Estudio de la viabilidad de la estimación del grado de adherencia a estilos de vida saludables a partir de medidas cardíacas oportunistas

In recent times, developed countries have experienced an increase of several diseases related to non-healthy lifestyles. Nevertheless, as an individual, it is difficult to know the degree of adherence to a healthy lifestyle (EVS) regardless the personal benefits of following it are multiple. Therefore, the aim of this thesis is to estimate the degree of EVS of an individual from the measurement of cardiovascular variables in a non-intrusive, simple, comfortable and fast way. So, the measurement has to be done with a common device like a smartphone. This thesis has employed the heart rate variability (HRV) and the pulse arrival time (PAT) as cardiovascular variables. Several previous studies have demonstrated that the HRV could be measured with a smartphone but they have not quantified their accuracy. Therefore, this thesis has assessed if it is possible to do an accurate HRV measurement with these devices. It has found that some smartphones are not suitable due to their flash and the quantification error of the HRV is mainly from the pulse wave and not from the smartphone. Moreover, the optimal cut off frequencies to filter the pulse wave depend on the method used to detect its arrival. Also, the estimation error in the HRV extracted from a pulse wave depends on the measurement conditions. These could happen if the measurement is done in non-controlled environments as an inexperienced person measuring HRV with a smartphone. Some of these influences have been thoroughly studied. This thesis has evaluated the breathing, the postural and the contact force between photoplethysmograph sensor and the finger effects showing that all of them should be controlled. Additionally, in order to reduce the measurements conditions effects, the pulse wave arrival should be detected at the maximum of its first or second derivative or with the parametric estimator TANH. Moreover, the respiratory sinus arrhythmia is systematically bigger in the HRV extracted from the pulse wave than from the electrocardiogram, the human body get used to the contact force and the pulse wave morphology changes with it and, likely, with the posture. The PAT estimation is also affected by external measurements conditions but, usually, their effects have not been studied. This thesis recommends controlling the breathing frequency, the posture and the contact force in the PAT measurement and detect the pulse wave arrival in the maximum of its first derivative or, in case of contact force not being controlled, with the parametric estimator TANH. Moreover, the breathing frequency and the posture affect the PAT potency but only the first and the contact force influence the PAT mean value. On the other hand, short good measurements segments could exist within long and not-controlled measurement of HRV and/or PAT. However, the effect of a very short cardiovascular measurement on the HRV and the PAT quantification is little analysed. Therefore, they are studied in this thesis. It has been found that four HRV indexes and two PAT indexes can be employed in measurements between 10 and 90 seconds long. In addition, averaging several of them is a simple method to increase their repeatability and similarity. Consequently, up to ten HRV indexes can be used if the indexes computed from several measurements of very short duration are averaged. Finally, this thesis has studied if a separated or simultaneously measurement of the HRV and the PAT can be used to quantify the degree of EVS of an individual. It has been concluded that the relation between them is likely insufficient to accurately predict most of the EVS indicators that quantify the degree of EVS. In addition, it has been found a moderate and complex relationship between these indicators and the HRV and small and complex with the PAT. Moreover, there is no interaction between the HRV and the PAT that reflects a significant bigger relationship with most of the EVS indicators.