Medical Information Processing Apparatus Learning Validation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

There is a lack of methods to validate whether the performance requirements for clinical use are met when machine learning-based changes occur in medical information processing systems, and users lack an easy way to determine the validity of these changes.

Innovation Solution

A medical information processing apparatus is configured with an obtaining unit, a learning unit, an evaluation data holding unit, an evaluating unit, and an accepting unit to assess and apply learning results, allowing users to evaluate the validity of machine learning outcomes and determine whether to update system parameters based on evaluation data and user instructions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If machine learning is used to improve image analysis accuracy, then the recognition accuracy is improved, but the user cannot easily determine whether the performance requirements for clinical use are met

Engineering Contradiction:
Improveimage analysis accuracyVSAvoidease of validity determination
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The patent implements a feedback mechanism where the evaluation unit provides feedback information to the user about the learning result's validity. The evaluation unit compares the learning result against evaluation data and provides feedback indicating whether the performance requirements are met, enabling users to easily determine validity without manual analysis of complex model outputs.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The evaluation unit acts as an intermediary between the learning unit and the user. It translates complex machine learning results into easily interpretable validity information by comparing learned patterns against stored evaluation data, thereby mediating the information flow and making validity assessment accessible to users without machine learning expertise.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If machine learning is used to change system performance, then the function improvement is achieved, but there is no configuration for users to easily perform determination on validity

Engineering Contradiction:
Improvefunction improvementVSAvoidvalidity examination configuration
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-storing evaluation data in the evaluation data holding unit before the actual learning process. This evaluation data serves as a reference standard that is ready to be compared against learning results, enabling rapid validity determination without requiring complex real-time analysis configurations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The evaluation unit performs self-service by automatically comparing learning results against stored evaluation data and generating validity determinations without requiring user intervention in the complex comparison process. The system serves itself by providing ready-to-use validity assessment functionality.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If learning results are applied to improve medical information processing, then the processing accuracy is improved, but the reliability of clinical application cannot be ensured without validity examination

Engineering Contradiction:
Improveprocessing accuracyVSAvoidclinical application reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent implements beforehand cushioning by conducting validity examinations before applying learning results to clinical workflows. The evaluation unit assesses whether learning results meet predetermined performance requirements before deployment, providing a protective check that prevents unreliable models from entering clinical practice.

Inventive Principle:
Principle #11Beforehand cushioning (Prior cushioning)

Data Source

PatentUS11914918B2Medical information processing apparatus, medical information processing method, and non-transitory computer-readable storage medium
Publication Date: 2024.02.27 CANON KK
  • US11914918B2 patent drawing
  • US11914918B2 patent drawing
  • US11914918B2 patent drawing

AI summary

A medical information processing apparatus comprises an obtaining unit that obtains medical information, a learning unit that performs learning on a function of the medical information processing apparatus using the medical information, an evaluation data holding unit that holds evaluation data in which a correct answer to be obtained by executing the function is known, the evaluation data being for evaluating a learning result of the learning unit, an evaluating unit that evaluates a learning result obtained through learning, based on the evaluation data, and an accepting unit that accepts an instruction to apply a learning result of the learning unit to the function.