A smart medical data interaction retrieval method and system based on a large language model and a medium

By mapping medical data into a dynamic waveform field and simulating the physical interference process, the problem that existing medical data retrieval methods cannot actively identify logical contradictions is solved, achieving efficient and reliable data retrieval and personalized interaction, thus improving user experience and information acquisition efficiency.

CN122432214APending Publication Date: 2026-07-21ANHUI ZHONGJI NAT MEDICAL MEDICAL TECH CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
ANHUI ZHONGJI NAT MEDICAL MEDICAL TECH CO LTD
Filing Date
2026-04-16
Publication Date
2026-07-21

Smart Images

  • Figure CN122432214A_ABST
    Figure CN122432214A_ABST
Patent Text Reader

Abstract

The application discloses a kind of wisdom medical data interaction retrieval methods, system and medium based on large language model, it is related to artificial intelligence medical technical field, method includes: inputting time sequence medical data into large language model and carrying out feature mapping, constructs dynamic data representation field, and data unit is encoded as frequency, amplitude and phase waveform element;User query is converted into semantic interference probe injection field;Monitoring interference state, extracting constructive interference element as preliminary recall path, positioning destructive interference area and marking as logic contradictory energy state collapse anomaly point;Drive large model to deduce injected compensation energy, so that data collapses into logically self-consistent coherent fact matrix;Based on the above path and matrix output interaction retrieval view, the present application actively resolves data deep contradictions through waveform interference, improves retrieval reliability and decision efficiency.
Need to check novelty before this filing date? Find Prior Art