Fig. 4: Subject-session level RMSE comparison between Deep Learning and ECTemp™ models across domains. | Communications Engineering

Fig. 4: Subject-session level RMSE comparison between Deep Learning and ECTemp™ models across domains.

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

Fig. 4: Subject-session level RMSE comparison between Deep Learning and ECTemp™ models across domains.

Box plots display RMSE values for ECTemp™ (blue) and the Deep Learning model (green), with lower RMSE values indicating better performance. The Deep Learning model shows consistently lower and less variable RMSE across most domains, suggesting improved predictive accuracy and stability. Domains represented are wildland firefighters (WFF, n = 152), race-car drivers (RACE, n = 72), nuclear plant workers (NUC, n = 29), factory workers (FACT, n = 116), mine workers (MINE, n = 14), and explosive ordnance disposal technicians (EOD, n = 31). Only subject-sessions with at least 10 data points were included to ensure robust RMSE estimates.

Back to article page