Theme segmentation method based on Chinese electronic medical record

• Main Authors: ZHANG XINGWANG, DU JINLIAN
• Format: Patent
• Language: Chinese; English
• Published: 14.05.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS

The invention provides a theme segmentation method based on a Chinese electronic medical record. Pre-defining the subject category of the Chinese electronic medical record, and constructing a subject segmentation data set of the Chinese electronic medical record according to the subject category of the sentence; the method comprises the following steps of: preprocessing by utilizing a Jiaba word segmentation model and a Word2Vec pre-training model; a sentence encoder is constructed by using a bidirectional long-short memory network and an attention mechanism, and feature encoding is performed on each sentence in a text. A bidirectional long-short memory network and a convolutional neural network are utilized to capture context features and local features, and the two features are dynamically fused through a gating mechanism to serve as global semantic features of a sentence sequence. A main task (theme segmentation) and an auxiliary task (theme classification) based on the network are constructed by using a j