Pattern Discovery Visual Analytics for Clinical Cohort Identification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing medical information systems are difficult to navigate for identifying appropriate patient cohorts, especially when selecting patients for clinical trials, as they are primarily designed for individual patient records and lack efficient tools for analyzing large datasets or incomplete information.

Innovation Solution

A pattern discovery visual analytics system that generates a patient data table, applies predictive patterns to attribute values, and displays statistics in a confusion matrix, allowing for the identification of patient cohorts by grouping patients based on attribute values and clinical stages, facilitating the selection of meaningful patient cohorts for clinical research and trials.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If medical information systems are designed for individual patient record retrieval, then ease of operation for individual patients is improved, but the ability to identify patient cohorts efficiently deteriorates

Engineering Contradiction:

Improveease of operation for individual patient retrievalVSAvoidefficiency of patient cohort identification

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The system segments the patient identification process into two distinct modes: individual patient retrieval (maintaining existing ease of use) and cohort-based pattern discovery (enabling efficient group identification). The pattern discovery module operates separately from the individual record system, allowing both functions to coexist without interference.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a pattern discovery visual analytics system as an intermediary layer between the existing medical information system and the cohort identification need. This intermediary translates individual patient records into discoverable patterns and cohorts, bridging the gap between individual-focused data storage and group-focused analysis requirements.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Quantity of substance

If medical information systems store diverse patient information in multiple formats, then comprehensiveness of patient data is improved, but difficulty of navigating and analyzing the data deteriorates

Engineering Contradiction:

Improvecomprehensiveness of patient dataVSAvoidcomplexity of data navigation and analysis

Core Design Contradiction:

Quantity of substanceVSDevice complexity

Solution Approach 1:

The system transforms diverse patient information into standardized attribute-value pairs, changing the parameter representation from unstructured diverse formats to a uniform tabular structure. This parameter transformation enables pattern matching and cohort identification while preserving all original data comprehensiveness.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces manual navigation through haphazardly organized patient information with an automated pattern discovery system that uses visual analytics and computational algorithms to identify cohorts, substituting mechanical manual searching with automated intelligent analysis.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If patient cohort selection requires complete medical information, then accuracy of cohort selection is improved, but the ability to select cohorts at early treatment stages deteriorates

Engineering Contradiction:

Improveaccuracy of cohort selectionVSAvoidtime delay in cohort identification

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary pattern discovery and cohort identification using available early-stage patient information without requiring complete medical records. By discovering patterns from partial data early in the treatment process, the system enables timely cohort selection while maintaining reasonable accuracy through iterative refinement as more data becomes available.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11087860B2Pattern discovery visual analytics system to analyze characteristics of clinical data and generate patient cohorts

Publication Date: 2021.08.10 KONINKLIJKE PHILIPS NV

AI summary

In pattern discovery visual analytics, a patient data table (14) is generated that tabulates, for each patient, attribute values for a set of attributes. A positive or negative prediction is generated for each patient for a target value of a target attribute using a prediction pattern (20) of attribute values for w attributes (22). The prediction is positive if at least a threshold fraction (26) of the w attributes of the patient match the prediction pattern, is negative otherwise. Patients are grouped into a selected proportion of a confusion matrix (30) in accord with the positive or negative predictions and actual values of the target attribute T in the patient data table. A display component (4) displays a representation (42) of patient statistics for the selected proportion of the confusion matrix on a per-attribute basis for attributes of the w attributes. A patient cohort (44) is identified using the representation.