Medical Information Processing Apparatus Validity Examination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a lack of validity examination in medical information processing apparatuses to determine whether machine learning performance meets clinical requirements and maintain data reliability, especially when performance changes occur.
Innovation Solution
A medical information processing apparatus is designed with an obtaining unit, a learning unit, an evaluation data holding unit, and an evaluating unit to perform machine learning, evaluate learning results using known correct answers, and update parameters based on evaluation results, ensuring that the performance meets intended clinical requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning is performed to improve image analysis accuracy, then the image analysis accuracy is improved, but the reliability of the learning result cannot be examined
Solution Approach 1:
The patent applies preliminary action by preparing evaluation data with known correct answers before performing machine learning. This evaluation data is stored in advance and used to verify the learning results, allowing the system to proactively ensure reliability rather than discovering issues after deployment.
Solution Approach 2:
The patent implements feedback by using the evaluating unit to compare machine learning results against pre-stored evaluation data with known correct answers. This creates a closed-loop verification system where the reliability of learning results is continuously assessed and fed back to determine whether the learned parameters should be updated.
2Measurement precision
If machine learning parameters are updated to improve performance, then the image analysis accuracy is improved, but it cannot be determined whether the performance meets clinical requirements
Solution Approach 1:
The patent prepares evaluation data with known correct answers in advance, storing it in the evaluation data storage unit. This preliminary preparation enables the system to assess whether learned parameters meet clinical requirements before actual clinical use, ensuring compliance is verified proactively.
Solution Approach 2:
The evaluating unit provides feedback by comparing learning results against pre-stored evaluation data and determining whether clinical requirements are met. This feedback mechanism controls the parameter update process, ensuring that only learning results meeting clinical standards are deployed.
3Device complexity
If evaluation data is not maintained properly, then the system is simpler, but the validity examination of machine learning cannot be performed
Solution Approach 1:
The patent segments the data management function into a dedicated evaluation data storage unit, separate from the main learning data storage. This segmentation isolates the critical evaluation data, ensuring its integrity and reliability while maintaining a clear distinction between training data and verification data.
Solution Approach 2:
The evaluating unit serves multiple functions: it reads evaluation data, performs validity examination of learning results, determines whether clinical requirements are met, and controls parameter updates. This multi-functionality achieves comprehensive validation without adding excessive system complexity.
Data Source
AI summary
A medical information processing apparatus comprises: an obtaining unit that obtains medical information; a learning unit that performs machine learning on a function of the medical information processing apparatus using the medical information; an evaluation data holding unit that holds evaluation data for evaluating a learning result of the learning unit; and an evaluating unit that evaluates a learning result obtained through machine learning, based on the evaluation data.