Calibration and validation of homeostasis parameters estimates produced by a DSP embedded in a wheelchair

Ballistocardiogram, capacitive electrocardiogram, and photoplethysmogram (PPG) sensors were embedded in a wheelchair together with a digital signal processor (DSP) with Wi-Fi capabilities. This setup allows monitoring of some homeostasis parameters of the wheelchair user, namely heart rate (HR), pulse arrival and transit times, and oxygen saturation (SpO₂). Ten subjects tested the wheelchair capabilities. SpO₂ and HR obtained from the PPG of reference equipment were also recorded for comparative analysis. This paper reports the results of the algorithms developed for HR estimation, for all the signals, and for SpO₂ from the PPG. From the data of the ten subjects the algorithms had their parameters calibrated, and presented an average RMS error of 1.903% for SpO₂, with the best value being 0.024%. The most unstable signal is the BCG from where the global calibration provides an HR estimate with an average 4.73 bpm error. The algorithms created were confirmed valid, and the signals from the hardware setup can be used to provide estimates with significant accuracy in heart rate estimation, but needing personalized calibration in the SpO₂ case.