Enhancing Energy Efficiency and Battery Lifetime in Cardiac Implantable Devices using Optimized RNN-LSTM

Abstract

Cardiac implanted electronic devices (CIEDs) help in elderly people who suffer from critical heart disorders. Patients who suffer from this condition are put at risk by several surgical treatments required to replace the batteries in CIEDs. To extract the ideal feature extraction and classification of Electrocardiogram (ECG) signals from CIED records, the proposed methodology uses a Recurrent Neural Network (RNN) along with a Long Short-Term Memory (LSTM) classifier. This research work aims to solve this issue by providing a novel technique intended to improve the battery life, longevity, and energy efficiency of CIEDs. Moreover, the noise in the recorded ECG data is considerably reduced by the Empirical Mode Decomposition (EMD). A novel integrated optimization approach that combines the Nesterov Accelerated Gradient (NAG) algorithm and Particle Harmonic Search (PHS) is deployed to improve constraints of the RNN-LSTM model. The hybrid PHS-NAG algorithm combines the generic outcomes of PHS's global search capabilities with NAG's local optimization advantages. MATLAB is used for the assessment of the formulated method's performance in improving CIED energy efficiency and extending battery life. The comparative outcomes clearly indicate that the RNN-LSTM model gets 94.5% accuracy and the PHS-NAG technique achieves 98.3% efficiency, which proves the effectiveness of the proposed methodology.