Physiological status monitoring and dynamic risk prediction for operators in extreme environments
Extreme environments are inherently dangerous places where there are high risks of injury, illness, and mortality to people who enter these arenas for occupational or vocational reasons. Mountaineers are one such group who are regularly exposed to extremely low temperatures, high wind speeds, high ultra-violet radiation, and low availability of oxygen. One solution to mitigating the risks posed to mountaineers is through the use of physiological monitoring systems. Physiological monitoring can provide an early insight into physiological disregulation, and allow for interventions to be realised in response to observed changes in an individual’s physiological state [1]. Laboratory or in-hospital devices are not suitable for deployment in harsh environments, therefore wearable devices are a popular alternative [2]. There is a vast array of devices which can measure an extensive range of physiological parameters; common metrics include heart rate, respiratory rate, and temperature. Wearable devices have been shown to be effective and non-intrusive on the subjects’ activities. Accordingly, they are the preferred option in extreme environments as they allow for non-invasive, continual evaluation of a user’s metrics [3]. The overall aim of this strategy is to reduce injury, mortality, and the morbidity of people operating within these environments. This, in turn, may increase the efficiency and effectiveness of the operations they are involved in.
Despite the promise of physiological monitoring and extensive research into the field, significant challenges still persist. These challenges prevent the routine wide scale adoption of physiological monitoring and mitigation systems from being implemented to protect the people operating in hostile environments. Various authors have identified automated data quantification as the most significant factor preventing the widespread use of physiological monitoring systems [4]–[7]. This thesis addresses this challenge through the development of the EXTReme Environment Risk Evaluation and MItigation Framework Score (EXTREMIS) which provides context to the physiological data collected via wearable sensors. The inclusion of acceleration and physiological sensors provides the ability for meaningful interpretation which takes into account the activities being completed and their respective physical demands. Thus, detecting more true incidences of negative health status, and reducing the incidence of false alarms.
The EXTREMIS framework utilises accelerometer data from body worn sensors, and supervised machine learning techniques to classify the activity of the user, and therefore allow for contextual analysis of measured physiological parameters. The approach of automated activity classification of simulated mountaineering specific activities through the use of accelerometers was validated in a controlled laboratory setting. The optimal number and location of accelerometers was established, and the effect on classifier performance of equipment states was quantified. A significant effect of external activity required equipment (12kg rucksack, and mountaineering boots) was noted, and an important conclusion from this research is the need to consider this additional equipment when designing activity classification models for use in extreme environments. Failure to do so will result in poor performance of the activity classifier. This result is potentially significant to fields far beyond mountaineering and extreme environments, and should be considered in any activity classification problem where extra equipment is utilised.
This optimised accelerometer configuration, discovered in the initial investigations, was then tested in a mountain environment where additional challenges such as, terrain surfaces and equipment variations were considered. Finally, simultaneous physiological measurements were collected and the EXTREMIS framework was evaluated against existing clinical tools for detecting negative health states. The National Early Warning Score (NEWS) [8], which is a prominent aggregate scoring system for captured physiological measurements, was chosen as a suitable comparison measure. The findings of the study demonstrate that there was a significant main effect of the type of activity being completed on mean heart rate (p < 0.001), mean respiratory rate (p < 0.001), and mean temperature (p < 0.001). This variation further reinforces the need to dynamically evaluate physiological metrics in relation to the activities being completed. As direct evaluation was not possible, the proposed EXTREMIS framework was evaluated against the clinically validated NEWS framework. When both evaluation frameworks are considered, the mean aggregate scores produced by both frameworks over the sessions were significantly different (p < 0.001), NEWS = 7.17(±1.33), EXTREMIS = 3.27(±1.70). Similarly, both frameworks produced significantly different scores for each activity (p < 0.001). The NEWS framework consistently overestimates the risk due to the omission of physiological loading, and therefore provides operational limiting restrictions on non-negative health events. 81.69% of total trial time was designated a ‘High’ risk state by the NEWS framework, which would indicate the need for immediate intervention despite physiological metrics being in normal ranges for the activities being completed and there being no occurrences of untoward events. EXTREMIS with 5.84% of total trial time was designated a ‘High’ risk state, was shown to more closely represent the lower risk state associated with the evaluated physiological measurands expected due to the metabolic demands of the activities being completed.
The main contribution of this body of research is a novel framework which allows for meaningful evaluation of activities based on their respective physiological demands. This is the first time relative metabolic intensities have been utilised to set expected intensity ratings for recognised activities, and adaptable thresholds defined. This new approach produces more meaningful parameter evaluations, better reflecting the health status of the user, and thus reducing the incidence of false alarms. The EXTREMIS framework provides an immediate solution to a current problem and retains the flexibility to be incrementally optimised and improved with further testing and inclusion of state of the art technologies and practices.
Funding
EPSRC Centre for Doctoral Training in Embedded Intelligence
Engineering and Physical Sciences Research Council
Find out more...History
School
- Mechanical, Electrical and Manufacturing Engineering
Publisher
Loughborough UniversityRights holder
© Stephen WardPublication date
2022Notes
A Doctoral Thesis. Submitted in partial fulfilment of the requirements for the award of the degree of Doctor of Philosophy of Loughborough University.Language
- en
Supervisor(s)
Massimiliano Zecca ; Sijung HuQualification name
- PhD
Qualification level
- Doctoral
This submission includes a signed certificate in addition to the thesis file(s)
- I have submitted a signed certificate