| Cardiorespiratory fitness,as the fifth major clinical vital sign,can reflect the health status of major body functions.The level of cardiorespiratory fitness is significantly related to the incidence of chronic diseases such as cardiovascular disease.However,accurate cardiorespiratory fitness evaluation methods require expensive equipment and high exercise load,which is difficult to popularize in the field of public health.The level of cardiorespiratory fitness is closely related to the function of cardiac autonomic nervous system.ECG signals can reflect the autonomic function of the heart,thereby reflecting the level of cardiorespiratory fitness.At present,there is a lack of research on cardiorespiratory endurance fitness by ECG signal.To solve these problems,this thesis conducts a related study on the evaluation of cardiorespiratory fitness through the ECG signal.A cardiorespiratory fitness evaluation system based on ECG signals is designed,which realizes an accurate evaluation of the level cardiorespiratory fitness through low exercise load.The main contents are as follows:(1)Design of experiment based on ECG signals and construction of ECG dataset for cardiorespiratory fitness evaluation.The experiment consists of two exercise tests: a squatting test and a cycle ergometer test for estimation of maximum oxygen uptake.The ECG signals of the experimental subjects were collected during the squatting exercise test with a total duration of 6.5 minutes.As the label of the corresponding ECG sample,Cardiorespiratory fitness level was achieved through the cycle ergometer test.(2)Extraction and analysis of ECG features.The wavelet transform method was used to denoise ECG signals in the dataset,and the R wave was identified by the Pan & Tompkins differential threshold algorithm to obtain the RR interval sequence.Heart rate features and heart rate variability features were extracted by time domain,frequency domain and nonlinear analysis methods.ECG features under different levels of cardiorespiratory fitness were analyzed and discussed.(3)Design of cardiorespiratory fitness evaluation models.Two types of machine learning models were built: supervised learning model and semi-supervised learning model.BP neural network and SVM were used for the supervised model.These two classifiers were combined with the SBS algorithm to select the best feature set which is the most important for evaluation of cardiorespiratory fitness.These two classifiers were trained and tested by the labeled samples of the ECG dataset.The semi-supervised learning model was built by the combination of IForest algorithm and XGBoost algorithm.All the data in the ECG dataset were used for training and testing,and the significance of features of the model was analyzed.Finally,through the analysis of the results of the three models,it is concluded that the XGBoost model has the best performance.(4)Design and implement of a cardiorespiratory fitness evaluation system based on ECG signals.The design of system includes the software and hardware design of the ECG detection platform and the software design of the cardiorespiratory fitness evaluation platform.The ECG detection platform includes the ECG signal acquisition module,the bluetooth module,the data storage module and the controller module.The ECG detection platform is designed by embedded software to realize the acquisition,analog-to-digital conversion,data storage and transmission of the ECG signal during the squatting exercise test.the cardiorespiratory fitness evaluation platform includes the cardiorespiratory fitness evaluation module,the humancomputer interaction module,the data processing module,and the data receiving module.the ECG processing algorithm,feature extraction algorithm and the XGBoost cardiorespiratory fitness evaluation model studied in this paper are integrated into the platform.The platform realizes the reception,processing and feature extraction of ECG signals,and provides a visual and operable human-computer interaction for cardiorespiratory fitness evaluation. |
