This paper proposes a novel hybrid framework for ECG signal classification and privacy preservation. The framework includes two phases: the first phase uses LSTM+CNN with attention gate for ECG classification, while the second phase utilizes adaptive least signal bit with neutrosophic for hiding important data during transmission. The proposed framework converts data into three sets of degrees (true, false, and intermediate) using neutrosophic and passes them to an embedding layer. In the sender part, the framework hides important data in ECG signal as true and false degrees, using the intermediate set as a shared dynamic key between sender and receiver. The receiver can reconstruct the important data using the shared dynamic key or the intermediate set. The proposed framework is more robust against attacks compared to other methods.
Digital Object Identifier (DOI)
Rezk, Abdallah; S. Sakr, Ahmed; and M. Abdulkader, H.
"Neutrosophic Adaptive LSB and Deep Learning Hybrid Framework for ECG Signal Classification,"
Applied Mathematics & Information Sciences: Vol. 17:
5, Article 9.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol17/iss5/9