Design Of Arrhythmia Monitoring System Based On Android Phone

In order to providing chronic cardiac patients and in-home elderly with long-term ECG (electrocardiogram) monitoring services, and enriching functions of wearable ECG medical App, this thesis works on a sort of Android App matching the portable ECG detection device. The App contains functions as:dynamic display and storage of ECG data; analyzing data intelligently and detecting the abnormal beats automatically, which can assist the doctors and provide with decision support.Intelligent analysis of ECG data consists of 3 steps:denoising, eigenvalue extraction, and identification of anomalous ECG. After the comprehensive references of existing algorithms, this thesis executes 4-scale wavelet decomposition on ECG data using biorthogonal & quadratic B-spline wavelet, which is suitable for Android system and has modest anti-interference performance and time complexity. Combining with the decomposition result, this thesis uses soft & hard thresholding to filter out muscle electricity and power line interference, and locates P-QRS-T waves via extracting the couples of extremums of wavelet coefficients modulus. This thesis adopts the 48 examples of ECG data in Arrhythmia Database of MIT-BIH as the training sample.7 types of common beats, such as sinus beat, premature beat (ventricular, supraventricular), escape beat (ventricular, supraventricular), etc, have been conducted pathologic analysis and concluded ECG characteristics. Eventually, diagnostic tree of branching logic for arrhythmia has been established to classify the beats and the abnormal rhythms automatically.Based on the XiaoMi 2s Android phone, this thesis develops a ECG monitor App matching the ECG acquisition device. This software uses Bluetooth Low Energy technology to accomplish ECG data transmission, and timely displays the ECG plot, and intelligently analyze it in Android environment. Besides, all the diagnostic records have been saved in SQLite database, and any abnormal plot can be replayed in need. Test results indicate that all the software functions run well, the effect of arrhythmia classification meets medical needs, and the real-time performance of this software is efficient enough for practical demands.