EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Development and validation of an AI-driven blood glucose prediction system for pilots based on continuous glucose monitoring

Dalong Guo, Guangyun Wang, Dexin Kong, Zhen Tian, Baosen Tan, Yufei Qin, Zhaoting Yang, Yubin Zhou, Haocheng Yin, Yimeng Zhang

Frontiers in Physiology · 14.08.2026 · 17

DOI: 10.3389/fphys.2026.1902975 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Physiology, online erschienen am 14.08.2026. Verfasst von Dalong Guo und 9 weiteren Autor:innen. Beteiligte Einrichtung: Air Force Medical Center, Air Force Medical University, Beijing, China. Die Arbeit nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

During flight, pilots are exposed to unique environmental factors such as low air pressure, acceleration, and psychological stress, which can lead to altered glucose metabolism that poses a threat to flight safety. Traditional fingerstick blood glucose monitoring has drawbacks such as the need for intermittent readings, time-consuming procedures, and passive monitoring. Conventional continuous glucose monitoring (CGM) can only display historical and real-time glucose data and lacks predictive capabilities. In the present study, an artificial intelligence (AI)-driven blood glucose prediction system for pilots based on CGM and a multimodal transformer was developed with the aim of achieving proactive early warning of blood glucose risks among pilots. Sixty healthy male volunteers aged 25–50 years were recruited. Subjects completed 180 simulated flight experiments, during which multimodal data such as CGM glucose level, heart rate, blood oxygen saturation, and physical activity levels, were simultaneously collected to establish a scenario-specific dataset in aviation. A multimodal transformer model was developed to predict future blood glucose levels at 15, 30, and 60 min, and its performance was compared with baseline models such as linear extrapolation, long short-term memory (LSTM), and gated recurrent unit (GRU).The results showed that the mean absolute percentage error (MAPE) of the model for the prediction of blood glucose at 30 min was 7.2%, and the root mean square error (RMSE) was 9.4 mg/dL, which indicated a significantly superior prediction performance compared with the baseline models. For hypoglycemia event prediction, sensitivity was 92.5%, specificity was 89.7%, and the area under the receiver operating characteristic (ROC) curve was 0.94. A Clarke error grid analysis revealed that 98.7% of the predicted points fell within the clinically accurate region, which could potentially provide pilots with approximately 32 min of lead time for intervention. This system proved capable of accurately predicting blood glucose fluctuations during flight. Therefore, it may aid in shifting the metabolic health management of pilots from passive monitoring to proactive prevention, and serve as a novel tool for assuring aviation medical safety.

Schlagwörter

Stichwörter der Autor:innen

aviation medicineaviation pilotcontinuous glucose monitoring (CGM)glucose predictionmultimodal fusiontransformer

Autor:innen und Institutionen

  • Air Force Medical Center, Air Force Medical University, Beijing, China.
  • Guangzhou Institute of Technology, Xidian University, Guangzhou, China.
  • School of Microelectronics, Key Laboratory of Wide Band-Gap Semiconductor Materials and Devices, Xidian University, Xi'an, China.

Zitieren

APA

Guo, D., Wang, G., Kong, D., Tian, Z., Tan, B., Qin, Y., Yang, Z., Zhou, Y., Yin, H., & Zhang, Y. (2026). Development and validation of an AI-driven blood glucose prediction system for pilots based on continuous glucose monitoring. Frontiers in Physiology, 17. https://doi.org/10.3389/fphys.2026.1902975

Vancouver

Guo D, Wang G, Kong D, Tian Z, Tan B, Qin Y, et al. Development and validation of an AI-driven blood glucose prediction system for pilots based on continuous glucose monitoring. Frontiers in Physiology. 2026;17. doi:10.3389/fphys.2026.1902975

BibTeX
@article{Guo2026,
  author = {Guo, Dalong and Wang, Guangyun and Kong, Dexin and Tian, Zhen and Tan, Baosen and Qin, Yufei and Yang, Zhaoting and Zhou, Yubin and Yin, Haocheng and Zhang, Yimeng},
  title = {Development and validation of an AI-driven blood glucose prediction system for pilots based on continuous glucose monitoring},
  journal = {Frontiers in Physiology},
  year = {2026},
  volume = {17},
  doi = {10.3389/fphys.2026.1902975}
}
Kostenlos starten →