Disease risk prediction method based on recurrent neutral networks

摘要: The invention discloses a disease risk prediction method based on recurrent neutral networks. includes the steps that 1, diagnosed is used as training sample, disease-name distributed word-vector conducted, and mapping matrix obtained saved; 2, sample again, network module obtained; 3, each in historic records of patient serves test input to module, results are obtained.The networks expression embedding technology adopted, solves problems too complicated, cost high accuracy low, wherein caused by dimensionality, sparse data, temporality other characteristics medical diagnostic data. achieves modeling process with conducted information.