NEAREST-NEIGHBOR MULTI-GRANULARITY PROFIT METHOD FOR SYNERGETIC REDUCTION OF KNOWLEDGE OF MASSIVE ELECTRONIC HEALTH RECORDS

A nearest-neighbor multi-granularity profit method for the synergetic reduction of knowledge of massive electronic health records: first, dividing a data set of massive electronic health records into different multi-granularity evolutionary subpopulations on a Spark cloud platform; next, building a nearest neighbor-based multi-granularity profit model, and constructing a coordinated nearest neighbor vector in the nearest neighbor radius; then finding super elite shared nearest neighbor profit weights and a weight profit vector thereof, and implementing an adaptive dynamic adjustment strategy of a super elite weight profit matrix; and finally, finding a data knowledge synergetic reduction set of the massive electronic health records and core attributes thereof, and storing the knowledge reduction set of the electronic health records on the Spark cloud platform. The described method is able to efficiently obtain an incomplete and fuzzy data knowledge reduction set in the massive electronic health records, which