PEM & wearables: heart rate, HRV, and the limits of pacing data

PEM does not always appear immediately after exertion. That makes it difficult to judge in the moment which demands may later be associated with worsening. Continuous or regular measurements can therefore be useful mainly as additional documentation.

The most useful signal is the personal trajectory: how do activity, heart rate, sleep, or resting heart rate change around better and worse days? These data can complement lived experience, not replace it.

A heart-rate monitor can provide immediate biofeedback during physical or orthostatic demand. Workwell describes an approach in which, when a ventilatory/anaerobic threshold (V/AT) has been measured with appropriate CPET expertise, the corresponding heart rate can be used as an individual pacing reference.

That is different from predicting PEM: heart rate reflects current cardiovascular demand. Whether PEM develops later depends on many more factors, including physical, cognitive, and emotional exertion and the person’s current baseline.

Generic formulas based on a fixed percentage of age-predicted maximum heart rate are not a reliable individual PEM threshold in ME/CFS. Workwell notes that altered heart-rate responses in ME/CFS can make standard age-based formulas a poor fit.

If heart rate is used for pacing, the number should therefore be treated as a personal reference rather than a medical clearance or a guarantee that staying below it will prevent a crash.

HRV and resting heart rate respond to many influences, including sleep, infection, medication, hydration, stress, menstrual cycle, and measurement conditions. Changes may be interesting within a personal trajectory, but a single value is not specific to PEM.

Our longitudinal analyses so far support HRV and resting heart rate more as accompanying context signals. They do not currently provide a sufficiently robust stand-alone prediction of a later PEM episode.

Elara brings wearable data and self-report onto one timeline. The goal is to make daily demand and longitudinal change easier to interpret against a personal baseline.

Elara currently does not provide a validated 24–72-hour PEM prediction. The app is intended to support documentation and pacing, not imply a medical forecast.

Consumer wearables vary in accuracy across devices and situations and generally are not diagnostic medical devices for the pacing questions discussed here. Their data should add context rather than become the only basis for decisions.

For some severely affected people, wearing, checking, or interpreting a device can itself add burden. A useful system should reduce cognitive load rather than create a new set of obligations.