Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments

Joel Strickland; Marco  Ghisoni; Hannah Marshall; Thomas  Whitehead; Bogdan  Nenchev; Ben  Pellegrini; Charles  Phillips; Karl  Tassenberg; Sarah Davey; Sandra  Dorman; Joseph Sol; David  Ferguson; Gareth  Conduit

doi:10.1038/s44172-025-00548-6

Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments

Joel Strickland
, Marco Ghisoni
, Hannah Marshall
, Thomas Whitehead
, Bogdan Nenchev
, Ben Pellegrini
, Charles Phillips
, Karl Tassenberg
, Sarah Davey
, Sandra Dorman
, Joseph Sol
, David Ferguson
, Gareth Conduit

Research output: Contribution to journal › Article › peer-review

4 Downloads (Pure)

Abstract

Accurate estimation of core body temperature (CBT) is essential for physiological monitoring, yet current non-invasive methods lack statistically calibrated uncertainty estimates required for safety-critical use. Here we introduce a conformal deep learning framework for real-time, non-invasive CBT prediction with calibrated uncertainty, demonstrated in high-risk heat-stress environments. Developed from over 140,000 physiological measurements across six operational domains, the model achieves a test error of 0.29 °C, outperforming the widely used ECTemp™ algorithm with a 12-fold improvement in calibrated probabilistic accuracy and statistically valid prediction intervals. Designed for integration with wearable devices, the system uses accessible physiological, demographic, and environmental inputs to support practical, confidence-informed monitoring. A customizable alert engine enables proactive safety interventions based on user-defined thresholds and model confidence. By combining deep learning with conformal prediction, this approach establishes a generalizable foundation for trustworthy, non-invasive physiological monitoring, demonstrated here for CBT under heat stress but applicable to broader safety-critical settings.

Original language	English
Article number	219
Number of pages	12
Journal	Communications Engineering
Volume	4
Issue number	1
DOIs	https://doi.org/10.1038/s44172-025-00548-6
Publication status	Published - 20 Nov 2025

Bibliographical note

© Crown 2025. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included
in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative
Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission
directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/

Funding

The authors thank the data providers for their contributions to this study. Special acknowledgement goes to David Ferguson and Abigail Faltus for acquiring the racing car driver (RACE) dataset. The explosive ordnance disposal (EOD) data were provided by Doug Thake, derived from an MRes thesis and ongoing research by Dirk Dugdale-Duwell, focusing on the efficacy of liquid cooling suits in mitigating thermal strain during explosive ordnance disposal tasks. The nuclear (NUC) data was acquired by Hannah Marshall and Sarah Davey, in collaboration with Tom Frigon at the Palo Verde Nuclear Power Generating Facility in Phoenix, Arizona, and was funded, in part, by Innovate UK under project code 13302. Sandra Dorman collected the mining (MINE) data during extensive fieldwork in operational mining environments at Laurentian University. The authors also thank Sarah Davey from Coventry University for providing the factory worker (FACT) dataset. The wildland firefighter (WFF) dataset was provided by the National Technology and Development Program, United States Department of Agriculture Forest Service. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the United States Department of Agriculture, Forest Service.

Funders	Funder number
Innovate UK	13302

Keywords

Occupational heat stress
Estimating Core Temperature
wearable technology

Access to Document

10.1038/s44172-025-00548-6Licence: CC BY

Strickland et al 2025Accepted author manuscript, 1.61 MBLicence: CC BY
Davey2025VoRFinal published version, 2.74 MBLicence: CC BY

Cite this

Strickland, J., Ghisoni, M., Marshall, H., Whitehead, T., Nenchev, B., Pellegrini, B., Phillips, C., Tassenberg, K., Davey, S., Dorman, S., Sol, J., Ferguson, D., & Conduit, G. (2025). Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments. Communications Engineering, 4(1), Article 219. https://doi.org/10.1038/s44172-025-00548-6

Strickland, J, Ghisoni, M, Marshall, H, Whitehead, T, Nenchev, B, Pellegrini, B, Phillips, C, Tassenberg, K, Davey, S, Dorman, S, Sol, J, Ferguson, D & Conduit, G 2025, 'Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments', Communications Engineering, vol. 4, no. 1, 219. https://doi.org/10.1038/s44172-025-00548-6

@article{7ef399acb13c409f90e71e6f24d4fc80,

title = "Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments",

abstract = "Accurate estimation of core body temperature (CBT) is essential for physiological monitoring, yet current non-invasive methods lack statistically calibrated uncertainty estimates required for safety-critical use. Here we introduce a conformal deep learning framework for real-time, non-invasive CBT prediction with calibrated uncertainty, demonstrated in high-risk heat-stress environments. Developed from over 140,000 physiological measurements across six operational domains, the model achieves a test error of 0.29 °C, outperforming the widely used ECTemp{\texttrademark} algorithm with a 12-fold improvement in calibrated probabilistic accuracy and statistically valid prediction intervals. Designed for integration with wearable devices, the system uses accessible physiological, demographic, and environmental inputs to support practical, confidence-informed monitoring. A customizable alert engine enables proactive safety interventions based on user-defined thresholds and model confidence. By combining deep learning with conformal prediction, this approach establishes a generalizable foundation for trustworthy, non-invasive physiological monitoring, demonstrated here for CBT under heat stress but applicable to broader safety-critical settings.",

keywords = "Occupational heat stress, Estimating Core Temperature, wearable technology",

author = "Joel Strickland and Marco Ghisoni and Hannah Marshall and Thomas Whitehead and Bogdan Nenchev and Ben Pellegrini and Charles Phillips and Karl Tassenberg and Sarah Davey and Sandra Dorman and Joseph Sol and David Ferguson and Gareth Conduit",

note = "{\textcopyright} Crown 2025. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/",

year = "2025",

month = nov,

day = "20",

doi = "10.1038/s44172-025-00548-6",

language = "English",

volume = "4",

journal = "Communications Engineering",

issn = "2731-3395",

publisher = "Springer Nature",

number = "1",

}

TY - JOUR

T1 - Degrees of uncertainty

T2 - conformal deep learning for non-invasive core body temperature prediction in extreme environments

AU - Strickland, Joel

AU - Ghisoni, Marco

AU - Marshall, Hannah

AU - Whitehead, Thomas

AU - Nenchev, Bogdan

AU - Pellegrini, Ben

AU - Phillips, Charles

AU - Tassenberg, Karl

AU - Davey, Sarah

AU - Dorman, Sandra

AU - Sol, Joseph

AU - Ferguson, David

AU - Conduit, Gareth

N1 - © Crown 2025. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/

PY - 2025/11/20

Y1 - 2025/11/20

N2 - Accurate estimation of core body temperature (CBT) is essential for physiological monitoring, yet current non-invasive methods lack statistically calibrated uncertainty estimates required for safety-critical use. Here we introduce a conformal deep learning framework for real-time, non-invasive CBT prediction with calibrated uncertainty, demonstrated in high-risk heat-stress environments. Developed from over 140,000 physiological measurements across six operational domains, the model achieves a test error of 0.29 °C, outperforming the widely used ECTemp™ algorithm with a 12-fold improvement in calibrated probabilistic accuracy and statistically valid prediction intervals. Designed for integration with wearable devices, the system uses accessible physiological, demographic, and environmental inputs to support practical, confidence-informed monitoring. A customizable alert engine enables proactive safety interventions based on user-defined thresholds and model confidence. By combining deep learning with conformal prediction, this approach establishes a generalizable foundation for trustworthy, non-invasive physiological monitoring, demonstrated here for CBT under heat stress but applicable to broader safety-critical settings.

AB - Accurate estimation of core body temperature (CBT) is essential for physiological monitoring, yet current non-invasive methods lack statistically calibrated uncertainty estimates required for safety-critical use. Here we introduce a conformal deep learning framework for real-time, non-invasive CBT prediction with calibrated uncertainty, demonstrated in high-risk heat-stress environments. Developed from over 140,000 physiological measurements across six operational domains, the model achieves a test error of 0.29 °C, outperforming the widely used ECTemp™ algorithm with a 12-fold improvement in calibrated probabilistic accuracy and statistically valid prediction intervals. Designed for integration with wearable devices, the system uses accessible physiological, demographic, and environmental inputs to support practical, confidence-informed monitoring. A customizable alert engine enables proactive safety interventions based on user-defined thresholds and model confidence. By combining deep learning with conformal prediction, this approach establishes a generalizable foundation for trustworthy, non-invasive physiological monitoring, demonstrated here for CBT under heat stress but applicable to broader safety-critical settings.

KW - Occupational heat stress

KW - Estimating Core Temperature

KW - wearable technology

UR - https://www.scopus.com/pages/publications/105025684395

U2 - 10.1038/s44172-025-00548-6

DO - 10.1038/s44172-025-00548-6

M3 - Article

SN - 2731-3395

VL - 4

JO - Communications Engineering

JF - Communications Engineering

IS - 1

M1 - 219

ER -

Degrees of uncertainty: conformal deep learning for non-invasive core body temperature prediction in extreme environments

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this