Deep Learning Associated with Linear Regression for Heart beat Rate (HBR) and Respiration Rate(RR) Measurement

The Heart beat Rate (HBR) and the Respiration Rate (RR) are considered to be a dynamic signs and indications of the patient“s health condition. Most probably it is neglected in normal treatment because of the deviation in the modest sensors considered for convenient measurement. The modulations identified in the respiratory tract by Photo Plethysmo Gram (PPG) and electrocardiogram (ECG) are much helpful in deriving the respiration rateand heartbeat rate by the use of different rules and markings followed in signal processing and voice processing techniques. In this work the HBR and RR is analysed by taking the speech signal as input. The Relationship found in between the human voice, respiration rate and heartbeat rate are discussed in this work. In this work a modified deep-learning algorithm is proposed which uses Hybrid Fuzzy based Convolutional neural networks (HFCNN) for acoustic voice recognition and a Hybrid Deep Support vector machine (HDSVM) for classification along with regression.After experimentation, the percentage of error margin is identified to be better while estimating the various patterns of breathing signs from the audio information. About 92.25% of sensitivity is recorded with 4.22% of error while estimating the RR and HBR. Then for estimating the HBR, the HFCNN-based regression model provides better prediction values with 0.2873 as root mean-squareerror and gives a better classification rate after implementing the HDSVM. The root mean square error is 0.3093 which is better when compared to the other methods. The accuracy of prediction in association with the precision and recall values are identified to be better when compared to the other existing methods.