Published May 17, 2026 | Version v1
Computational notebook Open

An orthogonal-oracle Gaussian-process surrogate with no-fitted-parameter oxygen-content coupling for rapid-onset G-LOC tolerance under FiO₂-equivalent hypoxia

Authors/Creators

  • 1. Colombian Aerospace Force
  • 2. Fuerza Aerea Colombiana

Description

High-altitude aviation routinely combines sustained +Gz acceleration with reduced inspired oxygen, yet no quantitative tolerance model jointly captures the two stressors. We develop and evaluate a non-classical multi-fidelity Gaussian-process framework — an orthogonal-oracle dual-surrogate composed through a closed-form physiological coupling — that predicts the corrected conditional G-LOC induction time, defined among event-positive trajectories of CGEM's E[time | event = 1] regressor channel, under combined acceleration and hypoxia in the rapid-onset regime. Two physiological simulators with disjoint feature support — the Civil Aerospace Medical Institute G-Effects Model (acceleration only) and the Pulse Physiology Engine (inspired-oxygen-fraction only) — are coupled through a no-fitted-parameter multiplicative term derived from the Hüfner arterial-oxygen-content equation. An orthogonal-oracle Gaussian-process surrogate, fit per arm on its active design axes, yields calibrated prediction intervals through split-conformal calibration. Validation anchors against the foundational 888-episode centrifuge corpus of Whinnery and Forster (2013). Mean absolute error against the published closed-form rapid-onset prediction is 0.58 s (n = 47 in-scope rows). 100 % of the eight rapid-onset bin-mean predictions with design coverage fall inside the published within-bin distributional envelope (|z| < 1.96; median |z| = 0.224) — a distribution-envelope plausibility diagnostic, not a mean-level agreement claim. Sobol variance decomposition attributes 25 % of corrected loss-of-consciousness-time variance — a first-order Sobol contribution of 0.2491 — to the Hüfner-coupled inspired-oxygen-fraction channel. Split-conformal calibration achieves 0.9067 marginal coverage at the nominal 0.90 target (calibration and test sets equal at 461; conformal quantile 0.0044 s). The framework is the first quantitative tolerance surrogate for joint acceleration and hypoxia in aviation physiology, scoped to rapid-onset profiles (Gz ≥ 4.7, onset rate ≥ 1.0 G/s), with a byte-level reproducible pipeline.

Files

Pulse_Research-public-0.1.0.zip

Files (240.0 kB)

Name Size Download all
md5:069550df95ed13abf1e55acffca21b7b
240.0 kB Preview Download

Additional details

Software

Repository URL
https://github.com/strikerdlm/Pulse_Research-public
Programming language
Python , TypeScript
Development Status
Active

References

  • Achermann, F., Stastny, T., Danciu, B., Kolobov, A., Chung, J. J., Siegwart, R., et al. (2024). WindSeer: real-time volumetric wind prediction over complex terrain aboard a small uncrewed aerial vehicle. Nat. Commun. 15(1):3507. doi: 10.1038/s41467-024-47778-4
  • Balandat, M., Karrer, B., Jiang, D. R., Daulton, S., Letham, B., Wilson, A. G., et al. (2020). BoTorch: a framework for efficient Monte-Carlo Bayesian optimization. Adv. Neural Inf. Process. Syst. (NeurIPS) 33, 21524–21538. Available at: https://proceedings.neurips.cc/paper/2020/hash/f5b1b89d98b7286673128a5fb112cb9a-Abstract.html
  • Besch, E. L., Werchan, P. M., Wiegman, J. F., Nesthus, T. E., and Shahed, A. R. (1994). Effect of hypoxia and hyperoxia on human +Gz duration tolerance. J. Appl. Physiol. 76(4), 1693–1700. doi: 10.1152/jappl.1994.76.4.1693
  • Bray, A., Webb, J. B., Enquobahrie, A., Vicory, J., Heneghan, J., Hubal, R., et al. (2019). Pulse Physiology Engine: an open-source software platform for computational modeling of human medical simulation. SN Compr. Clin. Med. 1(5), 362–377. doi: 10.1007/s42399-019-00053-w
  • Copeland, K., and Whinnery, J. E. (2023). Cerebral blood flow based computer modeling of Gz-induced effects. Aerosp. Med. Hum. Perform. 94(5), 409–414. doi: 10.3357/AMHP.6179.2023
  • Hutcheon, E. A., Vakorin, V. A., Nunes, A. S., Ribary, U., Ferguson, S., Claydon, V. E., et al. (2023). Comparing neuronal oscillations during visual spatial attention orienting between normobaric and hypobaric hypoxia. Sci. Rep. 13(1):18021. doi: 10.1038/s41598-023-45308-8
  • Kennedy, M. C., and O'Hagan, A. (2000). Predicting the output from a complex computer code when fast approximations are available. Biometrika 87(1), 1–13. doi: 10.1093/biomet/87.1.1
  • Le Gratiet, L., and Garnier, J. (2014). Recursive co-kriging model for design of computer experiments with multiple levels of fidelity. Int. J. Uncertain. Quantif. 4(5), 365–386. doi: 10.1615/Int.J.UncertaintyQuantification.2014006914
  • Lyons, T. J., Harding, R., Freeman, J., and Oakley, C. (1992). G-induced loss of consciousness accidents: USAF experience 1982–1990. Aviat. Space Environ. Med. 63(1), 60–66. PMID: 1550536.
  • Ohrui, N., Iino, Y., Kuramoto, K., Kikukawa, A., Okano, K., and Takada, K. (2024). G-induced loss of consciousness prediction using a support vector machine. Aerosp. Med. Hum. Perform. doi: 10.3357/AMHP.6301.2024
  • West, J. B. (2012). Respiratory Physiology: The Essentials, 9th Edn. Philadelphia, PA: Wolters Kluwer / Lippincott Williams & Wilkins.
  • Whinnery, J. E. (1989). Methods for describing and quantifying +Gz-induced loss of consciousness. Aviat. Space Environ. Med. 60(8), 798–802. PMID: 2673203.
  • Whinnery, J. E., and Forster, E. M. (2013). The +Gz-induced loss of consciousness curve. Extreme Physiol. Med. 2(1):19. doi: 10.1186/2046-7648-2-19