Storing sparse/non-continuous peripheral physiological data

[me-pic]

Your idea

Related to PR #1128 (see also our internal discussion)

Dear BIDS community,

As part of our effort to write the BEP on Peripheral Physiological Recordings (BEP045), we were discussing what format should be used to store sparse / non-continuous physiological recordings, and to store continuous data that has non-constant sampling rate. An example of sparse data would be pulse timestamps outputted by some pulse oximeters, which are not linked to any continuous timeseries.

The physioevents file introduced in BEP 020 Eye Tracking (PR #1128) seems to be what would fit the most this purpose, already offering an associable timeseries/duration column to the non-constant recording values. However, we are proposing some changes:

• Description of Physiology “events”: currently the description implies a dependence between the data stored in the <matches>_recording-<label>_physioevents.tsv.gz, and the <matches>_physio.tsv.gz. As shown in the example above, sparse data can be acquired without any continuous data associated.
• Relaxation of OnsetSource: for the reason stated in the point above, the requirement level for OnsetSource should be RECOMMENDED instead of REQUIRED.

We were wondering if those changes/suggestions make sense, or if it would be better to create a different type of file to store those types of data.

Thank you in advance for your input !

[neuromechanist]

Hi @me-pic,

Would you mind also providing examples that fits your question (sparse and/or with irregular sampling)?

Which time series data are you considering that benefits most from irregular recording?

For example, are you planning to enter EHR data domain?