Kalman Filter Models for the Prediction of Individualised Thermal Work Strain

Abstract

It is important to monitor and assess the physiological strain of individuals working in hot environments to avoid heat illness and performance degradation. The body core temperature (Tc) is a reliable indicator of thermal work strain. However, measuring Tc is invasive and often inconvenient and impractical for real-time monitoring of workers in high heat strain environments. Seeking a better solution, the main aim of the present study was to investigate the Kalman filter method to enable the estimation of heat strain from non-invasive measurements (heart rate (HR) and chest skin temperature (ST)) obtained 'online' via wearable body sensors. In particular, we developed two Kalman filter models. First, an extended Kalman filter (EFK) was implemented in a cubic state space modelling framework (HR versus Tc) with a stage-wise, autoregressive exogenous model (incorporating HR and ST) as the time update model. Under the second model, the online Kalman filter (OFK) approach builds up thetimeupdateequationdependingonlyontheinitialvalueofTcandthelatestvalueofthe exogenous variables. Both models were trained and validated using data from laboratoryand outfield-based heat strain profiling studies in which subjects performed a high intensity military foot march. While both the EKF and OKF models provided satisfactory estimates of Tc, the results showed an overall superior performance of the OKF model (overall root mean square error, RMSE = 0.31 C) compared to the EKF model (RMSE = 0.45 C).

Accuracy of Algorithm to Non-Invasively Predict Core Body Temperature Using the Kenzen Wearable Device

“…However, accurately monitoring individuals’ T C on a daily basis at the job site is prohibitively invasive (i.e., rectal and esophageal probes) or expensive (i.e., gastrointestinal pill). Thus, it has been a long-standing goal of many entities to develop a way to non-invasively estimate T C using other physiological parameters like heart rate, skin temperature, and heat flux [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ].…”

“…As computational and machine learning approaches have become faster and more accessible over the last decade, there has been a renewed interest in algorithms designed to continuously predict T C (in real-time) based on various environmental and physiological parameters [ 17 ]. Various research groups have published T C algorithms that meet field-established accuracy standards through the use of easily obtainable physiological measurements (e.g., heart rate and skin temperature) collected during physical activity [ 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ]. However, while model accuracy is high, many of these algorithms were trained and validated on datasets involving primarily young fit males [ 11 , 12 , 13 , 14 , 16 , 18 , 19 ], or only hot conditions [ 19 ], or with minimal data for ground truth T C ≥ 38.5 °C [ 10 , 14 , 16 , 19 ]—the temperatures above which heat injuries and illnesses are most prevalent, and thus accuracy is of utmost importance.…”

“…Another important factor to consider is that many of the previous T C algorithms have been designed using various laboratory-grade physiological monitoring devices like chest straps for monitoring heart rate or thermocouples for measuring skin temperature [ 10 , 11 , 12 , 13 , 14 , 15 , 16 ]. While these physiological measurements using laboratory-grade equipment are highly accurate, these sorts of measurements are typically cumbersome to implement daily as they can interfere with worker comfort.…”