A Security Threat Identification Method for Public Health Management Network Data Based on Deep Learning

Jiapeng Li^1*Jiacheng Liao²Youyou Chen³

• ¹Guangdong University of Technology, Guangzhou, China
• ²Hubei University of Automotive Technology, Shiyan, China
• ³Third Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, China

With the rapid growth of public health network data and its importance in management decisionmaking, ensuring the security of these network data has become a problem that needs to be addressed. Given that security threats such as data leakage may have a profound impact on the public health management network, this paper proposes CNN-Attention-BiLSTM-CRF based on convolutional neural network (CNN) and attention mechanism. It uses deep learning technology to identify security threat for public health management network data. CNN-Attention-BiLSTM-CRF first converts characters in public health management network data sequences into fixed dimensional vector representations, then CNN uses multiple convolutional kernels to capture local features in the network data, forming a comprehensive feature set. To enhance feature capture capability, attention mechanism is introduced and combined with bidirectional long short-term memory network (BiLSTM) to encode sequence data, effectively capturing long-term dependencies. Finally, the conditional random field (CRF) determines the optimal label sequence based on the conditional probability distribution and applies the Viterbi algorithm to complete the sequence labeling of public health management network data. Through extensive experimental verification, CNN-Attention-BiLSTM-CRF can comprehensively understand the semantics of network data, accurately identify security threat in public health management network data.