ATHENA_MHP: Metabolic Heat Production During Heat Tolerance Testing

Reproducible analysis pipeline for metabolic heat production (MHP) and associated cardiopulmonary responses during the IDF heat tolerance test, before and after a 5-day heat acclimation protocol. Companion code to the Bowie et al. (2026) study cohort; manuscript in preparation.

Cohort publication

Bowie JS, Szymanski MR, Struder JF, et al. Heat tolerance classification criteria require population-specific thresholds for accurate assessment of acclimation state in adults. Physiological Reports. 2026;14(4):e70745. doi:10.14814/phy2.70745

This repository contains the metabolic-heat-production analysis derived from the same cohort.

Analysis pipeline

Scripts run in numbered order. Each .Rmd knits to a self-contained HTML report.

Script	Purpose
00_preprocessing.R	Raw Excel to processed CSV (wide + long format)
01_MHP_Analysis.Rmd	Descriptive statistics, paired t-tests, sex comparisons, three-Aim framing
02_Manuscript_Analysis.Rmd	Primary. LMM + emmeans, publication tables/figures (main + supplement)
02_MHP_Prism_Figures.Rmd	Prism-style figure library (paired-dot, raincloud, slopegraph variants)
03_compensability_audit.Rmd	Partitional-calorimetry compensability audit using a Weir-validated workflow
03_Statistical_Methods_Justification.Rmd	Literature precedents for every statistical method used
04_Statistical_Methods_Walkthrough.Rmd	Worked statistical walkthrough for student/reviewer onboarding

A Shiny calculator for the MHP partitional model is provided at apps/mhp_calculator/.

Quick start

# 1. Open the project in RStudio (the .Rproj file at the repo root) or set
#    the working directory to the repo root.
# 2. First time only:
source("_init_renv.R")

# 3. Subsequent runs (requires raw cohort data — see "Data access" below):
renv::restore()                                  # pin packages
source("00_preprocessing.R")                     # only if raw data changed
rmarkdown::render("02_Manuscript_Analysis.Rmd")  # primary publication analysis
rmarkdown::render("01_MHP_Analysis.Rmd")         # descriptive companion

The rendered figure library in output_prism/ is committed — you can examine all main and supplementary figures without re-running the pipeline.

Methods summary

• N = 38 (Year 1 and 2 cohort; Subjects 4 and 152 excluded).
• Measures: V̇O₂, RER, MHP (W, W/m², W/kg LBM), T_re, HR.
• Windows: T30-40 and T90-100 minutes of HTT1, matching metabolic-cart measurement periods.
• Statistical approach: Linear mixed models fit with lme4 / lmerTest, estimated marginal means via emmeans, Kenward-Roger denominator degrees of freedom, bias-corrected Hedges' g effect sizes.
• Multiple comparisons: Benjamini-Hochberg FDR within physiological domain families.
• Stratification: combined cohort, male-only, female-only.
• Reproducibility: renv.lock pins every package; the full figure library (output_prism/) is committed for examination without re-running the pipeline.

Key findings

1. MHP reduced approximately 6 percent uniformly across measurement windows (consistent with Rivas et al. 2017).
2. HR is the only variable showing a Condition by Window interaction (temporal dissociation).
3. RER shows a sex-dependent HA response (Condition by Sex, p = 0.009).

Repository layout

ATHENA_MHP/
├── 00_preprocessing.R                          # Raw to processed CSV (needs external raw data)
├── 01_MHP_Analysis.Rmd                         # Descriptive analysis
├── 02_Manuscript_Analysis.Rmd                  # Primary publication analysis
├── 02_MHP_Prism_Figures.Rmd                    # Prism-style figure library
├── 03_compensability_audit.Rmd                 # Compensability audit
├── 03_Statistical_Methods_Justification.Rmd    # Methods literature scaffold
├── 04_Statistical_Methods_Walkthrough.Rmd      # Methods worked walkthrough
├── apps/mhp_calculator/                        # Shiny MHP partitional calculator
├── helpers/                                    # Shared R utilities + thermal helpers
├── output_prism/                               # Prism-style figure library (tracked PNGs)
├── docs/                                       # Methods + data manifest + literature reviews
├── renv/                                       # Pinned R environment (lockfile tracked)
├── CITATION.cff
├── LICENSE
└── README.md

Data access

No subject-level data ships in this repository. Raw and processed cohort data are available from the corresponding author on reasonable request; see the cohort publication for protocol details and inclusion criteria:

Bowie JS et al. Physiological Reports. 2026;14(4):e70745. doi:10.14814/phy2.70745

What this repo provides without external data:

• The full analysis pipeline (.Rmd files + helpers/)
• The rendered figure library (output_prism/) for review
• Methods documentation (docs/)
• Pinned package environment (renv.lock)

The original study was IRB-approved at the University of Connecticut.

Reproducibility notes

• Two males with missing DXA data use sex-specific mean replacement, handled deterministically in 00_preprocessing.R.
• WBGT data-entry errors for Subjects 9 and 12 are corrected in-script after data load.
• Seed is set in 00_preprocessing.R and propagated through renv to ensure stable randomization on re-runs.

Citation

If you use this code, cite both the cohort publication (above) and this repository. A CITATION.cff file is provided so GitHub's "Cite this repository" widget renders automatically.

@software{bowie_athena_mhp_2026,
  author       = {Bowie, Jacob S.},
  title        = {ATHENA\_MHP: Metabolic Heat Production Analysis Code},
  year         = {2026},
  url          = {https://github.com/JacobBowie/ATHENA_MHP},
  version      = {0.1.0}
}

License

MIT for code.