EUVIMEDEUVIMED entdecken
Crossref

TReMR: Temporal-Relevance mRMR for HRV-Based Glucose Prediction Under Physiological Delay

Cyrus Mexon Evrard Djindot, Bin Chen, Israel Mbala Ntumba, Jessie Chen, Weixiong Rao, Sylvain Laborde, Qinpei Zhao

IEEE Instrumentation & Measurement Magazine · 2026-11 · 29 (8) · 49-62

DOI: 10.1109/mim.2026.11224913 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Instrumentation & Measurement Magazine, 2026-11. Verfasst von Cyrus Mexon Evrard Djindot und 6 weiteren Autor:innen. Die Arbeit nennt 11 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Heart rate variability (HRV)-based machine learning for glucose prediction relies on feature selection methods that were developed for instantaneous physiological classification—where signal and target are temporally aligned. In metabolic prediction tasks, however, autonomic regulation influences glucosedynamics through delayed pathways, meaning that features most discriminative of the current state may not be the most predictive of future outcomes. We introduce Temporal-Relevance mRMR (TReMR), a minimal extension of the classical minimum Redundancy-Maximum Relevance (mRMR) feature selection framework that redefines relevance using the mutual information between HRV features at time$t$and glucose values at a future time t + τ. Evaluated on the D1NAMO free-living dataset under strict Leave-One-Subject-Out (LOSO) cross-validation, TReMR improves cross-subject glucose-state classification while reducing feature dimensionality by approximately 10-fold (97 to 10 features). Complementary regression experiments reveal that HRV encodes short-term glucose dynamics more reliably than absolute glycemic levels, with predictive performance peaking at a τ = 15-minute horizon consistent with known autonomic-metabolic delay mechanisms.

Zitieren

APA

Djindot, C. M. E., Chen, B., Ntumba, I. M., Chen, J., Rao, W., Laborde, S., & Zhao, Q. (2026). TReMR: Temporal-Relevance mRMR for HRV-Based Glucose Prediction Under Physiological Delay. IEEE Instrumentation & Measurement Magazine, 29(8), 49-62. https://doi.org/10.1109/mim.2026.11224913

Vancouver

Djindot CME, Chen B, Ntumba IM, Chen J, Rao W, Laborde S, et al. TReMR: Temporal-Relevance mRMR for HRV-Based Glucose Prediction Under Physiological Delay. IEEE Instrumentation & Measurement Magazine. 2026;29(8):49-62. doi:10.1109/mim.2026.11224913

BibTeX
@article{Djindot2026,
  author = {Djindot, Cyrus Mexon Evrard and Chen, Bin and Ntumba, Israel Mbala and Chen, Jessie and Rao, Weixiong and Laborde, Sylvain and Zhao, Qinpei},
  title = {TReMR: Temporal-Relevance mRMR for HRV-Based Glucose Prediction Under Physiological Delay},
  journal = {IEEE Instrumentation & Measurement Magazine},
  year = {2026},
  volume = {29},
  number = {8},
  pages = {49--62},
  doi = {10.1109/mim.2026.11224913}
}
Kostenlos starten →