Why don't fitness apps talk to each other? Where injuries hide

The serious self-tracker does not have a data problem. They have a fragmentation problem: the watch knows the sleep, the lifting app knows the squats, the run app knows the tempo, and nothing knows all three. The person is the integration layer, and the person is tired.

Why do fitness apps not share data?

Mostly because each one was built to be the place you log, and the place you log is the product. A lifting app that sends your sets elsewhere has given away the reason to open it. So each app holds its own slice, exports grudgingly, and imports almost nothing.

The platforms have quietly fixed the plumbing. Apple Health on iPhone and Health Connect on Android are shared stores that most wearable apps already write to, and Hevy and Strava have APIs. The data can flow. What is missing is something that reads all of it and does the arithmetic across the slices, which is a different kind of product from a logger.

Why is training load only meaningful in total?

Tim Gabbett's work on the acute-to-chronic workload ratio made a simple point that the industry has spent a decade proving: injury risk rises when this week's load spikes relative to the last four, regardless of the absolute number. A ratio above about 1.5 was associated with a sharp rise in injury in the original cohorts.

The ratio only works if the load is complete. A heavy leg day counted by the lifting app and a tempo run counted by the run app are, to each app, moderate weeks. Added together they are a spike. No single-purpose app can see it, so no single-purpose app can warn you. The detail of the ratio, and how contested the exact threshold is, is in how much training load is too much.

PeakU does this from your own data. A coach that reads everything you already track. Join the free beta →

What does a complete load count look like?

Every session, from every source, on one intensity-weighted scale. Duration alone is not enough: ninety easy minutes and forty hard ones are not the same withdrawal, so each session is duration multiplied by an intensity factor. For a run that factor can come from heart rate; for a lift it has to come from sets, reps and load relative to your maxes, which is data only the lifting app has.

1. Pull workouts from the shared health store, strength sessions from the lifting app, runs from the run app.
2. Remove duplicates: the same run arriving from watch and Strava, a treadmill warm-up logged inside a gym session.
3. Score each session for intensity and duration on the same scale.
4. Sum this week; average the previous four; compare.

Steps one and two are where fragmentation bites. Nobody does them by hand for long, which is why the ratio, however useful, is almost never computed for a recreational athlete.

Why recovery signals are cross-domain too

Yesterday's squats depress today's HRV. Tonight's short sleep will make tomorrow's threshold run harder than the plan assumes. The lifting app never sees the HRV; the run app never sees the squats; the sleep app sees neither. The question a coach is asked, can I train hard today?, needs all of them in one place.

Even something as basic as a heart-rate zone depends on the whole picture. A zone set from a run while fatigued by lifting is a different zone from one set fresh, which is part of why finding your heart-rate zones without a lab starts from your own data rather than a formula.

What does fragmentation cost you?

Three things, in rising order of expense.

• Time. You become the integration layer: opening four apps each morning, holding the week in your head, re-entering what one app knows into another.
• Bad decisions. Each app answers its own narrow question honestly and the combination is wrong. The run app says go; it did not see the squats.
• Injuries. The spike that the ratio would have flagged was invisible because it was split across three places. It becomes visible as an Achilles.

Why the fix is not another app to log into

Adding a sixth app that asks you to re-enter what the other five already know is how most "unified" products fail. The data already flows: Samsung Health, Garmin, Whoop, Oura and Polar all write to Health Connect on Android and Apple Health on iPhone; Hevy and Strava have APIs. The missing piece is a reader that sits on top, not another logger.

Doing that well has its own traps, mostly duplicates: the same run arriving from the watch and from Strava, steps counted by phone and watch both. The step-by-step is in how to combine Garmin, Samsung Health, Strava and Hevy without double-counting. And when you are choosing software to sit on top, reading your existing data rather than asking you to re-log it is the first thing to check; what to look for in an AI fitness coach app puts it at the top of the list.

What PeakU does with the whole picture

PeakU reads sleep, HRV, resting heart rate, steps, weight and workouts from Apple Health or Health Connect, strength sessions from Hevy and runs from Strava, and reconciles them into one timeline with duplicates removed.

Load is computed across everything on one scale, the acute-to-chronic ratio is shown with its drivers, and the coach answers in terms of your actual week: your HRV is 12 ms under baseline and you squatted heavy yesterday, so keep the session but drop the squats. You keep logging where you already log. PeakU is the one place you go to understand it.

Hen Eytan · Founder, UPeak

Builds the UPeak apps and writes up the research behind them.