Medical record intelligent mining system for diagnosis and treatment of nephropathy based on natural language processing

Abstract

The application discloses a natural language processing-based intelligent medical record mining system for kidney disease diagnosis and treatment, comprising a term preprocessing module, a structured extraction module, a diagnosis and treatment relationship mining module and an auxiliary decision-making module. The term preprocessing module performs specialty entity recognition on medical record texts through hierarchical vocabulary and conditional random field decoding; the structured extraction module converts heterogeneous medical records into standardized records through field recognition and relationship extraction; the diagnosis and treatment relationship mining module performs time sequence modeling and hierarchical association mining on multiple medical record records, and outputs a diagnosis and treatment knowledge vector set carrying a confidence weight; and the auxiliary decision-making module generates an auxiliary decision-making result with traceable basis through two-way parallel retrieval and evidence chain assembly.

Images