Methods, computer programs, and devices in data processing systems (scalable visual analysis pipelines for large datasets)

The scalable graph-based visual analysis pipeline addresses the inefficiencies in analyzing EHRs by generating time-series recognition graphs, enhancing pattern discovery and visualizations, thus improving the understanding of clinical pathways and reducing computation time.

JP7868935B2Active Publication Date: 2026-06-02INTERNATIONAL BUSINESS MACHINE CORPORATION
View PDF 5 Cites 0 Cited by

Patent Information

Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
INTERNATIONAL BUSINESS MACHINE CORPORATION
Filing Date
2022-09-01
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing healthcare data processing systems struggle to efficiently analyze large datasets from electronic health records (EHRs) due to the sheer number of medical features and lack of time-series awareness, leading to spurious correlations and limited visualizations that fail to capture clinical pathways and causal relationships.

Method used

A scalable graph-based visual analysis pipeline that generates time-series recognition graph data structures from EHRs, enabling efficient pattern discovery and interactive visualizations for domain experts, using a time-series recognition graph query engine and pattern discovery mechanism.

Benefits of technology

Enables efficient exploration of patient care trajectories, facilitating interactive pattern discovery and improved understanding of clinical pathways, reducing computation time by up to 100-fold compared to traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007868935000002
    Figure 0007868935000002
  • Figure 0007868935000003
    Figure 0007868935000003
  • Figure 0007868935000004
    Figure 0007868935000004
Patent Text Reader

Abstract

To provide a method, device, and computer program for implementing a visual analysis pipeline.SOLUTION: A mechanism generates, from an input database of records, a time series recognition graphical data structure comprising a plurality of record base features designated in an ontology data structure. The time series recognition graphical data structure includes: vertices indicating events or one or a plurality of record base features corresponding to the events; and an edge representing a time series relation between the events. The mechanism executes a time series recognition graphical query to the time series recognition graphical data structure, generates a filtered set of vertices corresponding to a standard of the time series recognition graphical query and corresponding features, executes a pattern finding action to the filtered set of the vertices and the corresponding features, and generates a visual analysis graphical representation for a subset of an identified vertex and a corresponding feature.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art