| Abstrakt: |
An electrocardiogram (ECG) is frequently used to assess heart state, which can guide more research into cardiac problems and identifying heart illnesses. The method is easy, quick, and non-intrusive. However, manually analyzing the ECG data takes time, making it difficult to identify and classify arrhythmia manually. The service of healthcare industries has become more effective and of higher quality due to integrating IoT elements into medical equipment. For arrhythmia detection, the proposed research undergoes two stage novel approach, which includes the novel routing model and the detection model. For an efficient data transfer, the routing strategy is crucial. Therefore, this research proposes a QoS based trust aware osprey geographic routing (QOS based TAOGR) by evaluating the trust score of the path (TSP). The multi-objective functions utilized here are total energy cost and delay for an optimal route selection from the source to the destination. The second stage is arrhythmia detection, which undergoes signal pre-processing for efficient detection. The triple phase cascaded Savitzky Golay smoothing (TPCSGSGS) filter is introduced by merging three Savitzky Golay smoothing (SGS) filters to eliminate the noise in the acquired ECG data from the IoT-based smart wearable devices. Afterwards, a binarized convolutional neural network and group decision-making-based ensemble learning (BCNN-GDM-EL) are proposed for detection. In the simulation scenario, the performance is evaluated under two sources: routing and detection performance, such as energy consumption, packet delivery ratio, delay, accuracy, specificity and sensitivity. Also, the tenfold cross validation analysis is performed in terms of accuracy, specificity and sensitivity and compared with existing models. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink