What is the best way to calculate sleep regularity index?

[dora201888]

Good afternoon,

I am currently attempting to calculate the Sleep Regularity Index (SRI) using GGIR.

(1) Based on the GGIR release notes, I understand that version 3.3.0 includes both SRI1 and SRI2 implementations, but I encountered errors when running Part 6.

_

**> Part 6

Busy processing ... see /data/guow4/ProjectX2_2021/0_mMARCH_actigraph/Korea/GGIR/output_data/meta/ms6.out for progress

Errors and warnings for 001jyj__103506_2024-07-18 15-47-35_T5A5.RData$message
[1] "object 'params_general' not found"

$call
main_part6_recordlevel(i, metadatadir, f0, f1, fnames.ms5raw,
ffdone, EXT, verbose)**

_

(2) The second question is if GGIR can output an epoch-level 0/1 sleep–wake status matrix, which is needed for sleepreg.

(3) As an alternative approach, I am considering calculating the original SRI directly using epoch-level data from GGIR (i.e., the /meta/csv output) together with the updated sleepreg package.

Any guidance or suggestions would be greatly appreciated.

Thank you very much for your help.

Best regards,
Dora

[vincentvanhees]

Hi Dora,

Re. 1) The issue was fixed in 3.3-1 (only on GitHub), and specific to running part 6 in parallel. So, either run it serially or use the latest GitHub release. It will be part of the next CRAN release in the new year.

Re. 2) and 3) I am afraid there is not. However, GGIR part 6 works with the time series exported by part 5. So, in theory you could take those time series files (stored in meta/ms5.rawout) and copy the code from g.part6 and step by step go through the code to understand how GGIR calculates GGIR around line 532.