Helix - Longitudinal Health Analytics

The Problem

Health data exists. Health understanding doesn't.

Consumer health apps show you today's step count. They don't help you understand why you feel worse than you did two years ago. The data exists - HealthKit has been logging your heart rate variability, resting heart rate, sleep quality, and activity for years. Your blood panel results sit in PDFs from your lab. Your Apple Watch knows more about your physiology than you do. But none of these data sources talk to each other, and none of them are analyzed over time in a way that means anything.

Health data exists in silos. HealthKit holds one category. Paper lab reports hold another. Your Oura ring holds a third. Your Apple Watch holds a fourth. No single tool synthesizes them. When you see your physician for an annual physical, they review a snapshot - your labs from last week, your vitals from today, perhaps a brief summary of how you have been feeling. The longitudinal signal that might have been visible for months is invisible in that snapshot.

This is not a data availability problem. The data is there. It is a synthesis and analysis problem - and it is exactly the kind of problem that on-device machine learning is now capable of solving.

What Helix Does

A longitudinal model built from everything you already have.

Helix does not ask you to do anything new. It reads the data that already exists on your device and in your HealthKit library, adds the lab results you photograph or import, and builds a unified biomarker model over time. The model improves as data accumulates. After six months, it is meaningfully better than at three. After a year, it surfaces patterns that no individual data point could have revealed.

Biomarker Modeling

Builds a longitudinal model from all your health data inputs - HealthKit history, Apple Watch continuous measurements, imported lab results, and third-party wearable integrations. The model tracks biomarkers over time and establishes personal baselines, distinguishing your normal variation from meaningful deviation.

Trend Detection

Identifies patterns and deviations months before they become symptomatic. Helix does not tell you what a trend means medically - it tells you what it noticed, with a confidence score, and suggests it may be worth discussing with your physician. It surfaces the signal; it does not interpret the clinical significance.

Lab Integration

Reads lab PDFs and maps results to standard reference ranges with longitudinal context. Rather than showing you a single lab value and a reference range, Helix shows you where that value sits relative to your own historical range - which is often more informative than the population reference range.

Wearable Synthesis

Unifies Apple Watch, HealthKit, and third-party wearable data into a single timeline. Data from different sources is normalized, deduplicated, and integrated into the biomarker model. You don't need to manage which app reads which sensor - Helix handles that.

Privacy

All of this happens on your phone. None of it leaves.

All processing happens on-device. No health data is transmitted to Halzer Group or to any third party. No accounts are required. Your data never leaves your phone. This is not a philosophical stance - it is an architectural requirement we set before writing a single line of code. The biomarker model runs on the Neural Engine. Lab parsing runs on-device. Trend detection runs on-device. You own your health data, and it stays with you. Read our full privacy policy for details on what Helix does and does not collect.

Results

4.9

App Store Rating

Maintained across 2.1 million downloads. The reviews that matter are not the five-star ratings - they are the detailed ones from users who describe specific trends Helix identified that they then brought to their physician. Those are the reviews we read.

2.1M

Downloads

Without advertising, without app store optimization campaigns, without growth hacking. The distribution is predominantly word of mouth - users sharing Helix with family members and friends who ask what app they are using. That is the kind of growth we build for.

94d

Avg Lead Time

The average time between when Helix flags a trend and when a physician would have otherwise detected it through standard-of-care annual or semi-annual review. Not every flag leads to a clinical finding. But the ones that do, on average, would have taken 94 days longer to surface through routine care.