Early-warning health · private by design
Aegis turns the numbers your wearable already tracks into an early warning you can understand. Every reading is analysed and encrypted on your own device. There is no account and no server, so there is nothing to leak.
The problem
A modern watch or ring records resting heart rate, heart rate variability, sleep, movement and skin temperature around the clock. Most apps show you yesterday's numbers in a chart and stop there.
Research on wearable data has shown that changes in these signals, especially a resting heart rate that climbs above your own normal, often appear a day or more before you feel sick. That is an early warning most people never see.
The usual fix is to upload everything to someone's cloud. Health data is some of the most sensitive data there is, so Aegis takes the opposite approach: the analysis comes to your data, not the other way round.
How a check-in works
Enter today's resting heart rate, sleep and stress, plus HRV, temperature and symptoms if you have them. Takes about 20 seconds.
Your reading is sealed with AES-256-GCM using a key made from your passphrase. The key lives only in this tab's memory.
Aegis compares today with your personal 14-day baseline and returns a 0–100 risk score, with the reason behind every point.
You see each factor and how many points it added, like "resting heart rate 10 bpm above your normal: +21". No mystery numbers.
After three check-ins, Aegis judges you against your own baseline instead of a population average, because a fit 58 bpm and a relaxed 75 bpm are both normal for someone.
The site has no backend. Your history is an encrypted file on your device that only your passphrase can open. You can export it, and delete it in one click.
Chest pain, trouble breathing, a very high fever or heart rate skip the trend entirely and tell you to get help now.
Try it with your own numbers, or load 14 days of sample data to watch an infection show up in the trend.
How it works
Everything Aegis does happens inside your browser tab. This page explains exactly what happens to your numbers, how the score is built, and what is still on the roadmap.
Resting heart rate, sleep and stress are required. HRV and temperature are optional but make the read sharper. Symptoms are simple checkboxes. Inputs are validated with clear limits (for example, resting heart rate 30–220 bpm) so typos don't turn into false alarms.
When you create a vault, your passphrase is stretched into a 256-bit key with PBKDF2 (SHA-256, 310,000 rounds, random salt). Each save encrypts your whole history with AES-256-GCM and a fresh random IV, then stores only the ciphertext in your browser. GCM also detects tampering: a changed byte means the file will not open.
Without a vault, a random key is created for the session and everything disappears when you close the tab.
The engine adds up points from each factor below. With three or more earlier check-ins it uses your personal baseline (the average of your last 14 days) for heart rate, HRV and sleep. Some combinations add extra points, because together they mean more than apart.
| Factor | When it adds points | Max |
|---|---|---|
| Resting heart rate | More than 3 bpm above your baseline (3 points per extra bpm), or above 80 / 90 / 100 bpm before you have a baseline | 25 |
| Symptoms | Weighted by symptom, from headache (5) to chest pain (15) | 30 |
| Temperature | 37.5 °C or above; 38 °C or above counts as a fever | 20 |
| Sleep | Under 7 hours (5 points per missing hour), or over 10 hours | 20 |
| HRV | More than 10% below your baseline, or under 20 ms before you have one | 15 |
| Stress | 3 points per level above 1 on the 1–5 scale | 12 |
| Infection-like pattern | Raised heart rate or temperature together with cough, sore throat, fatigue or headache | 8 |
| Strain pattern | Stress 4 or 5 with less than 6 hours of sleep | 6 |
Red flags override the score. Chest pain, shortness of breath, a resting heart rate of 130 bpm or more, or a temperature of 40 °C or above (or below 35 °C) set the score to at least 80 and show an urgent message telling you to seek help.
Resting heart rate is one of the best-studied early signals in wearable research. Large studies have found that a rise above a person's own normal often shows up before symptoms, for example Mishra et al. (2020, Nature Biomedical Engineering) on detecting COVID-19 from smartwatch data before symptoms, and Radin et al. (2020, The Lancet Digital Health) on tracking influenza-like illness with Fitbit data.
That is why Aegis cares more about your change from normal than about any fixed number.
| What | Where it lives | Who can read it |
|---|---|---|
| Today's reading, while you type | This browser tab's memory | Only you |
| Your history | Your browser's storage, as AES-256-GCM ciphertext | Anyone with your passphrase |
| Your passphrase and key | Never stored; the key exists only in memory while unlocked | Only you |
| Exported backup | A JSON file you download, still encrypted | Anyone with your passphrase |
| Anything on a server | Nothing. The site has no backend and blocks network requests from its scripts. | No one |
No. Aegis is a wellness and self-awareness tool. It does not diagnose anything. If you feel unwell or are worried, talk to a doctor. In an emergency in Australia, call 000.
No. There is no Aegis server. The page's security policy blocks its scripts from making network requests, so even a bug could not send your readings anywhere.
Your history cannot be recovered. That is the trade-off for real encryption: no one, including us, holds a copy of your key. You can delete the vault and start a new one.
Not yet. Today your reading is analysed in plain form inside your own browser, then stored encrypted. Computing directly on encrypted data (homomorphic encryption) matters once data leaves your device, which is why it is on the roadmap for population-level signals.
A login means an account on someone's server. A passphrase lets the key be made on your device and never leave it.
About
Aegis Health started as a 72-hour hackathon project by team CipherPulse. Our mission is simple: predictive, privacy-first health signals for anyone with a wearable, not only people with premium healthcare.
Concept, user experience, visual identity and the story behind the product.
The scoring engine that turns heart rate, sleep and symptoms into an explainable risk read.
The encrypt-before-storing architecture and the plan for privacy-preserving population signals.
Healthcare is mostly reactive: you get help after you feel sick. We asked what the data people already carry could do earlier.
Built the first scoring engine and the interactive check-in.
Created the Aegis identity and the Sense, Encrypt, Predict story.
Deployed the site and tested across phones and laptops.
Rebuilt as a working product: real AES-256-GCM encryption, personal baselines, trend charts, encrypted backups, red-flag detection and a tested engine.
Aegis Health is a wellness tool, not a medical device, and does not provide medical advice or diagnosis. Always seek the advice of a qualified health professional about any health concern. In an emergency in Australia, call 000.
Check-in
Enter today's numbers. Aegis encrypts them and returns a risk read with the reasoning behind it, all inside this browser tab.
Run an analysis to see your signal.
Want to see it working first?
Not saved. Create a vault to keep an encrypted history on this device.
| Date | Risk | HR | Sleep | Symptoms | Actions |
|---|