od-activity-rhythm

Analysis code, manuscript, and supplementary materials for:

Olfactory dysfunction, daytime activity reduction, and 24-hour rhythm fragmentation: but not food-odor recognition: in NHANES 2013-2014

Ceyhun Olcan, Dartmouth College.

• medRxiv preprint: https://doi.org/10.1101/2026.MM.DD.XXXXXXXX (DOI pending screening; typically 2-4 business days after submission)
• OSF pre-registration (paper #2 follow-up): https://doi.org/10.17605/OSF.IO/ZX8RN
• Zenodo archive of this repo (citable concept DOI): https://doi.org/10.5281/zenodo.20132927

What's here

paper/
  manuscript.md              Markdown source
  manuscript.pdf             Text-only PDF (14 pages)
  manuscript_complete.pdf    Text + tables + 5 figures (22 pages)
  STROBE_checklist.md/.pdf   STROBE Statement v4 compliance
  tables/                    Main tables (1-4) as md, pdf, and CSV
  figures/                   Main figures (1-5) as pdf and png
  supplementary/             Supplementary appendix + 21 supp tables + 2 supp figures
src/
  stage1_build_analytic.py             construct analytic dataset from 18 NHANES XPTs
  stage25_extract_hourly.py            hourly mean MIMS per participant
  stage8_minute_level_fragmentation.py bouts/transitions/hazards from PAXMIN_H
  stage30_analysis.R                   primary regression, FDR, MICE pooled
data/
  attrition_log.csv          STROBE attrition counts
  analytic_seqn_list.csv     2,327 SEQNs in the final analytic sample
docs/
  variable_dictionary.md     data dictionary for analytic_full.csv

The actual NHANES XPT files are not redistributed: they live at https://wwwn.cdc.gov/nchs/nhanes/ and are downloaded into the working dir before running stage 1.

Reproducing the analysis

1. Download NHANES 2013-2014 cycle files: DEMO_H, BMX_H, BPX_H, CSX_H, CSQ_H, PAXMIN_H, PAXHD_H, PAQ_H, SMQ_H, DIQ_H, GHB_H, MCQ_H, BPQ_H, DPQ_H, INQ_H, HUQ_H, RXQ_RX_H, RDQ_H, SLQ_H.

2. From the directory containing the XPTs:

   python src/stage1_build_analytic.py
   python src/stage25_extract_hourly.py \
       --paxmin PAXMIN_H.xpt --paxhd PAXHD_H.xpt \
       --features paxmin_features.csv \
       --seqn analytic_seqn_list.csv \
       --out paxmin_output
   python src/stage8_minute_level_fragmentation.py
   Rscript src/stage30_analysis.R

   Stage 8 takes ~25 min on a 2022 MacBook Air; everything else is faster.

Dependencies

• Python 3.10+ with pandas, numpy
• R 4.3+ with survey, mice, dplyr, readr

See requirements.txt and r_requirements.txt.

Data dictionary

See docs/variable_dictionary.md. The most-used variables:

name	what
SEQN	NHANES participant ID
PST_correct	Pocket Smell Test, # correctly identified (0-8)
od_binary	1 if PST_correct ≤ 5, else 0
mean_mims	per-day mean MIMS units
mvpa_min	min/day in MVPA (Karas: ≥ 37.5 MIMS)
IS, IV	interdaily stability, intradaily variability
ASTP_minute	active-to-sedentary transition probability (minute-level)
SATP_minute	sedentary-to-active transition probability
WTMEC2YR, SDMVSTRA, SDMVPSU	NHANES MEC weights and clustering

Citing

If you use this code or build on these findings, please cite:

Olcan C. Olfactory dysfunction, daytime activity reduction, and 24-hour rhythm fragmentation: but not food-odor recognition: in NHANES 2013-2014. medRxiv 2026 (preprint, DOI pending).

For the code archive specifically: cite the Zenodo concept DOI 10.5281/zenodo.20132927.

License

• Code: MIT (see LICENSE)
• Manuscript and figures: CC BY 4.0 (matches Lancet Healthy Longevity policy)
• NHANES data: U.S. public domain (CDC/NCHS)

Contact

ceyhun.olcan.27@dartmouth.edu | ORCID 0000-0002-6326-6071