Sampling rate for health data on Apple Watch

I have collected three months of data and exported it with the Health Auto Export App. I am interested in the sample rate (frequency) as well as the duration and timing of sampling epochs (e.g., intermittency).


I suspect that the intermittency of sampling for different health metrics varies, and in different ways, depending on the situation (e.g., activity level), but are there some general rules of thumb for the duration and timing of sampling epochs for such metrics as energy burned, heart rate, and steps?


Are the differences in sampling epochs (e.g., in my data depicted below) due to the way data are sampled, the way they are processed, the way they are stored, or the way they are exported?

Posted on Mar 19, 2023 5:46 PM

Reply

Similar questions

3 replies

Mar 24, 2023 5:03 AM in response to LD150

Below is a qualitative summary of Healthkit data that are available through the Health Export App. I have been in contact with developers of that app, and they have been very responsive. They currently are looking at my exports of quantitative data to answer these questions, but they may not be able to provide answers to my questions, so here is a bit more detail about what I am looking for.


The reply from LD150 is getting closer to the answer I need. Thank you. In explaining that "all data is there," I assume LD150 means that the Apple Healthkit does not try to filter out outliers or artifacts and, similarly, does not do any data aggregation in data that are made available to app developers. Is that right? If so, that is quite useful to know.


It also begs the question about what LD150 means by "all" and "there." When an app queries Healthkit, what exactly does it have access to? It surely isn't an analog signal, so it must be quantized at some density per unit of time (e.g., days, hours, minutes, seconds, fractions of a second). Even if these time scales vary, it would quite useful to know the rough order of magnitude for the time scales in which data on various metrics are quantized. I would value any thoughts anyone has about this.


Finally, some metrics available in Healthkit are summary statistics (e.g., Heart Rate Variability, Wrist Temperature) or metrics derived from multiple analog signals (e.g., Resting Energy, Active Energy, Walking Asymmetry, Double Support Time). There must be analysis epochs over which these data are derived and stored. Any insight into the length of those analysis epochs would be quite useful, even if only rough order of magnitude.


This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

Sampling rate for health data on Apple Watch

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.