong in at least one direction.

Written and reviewed

Queued

Drafted against the same format, awaiting citation review.

Where we stand

Salú does not sell supplements and takes no money from anyone who does. That is the only reason an entry here can end with "don't bother" — and several of them will.

What your device is telling you

Heart rate variability

Your watch hands you a number every morning with no story attached, and the implication is that something is wrong with you. Usually nothing is.

EvidenceRated per claim, below
Reading time7 min
Last reviewedSep 2026
ShelfBody Systems

What it is

Your heart does not tick like a metronome. The gap between one beat and the next is changing constantly, and heart rate variability is simply a measure of how much it changes.

also filed under · HRV · RMSSD · beat-to-beat variability

The mechanism

Two nerves argue over your heart's pace. One is the accelerator — the sympathetic branch, the one that wakes up when something needs handling. The other is the brake: the vagus nerve, which slows the heart between demands. Both are connected at once, pulling in opposite directions, and the tug-of-war is what produces the variation.

The brake is faster than the accelerator. It can adjust between individual beats; the accelerator works on a slower chemical timescale. So most of the beat-to-beat wobble your device is measuring comes from the brake being applied and released.

That's why a high number isn't a strong heart and a low number isn't a weak one. Plenty of variation means both systems are online and responsive. Very little variation means one of them has taken over — usually the accelerator, holding the pace steady because something is being dealt with. A low reading doesn't say your heart is failing. It says your body is braced.

Which might be a hard session yesterday, a late dinner, two drinks, a cold coming, a deadline, or a bad week. The number tells you the brace is on. It does not tell you what you are bracing against.

What it can and can't tell you

HRV is sold as a single verdict on your body. It supports some claims well and others not at all, so we rate them separately.

It reflects your autonomic nervous system Strong & consistent Yes — this is the solid ground. Beat-to-beat variability is a well-established, decades-old index of vagal activity. The physiology is not in dispute. Everything else on this page is an inference built on top of it.
Your own trend tells you something useful Supported, open questions Within yourself, over time, yes. A run of mornings well below your own baseline is real information about load — training, illness coming on, alcohol, sleep debt, sustained stress. What's still unsettled is how much of a drop matters, and for whom. Nobody can give you a threshold that means something.
You can compare your HRV to other people's Backwards This is the one that does real harm, and it's the one every leaderboard is built on. Among healthy adults the spread is enormous — published normative work puts the common measure around 42 ms on average with a range running roughly 19 to 75 ms. Two people can both be entirely healthy and sit at opposite ends of that. Age pushes it down on its own. Comparing your number to your friend's is close to meaningless, and any product ranking you against strangers is selling you a feeling.
A single low morning means take the day off Early or contested Thin, and the honest answer is that one reading is noise. HRV moves with posture, breathing rate, time of measurement, hydration, and what you ate last night. Guided-training studies that adjust sessions from HRV trends are genuinely interesting, but they use rolling averages over days, not a single morning's figure. One low reading is information about last night.
You can raise your HRV with a device or supplement No data either way Largely unevidenced, and the framing is wrong anyway. HRV is a readout, not a dial. Things that genuinely reduce load — sleeping enough, drinking less, training sensibly — tend to show up in it. Products sold to move the number directly are usually selling you the readout rather than the thing it reads.

How good is the number itself?

Better than you might fear, with one large caveat. Wrist and finger devices don't read electrical activity the way a clinical ECG does — they read blood volume changes through the skin with light, and infer the beat intervals from that. Under still conditions, overnight, that inference is reasonable.

The caveat is that the number is only comparable to itself under matched conditions. Values taken over 24 hours, over five minutes, and over one minute are not interchangeable, and neither are values taken standing versus lying down, or at different points in the night. This is why your device insists on measuring during sleep and why the figure it shows you will not match a reading from a different app or a different device. Neither one is lying. They are measuring different windows.

What to actually do with it

Read it as a trend against your own baseline, never as a scoreSeven mornings is information. One morning is weather.weekly
When it drops, look for the cause before changing the trainingAlcohol, a short night, a late heavy meal and an incoming cold all do this. So does a genuinely hard week.when it falls
Stop comparing it to anyone else'sIncluding your partner's, your friend's, and the average for your age bracket.always
Ignore the number entirely on days you feel fineIf your body and your device disagree and you feel well, the device is one input, not a verdict.often

The bottom line

In one paragraph

HRV is a real measure of a real thing: how much your nervous system's brake is working. Your own trend, over days, is genuinely useful. Your number against someone else's is nearly meaningless, because healthy people span a threefold range. Treat a low morning as a question, not an instruction — and if you feel well, you probably are.

Sources

  1. An Overview of Heart Rate Variability Metrics and Norms. Frontiers in Public Health, 2017. Source of the normative range and of the warning that recording windows are not interchangeable.
  2. Heart rate variability as a way to understand associations between the autonomic nervous system and affective states: a critical review. Biological Psychology.
  3. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology — standards of measurement, physiological interpretation and clinical use.

Salú is health education, not medical advice. Nothing here is a diagnosis, and no conclusion on this page is for sale.

Explore next

What your device is telling you

Your sleep score

A confident pie chart of light, deep and REM, built from a wrist that cannot see your brain. Some of it is solid. One part of it is much shakier than the graphic suggests.

EvidenceRated per claim, below
Reading time8 min
Last reviewedSep 2026
ShelfSleep & Recovery

What it is

A sleep score is a device's best guess at how your night went, assembled from movement, heart rate, and the timing of both — then compressed into one number and a coloured chart of stages.

also filed under · sleep stages · sleep staging · light / deep / REM

How a wrist guesses at a brain

In a sleep laboratory, stages are scored from electrical activity: electrodes on the scalp read brain waves, others read eye movements and muscle tone. Deep sleep is identified by slow waves in the brain. REM is identified by the eyes moving while the body stays still. These are direct observations of the thing itself.

Your wearable has none of that. It has an accelerometer that knows how much you moved and an optical sensor that knows how your pulse is behaving. From those two streams it infers the rest — and it can infer quite a lot, because the stages really do have heart-rate and movement signatures.

That's why the overall shape tends to be roughly right while the individual slices are the weakest part. Knowing you were asleep is easy: you stopped moving and your heart slowed. Knowing whether a given thirty seconds was light sleep or REM is a much harder call from the outside, and it's exactly the call the pie chart presents most confidently.

What the score can and can't tell you

Different parts of that screen deserve very different amounts of your trust.

It knows roughly when you fell asleep and woke up Strong & consistent Yes, and this is the most useful thing it does. Across validation studies, devices identify sleep itself with high sensitivity — reliably in the nineties. If you want to know whether your bedtime is drifting later across a month, the device will tell you honestly.
It can track changes in your sleep over weeks Supported, open questions This is what these devices are genuinely for. Validation work concludes they can track prolonged, meaningful changes in sleep patterns even while disagreeing with the laboratory on any given night. The trend survives the noise; a single night does not.
It accurately measured your deep sleep last night Shakier than it looks This is the number people panic about, and it's the one to trust least. In a 2025 study of six consumer wrist devices tested against laboratory sleep studies in 62 adults, agreement on four-stage classification ranged from fair to moderate — the best device reached a kappa of 0.53, the weakest 0.21. Kappa is a measure of agreement beyond chance, and 1.0 would be perfect. None of them is close. The devices are not broken; they are inferring something they cannot see. But "45 minutes of deep sleep" is presented with a precision the underlying measurement does not have.
It knows how long you lay awake Systematically optimistic This is the clearest, most consistent failure, and almost nobody mentions it. These devices are excellent at spotting sleep and poor at spotting wake — in that same study, specificity for detecting wakefulness ran from about 29% to 52%, against sensitivity for sleep in the low-to-mid nineties. Every device underestimated time spent awake after falling asleep, by twelve to forty-eight minutes.

If you lie still with your eyes closed and your mind racing, your wrist has very little to distinguish that from sleep — so it calls it sleep. Which means the people most likely to be told they slept well are the people who lay awake without moving.
A low score means you should be worried No data either way No score has been validated against health outcomes, because the formulas are proprietary and differ between companies. There is also a documented pattern of people becoming anxious about their sleep data in a way that worsens their sleep. If the number and how you feel disagree repeatedly, trust how you feel — and take persistent unrefreshing sleep to a doctor rather than to an app.

Why we can say this and your device can't

None of the above is an accusation. Inferring brain states from a wrist is a genuinely hard engineering problem, and the devices have got dramatically better at it. The companies building them employ serious sleep scientists.

But no company can be the neutral authority on its own instrument's accuracy. That isn't a comment on anyone's honesty — it is simply what a conflict of interest is. Someone has to be able to say the pie chart is more confident than the measurement, and it cannot be the company that drew the pie chart.

Salú sells no devices and takes no money from anyone who does. That is the only reason this page can exist.

What to actually do with it

Watch bedtime and wake time, not the stage breakdownThose are the two numbers the device actually measures well, and they're the two you can change.daily
Compare months, not nightsThe trend is the signal. One night's minutes are mostly the algorithm's opinion.monthly
If you want more deep sleep, drop the room temperature and move the last drink earlierAlcohol and a warm bedroom are the two reliable suppressors. Both act on the mechanism directly.tonight
If checking the score is making you anxious, stop checking itSleep anxiety is a documented way to make sleep worse. The device is optional; the sleep isn't.if needed

The bottom line

In one paragraph

Your device is good at knowing when you slept, weaker at telling the stages apart, and systematically optimistic about how long you lay awake. Use it for the shape of your month. Don't litigate a single night's minutes, and don't let a number tell you that you feel worse than you do.

Sources

  1. A performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography. SLEEP Advances, 2025. 62 adults; source of the kappa range, the sleep-sensitivity and wake-specificity figures, and the underestimation of time awake.
  2. It Is All in the Wrist: wearable sleep staging in a clinical population versus reference polysomnography. Nature and Science of Sleep.
  3. American Academy of Sleep Medicine position statements on consumer sleep technology.

Salú is health education, not medical advice. Nothing here is a diagnosis, and no conclusion on this page is for sale.

Explore next

What your device is telling you

Recovery and readiness scores

One number out of a hundred, telling you what kind of day your body is ready for. It is a real measurement wrapped inside an opinion — and the opinion belongs to a company.

EvidenceRated per claim, below
Reading time6 min
Last reviewedSep 2026
ShelfMovement

What it is

A recovery or readiness score is several separate measurements — usually overnight heart rate variability, resting heart rate, sleep, and sometimes breathing rate and skin temperature — combined by a private formula into a single number.

also filed under · readiness · body battery · daily score

What's underneath the number

The inputs are real physiology and most of them are individually defensible. Resting heart rate rising overnight, HRV falling below your own baseline, breathing rate creeping up, a short or broken night — each of those genuinely tends to accompany a body under load, whether from training, illness, alcohol or stress.

What you cannot see is the recipe. How much weight each input carries, how the baseline is calculated, how many days it looks back — all of that is proprietary, and it differs between companies. Two devices on the same body on the same morning will frequently give you different scores.

That's why the score can be both useful and unverifiable at the same time. The ingredients are measurable. The combination is an editorial decision made by a company, expressed in the language of measurement. A number out of a hundred looks like a reading. It's closer to a recommendation.

What it can and can't tell you

Its inputs track real physiological load Supported, open questions The ingredients are sound. Overnight HRV and resting heart rate are well-established autonomic markers, and both move in the expected direction under training load, illness and alcohol. What's open is how much any particular deviation should change what you do.
A run of low scores is worth paying attention to Supported, open questions Several days in a row is a genuine signal. It usually means something real: accumulated training, an infection starting, sustained poor sleep, or a hard stretch of life. It won't tell you which — that's your job, and you usually already know.
The number itself is a validated measurement Category error It cannot be independently validated, because the formula isn't published. An outside researcher can test whether HRV predicts something. Nobody outside the company can test whether this particular weighting of HRV, resting heart rate and sleep predicts anything, because they can't see it. A number presented to two decimal places implies a precision that an undisclosed formula cannot claim.
Scores from different devices mean the same thing Backwards They're not on a common scale and were never designed to be. A 70 on one device is not a 70 on another — different inputs, different weights, different baselines. Comparing scores across brands, or comparing yours to a friend's, is comparing two companies' opinions, not two bodies.
Training by the score improves outcomes No good data either way Genuinely unresolved. HRV-guided training — adjusting sessions from a rolling trend rather than a single proprietary score — has some promising trial work behind it, mostly in trained athletes. Whether following a commercial daily score beats simply training sensibly and listening to your body has not been established for ordinary people.

What to actually do with it

Treat it as a prompt to check in, not an instruction"Low today" is a good reason to ask yourself how you actually feel. It is not a reason to skip something you feel fine for.mornings
Look at the inputs, not the compositeMost apps let you see HRV and resting heart rate separately. Those are the measurements. The score is the interpretation.when it drops
Take three or more low days seriouslyThat is usually real. One is usually last night.weekly
Never let a good score talk you into training through illness or injuryThe formula does not know your knee hurts.always

The bottom line

In one paragraph

Real measurements, combined by a recipe you can't see, presented as a verdict. The underlying signals are worth watching and a run of low days usually means something. The number itself is a company's opinion about how to weigh them — useful as a prompt, not as an authority, and not comparable to anyone else's.

Sources

  1. An Overview of Heart Rate Variability Metrics and Norms. Frontiers in Public Health, 2017.
  2. Reviews of HRV-guided training in athletes — the closest thing to a tested version of this idea.
  3. Validation studies of consumer wearables against polysomnography, for the sleep component of the composite.

Salú is health education, not medical advice. Nothing here is a diagnosis, and no conclusion on this page is for sale.

Explore next

Cross-cutting index

Your numbers, explained by nobody who sells them.

Millions of people now wake up to a score. Almost nobody is told what the number is a number of — which part of it is measured, which part is inferred, and which part is a company's opinion wearing the clothes of a measurement.

These entries explain the metrics on your wrist the way we'd explain any other physiology: mechanism first, every claim rated separately, and the limits stated plainly. Salú sells no devices and takes no money from anyone who does, which is the only reason a page like this can say what it says.

Written and reviewed

Queued

Drafted against the same format, awaiting citation review.

About Salú

Health in your palm.

A knowledgeable friend who understands the science and can explain it clearly.

That sentence is the whole brief. Not a journal, not an influencer, not a clinic — the person at the table who read the research and can tell you what it said without making you feel even more confused or scared.

Most health information comes in one of two forms: a study written for other scientists, or a post that flatters you into buying something. Salú is the third option — plain language, real physiology, no shame, no fear, and nobody paying us to tell you otherwise.

And it's built to be left. Read the entry, take the one thing, close the tab.

Why this exists

I built this because I needed it, and it wasn't there.

At sixteen I was told I had supraventricular tachycardia — AVNRT. I understood two things: that it had to do with my heart, and that whatever else I needed to know, I would be looking up myself.

One episode ran about ninety minutes and took adenosine to stop — a drug that briefly blocks the loop the signal is stuck in, so the rhythm can start over. Two years later, weeks after my eighteenth birthday, I had surgery. It was 2020 — the middle of the pandemic, when a hospital was the last place anyone wanted to be, and going in mostly meant going in alone.

A few years after that, bloodwork at the end of my first year of university came back prediabetic. Same experience, different organ: a label, a leaflet, and no explanation of the process I was apparently already in the middle of. I changed how I ate and how I lived. A year later the number had moved. Nothing about that year was complicated — it had just never been explained to me.

I'm also the one in my family who reads the paperwork. First-generation kids do a particular kind of work: you translate the diagnosis at the same moment you're absorbing it, and you learn early that the quality of what you can find decides what your whole family ends up understanding.

None of that made me an expert. It made me a very specific kind of reader — the one standing in a parking lot with a phone, a word they've never heard, and a family waiting on an answer. Everything about Salú is built for that person, because I was that person more than once.

“Health in your palm” was never a line about phones. It's about who gets to hold it.

What that turned into

Nobody gave me one sentence I could hold onto. →KNOW is a single plain definition, set large, with nothing competing.
I had the label but not the mechanism. →UNDERSTAND explains how it works, not just what it's called.
I needed to know what to do on Monday. →APPLY is two to five specific actions, tagged to a time of day.
I was translating for my family in real time. →Every entry exists in English and Spanish, switchable in one tap.
The frightening version was always easier to find than the useful one. →No fear as a motivator, and no diagnosis dressed up as education.
Almost everything I found was selling something. →No supplements, no sponsors, and nobody paying for a conclusion.
I couldn't tell strong evidence from a confident headline. →Three ridges, printed on the entry, saying how much to trust it.
Where this is meant to go

The people who need this most are the furthest from it.

Rural communities, remote areas, and neighbourhoods where the nearest clinic is an appointment you have to plan a day around — that's where the gap between having a diagnosis and understanding it is widest, and where good health information arrives last if it arrives at all. Salú starts as something one person can open on their own phone, for free, with no account. That's the stepping stone, not the destination.

From there it belongs somewhere with reach: community health centres handing an entry to someone at the moment of diagnosis, school districts using it where health class runs out of hours, clinics printing a snapshot instead of a leaflet nobody reads. Built to be given away, in both languages, by whoever is already standing closest to the person who needs it.

Individual resource Health centres & clinics School districts Community programmes
Design principles

What we commit to, and what we refuse.

Six operating rules. Each one is a commitment and a refusal, because a principle that never costs you anything isn't one.

01

Clarity over cleverness

One idea per page, one accent colour, and a measure set for comprehension rather than density. Every element earns its place by carrying meaning.

We refuseDecoration that draws attention to itself instead of the content.

02

One structure, every time

Know, understand, apply — in that order, on every entry. A reader who learns the rhythm once can move through the whole library without re-orienting.

We refuseFormat experiments that make each page a puzzle to solve again.

03

Evidence, stated plainly

A visible strength rating on every entry, a review date at the top rather than the footer, and a traceable source behind every figure.

We refuseConfidence we haven't earned. We would rather publish “we don't know yet” than round it up.

04

Education, never pressure

Information that leaves someone more capable of making their own decision, and small actions they can take the same day.

We refuseFear as a motivator, shame attached to food or bodies, streaks and badges built to pull people back, and diagnosis dressed up as education.

05

Built for more than one kind of life

Every entry in English and Spanish. Illustrated people who vary in size, age, tone and dress. Ingredients from more than one continent.

We refuseA default reader whose household is assumed to look like everyone's.

06

Independent, and designed to be left

The Encyclopedia stays free to read, with no account required, and no revenue from anyone selling supplements or treatments — which is what lets the evidence ratings mean something.

We refuseSponsored entries, affiliate links, and time-on-page as a measure of success.

Brief interaction→Useful information→Real-world action→Leave the phone
The illustration language

One primitive, drawn over and over.

A fingerprint is a field of open contour lines. So is a hillside, a hormone curve, an avocado and a person asleep. Every drawing in Salú is made the same way — one stroke weight, no fills, every body a closed contour with a single interior echo — which is why a diagram, a food and a figure can sit on the same page without arguing.

Lifestyle — people, drawn in the plural
Sleeping
Cooking
Walking
Exercising
Meditation
Eating together
Company
Outdoors
Food — friendly, never moralised
Avocado
Grain
Greens
Fish
Legume
Berries
Anatomy — simplified, never cartooned
Heart
Lungs
Gut
Neuron

Anatomy is drawn accurate enough to be trusted and simple enough to be read at thumbnail size. Where a drawing has to leave something out to stay legible, the caption says what.

How Salú travels

A carousel is one entry, folded into eight.

Same structure, same colour, same ridge marks — a slide should read as a page torn out of the encyclopedia, not as a social account with its own personality. Scroll it.

Kitchen carousels swap the accent to green and the eyebrow to NOURISH; Blueprint runs neutral with LIVE. Nothing else changes.