A novel method to determine dynamic temperature trends applied to in-shoe temperature data during walking

Abstract

Body temperature is one of the fundamental measures considered in the assessment of health and well-being, with various medical conditions known to give rise to abnormal changes in temperature. In particular, abnormal variations in dynamic temperature patterns during walking or exercise may be linked to a range of foot problems, which are of particular concern in diabetic patients. A number of studies have investigated normative temperature patterns of a population by considering data from multiple participants and averaging results after an acclimatisation interval. In this work we demonstrate that the temperature patterns obtained using such an approach may not be truly representative of temperature changes in a population, and the averaging process adopted may yield skewed results. An alternative approach to determine generic reference temperature patterns based on a minimization of root mean square differences between time-shifted versions of temperature data collected from multiple participants is proposed. The results obtained indicate that this approach can yield a general trend that is more representative of actual temperature changes across a population than conventional averaging methods. The method we propose is also shown to better capture and link the effects of factors that influence dynamic temperature trends, which could in turn lead to a better understanding of underlying physiological phenomena.