Real time non-invasive estimation of oxygen uptake using smartphones

Sammanfattning: Oxygen uptake is a great indicator of cardiopulmonary health but requires spe- cialized equipment and is time consuming to measure. There are ways to estimate oxygen uptake based on other factors such as heart rate and acceleration. The aim of this master thesis was to investigate whether models estimating oxygen uptake based on acceleration data collected from the sensors in smartphones, and steps and heart rate data from wearables could be created. To do this ten healthy volunteers performed a cardiopulmonary exercise test (CPET) where their oxygen uptake was measured using a portable system in ad- dition to this acceleration was measured using both an MSR 165 accelerometer and a smartphone, and their heart rate was measured using a Fitbit Edge 4. Dif- ferent ways to process the raw acceleration data were used to create models link- ing acceleration to oxygen uptake. The models were compared to each other to investigate which model best estimated oxygen uptake. The results show that models using the L2 norm of the acceleration, with the effect of gravity removed, performed better than models using the integral of absolute acceleration. Additionally, the models based on data collected using the smartphone outperformed the models based on data from a purpose made accelerometer. A model based on 6-axis motion data collected using the smart- phone performed the best with an R^2 value of 0.98 showing great potential for estimating oxygen uptake using only a smartphone.