Abstract
This thesis addresses the need for accessible, fieldable measurement of integrated cardiorespiratory and locomotor function. The central hypothesis was that a low-cost, fieldable system based on existing Movesense sensors, combined with tailored firmware and transparent signal-processing algorithms, can yield metrics comparable with laboratory standards. To test this hypothesis, FreeLab, an open-source, end-to-end pipeline, was developed and tested against laboratory criterion metrics. FreeLab integrates customised Movesense firmware, Android data-capture applications, synchronisation routines and reproducible signal-processing tools to bridge laboratory validity and ecological application. Additional, functional metrics were created that combine cardiorespiratory and locomotor function.
A systematic review identified shortcomings in wearable-derived metrics and guided the selection of primary outcomes: ECG-derived RR′ intervals, IMU-based step timing and locomotor–respiratory coupling. Algorithm development focused on robust peak detection and artefact correction. Laboratory validation (n = 21; 14 running, 7 cycling) demonstrated that a bespoke two-stage R/S peak-detection and artefact-correction algorithm applied to single-lead Movesense ECG (500 Hz) produced RR′, instantaneous heart rate and time-domain HRV measures in strong agreement with a three-lead reference and with superior reliability to a commercial chest strap during high exertion. A complementary treadmill study (n = 14) showed that chest-mounted Movesense IMUs (~208 Hz), processed using optimised peak-to-peak and threshold-crossing methods, estimated step timing, laterality and step-time asymmetry with errors approaching the device’s sampling resolution and high correlation with a 1000 Hz force-plate reference.
The thesis also presents a multimodal breathing-frequency estimator that fuses ECG amplitude, RR′ dynamics and chest acceleration with a heart-rate-informed correction, and introduces the Cadence–Breathing Frequency Ratio (CBFR) to quantify locomotor–respiratory strategies. Field deployments, pairing bilateral foot and chest units with GNSS corrected by a digital-elevation model, produced terrain-aware outputs such as step-level cadence, ground-contact time variants and a Heart-Beats-per-Kilometre (HbpKm) efficiency index. Complementary open-source tools (imu2sto, a Kubios-compatibility module, Movesense Flash converters and the FreeLab GUI) were developed to enhance reproducibility and application.
Key delimitations include modest and uneven sample sizes, treadmill-based laboratory conditions, hardware sampling bounds and the absence of a breath-by-breath outdoor respiratory criterion. The findings indicate that transparent algorithms, customised firmware and open tools can enable low-cost wearables to produce metrics approaching laboratory comparability, supporting their use in sports-science research and applied monitoring.