NURA · The app

The Health Engine: organize, track, flag — never diagnose

NURA's Health engine is a deterministic, private, on-device organizer. It records the readings you or your devices report — blood sugar, blood pressure, heart rate, weight, sleep, steps, and mood — tracks how they trend, and classifies each against widely-published general reference ranges so that a dangerous value can be surfaced loudly and routed to a professional. It is an organizer, and the code is written to keep it one.

What it will not do, enforced in code

The engine never diagnoses. Every classification returns a plain description plus, always, the line: "This is general information, not medical advice — please talk to a healthcare professional." There is no code path that names a condition or prescribes anything, and no function that returns "take X." A reading that crosses a dangerous threshold escalates to "This reading can be dangerous. If you feel unwell, seek medical care now" — routing you to a human, not acting as one. Because the whole module is pure — no I/O, no clock, no randomness — the reference ranges and the safety escalation are unit-tested exactly.

The reference ranges

Classification is descriptive only, against general references in the style published by bodies such as the ADA and AHA. Context — fasting versus post-meal, athlete versus non-athlete — changes what a number means, which is precisely why the engine flags and defers rather than diagnosing.

Trends, described — not judged

NURA reports whether a metric is improving, holding, or worsening by comparing the mean of your newer readings against the mean of your older ones. It needs at least four readings before it will say anything, and it ignores changes under five percent as noise. This is deliberately the direction of the number — up, down, or flat — and never a judgement about your health. A glanceable status shows the single most urgent flag across a set of readings, so a critical value is never buried under normal ones.

Built to hand to a doctor

The engine can produce a plain-language summary — the latest value of each metric, how far back it goes, and its trend direction — with no interpretation attached, because the interpreting belongs to the professional. And it always keeps the clock on a reading: 140 before breakfast and 140 after dinner are different facts, and "was my pressure high this morning?" cannot be answered by a number with no time on it, so every reading carries a natural-language when — "today at 8:15 AM," "yesterday at 9:40 PM."

Private by architecture

Health readings are stored on your device, in local storage, behind an explicit consent switch — the engine records nothing until you turn it on, keeps a rolling window of your most recent readings, and lets you delete any single reading or clear them all. There is no NuroNova server for the core app, so nothing here is transmitted to us. The optional AI can read your latest readings through a get_health tool only when you have enabled it with your own provider key — and even then it inherits the same rule the engine lives by: it organizes and flags, it does not diagnose.

The Health engine is built by NuroNova Genix Co. (Delaware; headquartered in Houston, Texas). It runs on-device with no account. The free app is coming soon to iPhone.