Author(s): AUSTIN, K., JONES, C.M., LOOSEMORE, M. , Institution: UNIVERSITY OF CHICHESTER, Country: UNITED KINGDOM, Abstract-ID: 730

INTRODUCTION:
Heart rate (HR) monitoring is critical in making informed decisions on training load and recovery prescription [1, 2]. Advancements and miniaturization of sensors have allowed for the monitoring of HR during sports performance, typically at the wrist or chest [3, 4]. There are however multiple sports where these sensors are not permitted, such as combat sports. The inclusion of an LED proximity sensor in some instrumented mouthguards (iMG), typically used for quantifying head impacts in sport, could estimate HR utilising photoplethysmography (PPG). Validation of such sensors is required before performance can be evaluated within combat sports. The aim of this study is to present an initial methodology and evaluate validity for estimating HR through an iMG mounted sensor under controlled conditions.
METHODS:
Study methodology was devised in line with the INTERLIVE network recommendations [5]. Four healthy active completed a single laboratory trial of three conditions: five minutes of seated rest, five minutes of low intensity (1.5 W/Kg) and medium intensity (3 W/Kg) cycling (Wattbike, Nottingham, UK). Participants were instrumented with the PROTECHT iMG (Sport and Wellbeing Analytics, Swansea, UK) and a Polar H10 chest strap (Polar Electro, Kempele, Finland) as a gold standard measure [6]. The iMG contains an LED sensor positioned to sit firmly against the gum [7]. The iMG samples at 100Hz and utilizes red light PPG. A novel processing pipeline was devised to accounts for significant baseline noise present within the raw PPG signal. For both methods, the final three minutes of each condition was analysed, with HR averaged over five second intervals. Agreement was assessed using mean absolute percentage error (MAPE; %), root mean square error (RMSE; beats per minute [bpm]) and Lin’s concordance correlation coefficient (CCC), with cutoffs of 10% and 0.9 generally accepted for MAPE and CCC, respectively.
RESULTS:
Average MAPE was 6.7%, 6.8% and 4.6%, average RMSE was 6.7bpm, 8.1 bpm and 7.5 bpm and average Lin’s CCC was 0.93, 0.92 and 0.73 for sitting, low and medium intensity exercise respectively.
CONCLUSION:
The initial exploration of an iMG mounted HR sensor found that valid measurements of HR can be obtained from the mouth during sitting and low intensity exercise. Although agreement during medium intensity exercise fell below the pre-determined 0.9 threshold for CCC, average MAPE was the lowest of all conditions and RMSE 7.5 bpm. Although further data is required in sports specific scenarios, findings indicate that sufficient physiological signal can be collected from an iMG, which could allow for HR quantification in sports where traditional sensors are not viable.
1) Schneider et al., Front Physiol, 2018; 2) Halson, Sports Med, 2014; 3) Ruiz-Malagón et al., Proc Inst Mech Eng P J Sport Eng Technol, 2021; 4) Biswas et al., IEEE Sens J, 2019; 5) Mühlen et al., Br J Sports Med, 2021; 6) Gilgen-Ammann et al., Eur J Appl Physiol, 2019; 7) Park et al., Front Physiol, 2022