3.3-5

• Part 2: Walking bout detection expanded to allow for multiple cadence and acceleration thresholds. #1466

• Part 2: Re-factor and tidy-up code related to data_masking_strategy 3 and 5 and study dates file. The resolution for the rolling window used in data_masking_strategy 3 is now higher (5 seconds by default). #1458

• Part 4: Remove duplicated column N_nights_guider_corrected in output file part4_summary_sleep_cleaned.csv when parameter guider_cor_do=TRUE. #1448

3.3-4

• Part 2: Improved handling of corrupted RData files in part 1 or 2. #1447 and #1449

• Part 1: Minor tidy up of code related to splitting recordings. #1446

• General: Zenodo release generation again has not been triggered since release 3.2-8, this is now fixed by correcting mistake in json file. #1426

3.3-3

• Part 2: Step/cadence expanded with a variety of output variables #1398. Credits for funding this enhancement: Ben Maylor and Alex Rowlands.

• Part 1: Fix bug that caused GENEAsleep binary data to no longer to be recognised as such since the introduction of functionality for Parmay binary data. #1432

• Part 4:

  • Now attempts to handle scenario where an integer participant ID is saved by Excel with decimal places, e.g. 123 versus 123.00. #1256

  • Fix bug causing the error about missing object defaultGuiderCorrected. #1438

3.3-2

• Part 1: Fractional sampling rate for GENEActiv now automatically resampled to rounded sampling rate to ease and avoid ambiguity in epoch level aggregation #1408.

• Part 3: Fixed a bug in handling HASPT_ignore_invalid = NA when no guider (SPT) window was detected #1414

• Part 3: Fixed bug in the guider correction algorithm where the main SPT window was directly preceded or followed by candidate (discarded) windows. #1415

3.3-1

• Part 2: Replaces QC plot as pdf file by QC plots as png file per recording to help speed up report generation #1402

• Part 3: Fix bug in calculation of Oakley1994, Galland2012, and ColeKripke1997 #1404

• Part 6: Fixed bug triggering an error when running part 6 with do.parallel = TRUE. #1400

• Error handling: Added parameter "use_trycatch_serial" to optionally log file-level errors without interrupting GGIR when "do.parallel = FALSE". #1308

3.3-0

• Part 3: Adding option to auto-correct sleep guider that failed to detect the full main sleep period or incorrectly focused on a daytime nap. The new algorithm comes with new parameters guider_cor_maxgap_hrs, guider_cor_min_frac_sib, guider_cor_min_hrs, guider_cor_meme_frac_out, guider_cor_meme_frac_in, guider_cor_meme_min_hrs #1307

3.2-12

• Part 5: Deprecate experimental nap-detection functionality that was added in 2.5-2. This has been superseded by a much more simplified nap detection which is now discussed in Annex NapDetection.

• Part 4: Fix bug #1376 causing more nights to be included than desirable if sleeplog dates for at least one recording do not overlap with accelerometer recording.

• Part 2: Always omit spaces from the extracted ID unless the entire filename is used as ID. #1372

• Vignette: Clarify definition of acceleration and time ranges in gh-pages chapter 12. #1370

• All parts: Value of parameter desiredtz is now passed on via milestone data, which enables jointly processing data from multiple timezones in part 2-6 #1380

3.2-11

• Part 2: In day segment analysis N hours and N valid hours per day segment are now correctly averaged over valid days only, previously calculation included invalid days. This affects only the person level summary and no other columns are affected. #1350

• Part 4: Reload sleeplog when part 3 output is newer than the latest sleeplog export, previously sleeplog was only reloaded if sleeplog itself had changed. #1366

3.2-10

Part 1: Correct timestamp extraction from ActivPAL data. #1287

Part 5 and 6: Fix bug for the specific situation where parameters datadir or outputdir are set to a path that has the letters "csv" in one or multiple parent folder names. #1341

Visualreport: Now also processes multiple files in parallel like the rest of GGIR. #1351

Part 4: Fix bug that made it impossible to use basic format sleeplog in combination with sleepwindowType "TimeInBed" #1359

3.2-9

• Part 3: For Fitbit data. Fixed bug in storage of ID format was list should be character, by which part5 could not run. #1330

• Part 5:

  • Fix bug in extracting sleep diary when recording starts between midnight and 4am. #1337

  • Fix bug that caused sibreport to be stored with timestamps in UTC timezone. #1332