Intelligent matching method for similar medical records

• Main Authors: RUAN SHUHUA, DENG MUQING, DENG FENGYANG, MO BEIXI, ZHENG YIYING, OU JIAMAN
• Format: Patent
• Language: Chinese; English
• Published: 14.10.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; 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 discloses an intelligent matching method for similar medical records. The intelligent matching method comprises the following steps: step 1, extracting key field information in a to-be-analyzed electronic medical record, deleting missing and repeated data in extracted field data, segmenting a key field text into single lexical symbols, finding out a maximum segmentation combination, and obtaining word vector representation of a key field; 2, the key field word vectors serve as input of a deep neural network pre-training model, three semantic models obtained through large-scale text database training serve as feature extractors, and migration feature learning is conducted on the key field word vectors in cooperation with a full-connection network; and 3, according to the feature vector similarity between the verification field and the trained field of the training library, carrying out weighted fusion on the feature matrix obtained after deep migration feature learning, and combining different cl