Nearest-neighbor multi-granularity profit method for collaborative knowledge reduction of large-scale electronic health records

• Main Authors: DING, Weiping, FENG, Zhihao, CAO, Jinxin, JU, Hengrong, DING, Shuairong, CHEN, Senbo, REN, Longjie, LI, Ming, WAN, Jie, ZHAO, Lili, SUN, Ying, ZHANG, Yi
• Format: Patent
• Language: English
• Published: 09.09.2021
• Subjects: 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

Disclosed is a nearest-neighbor multi-granularity profit method for collaborative knowledge reduction of large-scale electronic health records, comprising: first dividing large-scale electronic health record datasets into different multi-granularity 5 evolutionary subpopulations on a Spark cloud platform; subsequently, constructing a nearest-neighbor multi-granularity profit model and constructing collaborative nearest-neighbor vectors in nearest-neighbor radii; calculating shared nearest-neighbor profit weights of super elitists and their weight profit vectors, and executing an adaptive dynamic adjustment strategy for a super-elitist weight profit 10 matrix; and finally, calculating a collaborative knowledge reduction set of the large-scale electronic health record data and its core attribute, and storing the knowledge reduction set of the electronic health records in the Spark cloud platform. The present invention can efficiently acquire a knowledge reduction set of incomplete and ambiguous data in the large-scale electronic health records, thus being of great 15 significance and value for decision support and analysis of the large-scale electronic health records.