Abstract
For e-health monitoring, concurrent data acquisition and processing of electrocardiogram and respiratory rate biosignals is needed to provide clean input signals for the detection algorithms. Given that cardiac and respiration rhythms are correlated, but they have different characteristic rates, we explored a sensor fusion approach to model the filtered future state of both signals. Therefore, a time-delayed structure with raw and filtered data fusion was used to improve the quality of the model. We apply Lasso and Ridge regression to the time-delayed data structure and study the approach using low and high sample rates for the data processing. At low sample rates, the validation for respiratory rate is high, while for electrocardiogram is comparatively low. With higher sample rates, the results are improved for both cases.