The invention discloses a population elite distribution cloud collaboration equilibrium method used for feature extraction of an electronic medical record. The method includes the steps that the large-scale electronic medical record is divided into different evolution sub populations under a distributed cloud computing MapReduce framework, and optimal solutions Psolui of all the sub populations are extracted; an N population elite equilibrium dominant strategy in a bounded rationality region is designed, a distributed type collaborative dynamic elite evolution region omega is constructed, and therefore all sub population elites taking part in feature extraction of the electronic medical record can arrive at a dominant region Z under Nash equilibrium under respective average weight probabilities; a concurrent operation mechanism is adopted in all the sub population elites to collaboratively extract feature subsets Fi in respective electronic endemic region dominant subregions, and therefore an overall equilibrium feature set of the electronic medical record is obtained stably. The multi-dimensional complex inline relation among data attributes of the electronic medical record can be handled well, and the efficiency of the concurrent feature extraction of the large-scale electronic medical record is improved.