Mattress Sensor-Based Respiration Rate Estimation Using Unsupervised Clustering

In this paper, a method is proposed to measure the human respiration rate using mattress sensor. An unsupervised clustering method, combined with cluster selection algorithm, was used rate. The volume change of upper body during breathing process indicated by pressure on sensor and waveform estimated measuring change. measured changes were separated into respiratory-related clusters noise method. Finally, algorithm find combination that best represented respiration. compared other methods. results showed had mean Pearson correlation coefficient 0.8746, which highest among all difference between these methods statistically significant (p<0.001). addition, regression analysis performed accuracy calculated respiratory most accurately.