An artificial intelligence-based clinical drug abnormal use identification method

CN122117218APending Publication Date: 2026-05-29SUZHOU KAIRUIYUAN INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SUZHOU KAIRUIYUAN INFORMATION TECH CO LTD
Filing Date
2026-02-24
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies are ill-suited for identifying abnormal drug use in clinical settings, particularly in real-world scenarios involving multiple drug combinations and dynamic combinations across time windows. Furthermore, they fail to adequately consider differences in departmental, diagnostic group, and liver and kidney function stratification, resulting in both false positives and false negatives, and lacking timely processing of abnormal drug use.

Method used

A drug co-occurrence relationship graph was constructed and egonet was extracted. Pharmaceutical risks were weighted by combining clinical pharmaceutical risk constraint information. Stratified power-law baseline and OddBall deviation sequence were used for change point detection to locate abnormal intervals and output key drug pairs, thereby achieving accurate identification, classification and interpretable verification of abnormal drug use.

Benefits of technology

It improves the accuracy and interpretability of identifying abnormal drug use in clinical practice, reduces false alarms and missed alarms, enhances the temporal stability and verifiability of abnormal drug use, and provides a traceable chain of evidence and priority criteria.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122117218A_ABST
    Figure CN122117218A_ABST
Patent Text Reader

Abstract

The application discloses a kind of clinical drug abnormal use identification method based on artificial intelligence, comprising the following steps: obtaining target patient information, and data preprocessing is generated medication event sequence;According to medication event sequence, and extract one-hop neighborhood local subgraph as egonet;Risk weighted egonet structure features are calculated;According to hierarchical priority rule, medication event is layered, and the power law baseline parameter corresponding to each layer is obtained;Using OddBall graph anomaly detection, the power law deviation degree of each time window is calculated and forms deviation degree sequence;Variable point detection is carried out to deviation degree sequence to determine abnormal interval, and output abnormal use identification result;In abnormal interval, edge removal contribution evaluation is carried out to egonet edge set, and determines key drug pair and is associated with abnormal use identification result and outputs.The application improves the accuracy and reviewability of clinical drug abnormal use early warning.
Need to check novelty before this filing date? Find Prior Art