Seeking Clarification on Data Exclusion Principles Across GGIR Parts 2, 4, and 5

[zhu0313]

I am writing to seek your expert guidance regarding the data cleaning and exclusion logic in my recent GGIR analysis. I apologize for the frequent inquiries lately, and I truly appreciate your continued patience and support as I navigate the complexities of this tool.
While comparing the summary reports across different stages, I observed a stepwise decrease in my sample size:
· Part 4 vs. Part 2: 4 IDs were excluded.
· Part 5 vs. Part 2: 6 IDs were excluded in total.
· ms3.out vs. ms2.out: 2 IDs did not generate intermediate files in the ms3.out folder.
I have carefully examined the Quality Control (QC) metrics in the part2_summary and the _full sleep reports. However, I am struggling to identify a consistent pattern or a single common reason (such as a specific calib_err or nonwear threshold) that explains why these specific participants were dropped at different stages.
Could you please clarify the specific exclusion principles for each Part? Specifically:
1. What criteria must a file meet to transition from Part 2 to Part 4 (Sleep) beyond basic calibration?
2. What additional filters are applied in Part 5 (Activity) that might lead to further exclusions (e.g., includedaycrit or 24-hour coverage)?
3. Is there a specific section in the GGIR documentation or a published paper that provides a detailed flowchart of these "hidden" exclusion checkpoints?
I want to ensure that my data cleaning process is transparent and reproducible for my study. Thank you very much for your time and for helping me better understand the underlying logic of GGIR.

[vincentvanhees]

The GGIR documentation https://wadpac.github.io/GGIR/ explains the differences between the parts and clarifies that part 2, 4 and 5 all store extra csv files in the QC-output subfolder to help you diagnose why a certain individual was excluded from the 'cleaned' output.

Each part of GGIR looks at different time segments and puts different criteria on accepting those time segments as valid. This means that one individual can pass the criteria in part 2 but not in part 4 or 5. Part 2 is most tolerant, while part 5 is most restrictive.

I am afraid this is all the free help I can offer here. Happy to help you as part of a paid consultancy. Another option for you to consider is the GGIR training as offered by @jhmigueles via his website https://jhmigueles.com