Biological Data Selection for Agitation Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately detect an agitated state in patients, leading to suboptimal quality of biological data for learning models, which in turn affects the accuracy of agitation detection.
Innovation Solution
An information processing device that acquires biological data and determination results from a patient, and selects learning data based on a skill value representing the determiner's ability to assess agitation, thereby improving the quality and accuracy of data for model generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If biological data is collected from patients without skill-based filtering, then data collection is simple and comprehensive, but the quality and accuracy of learning data is insufficient
Solution Approach 1:
The system performs preliminary evaluation of determiners' skill levels before using their data for model training. By pre-assessing and filtering determiners based on their agitation detection capability, the system ensures that only high-quality data from skilled determiners is used, thereby improving measurement precision without requiring complex real-time filtering mechanisms during data collection
Solution Approach 2:
The system incorporates feedback mechanisms where determiners receive evaluations on their skill levels, and this feedback is used to filter and select learning data. The feedback loop allows continuous improvement of data quality by using performance metrics to guide which data is selected for model training, resolving the contradiction between data quality and system complexity
2Measurement precision
If all biological data is used for learning without skill-based filtering, then data quantity is maximized, but the quality of learning data deteriorates
Solution Approach 1:
The system applies local quality filtering by evaluating and selecting data based on the specific skill level of each determiner. Instead of uniformly treating all data equally, the system differentiates data quality based on the determiner's expertise, ensuring that high-quality data from skilled determiners is prioritized while still incorporating data from less skilled determiners to maintain sufficient data quantity
Solution Approach 2:
The system changes the parameter of data selection criteria from simple quantity-based inclusion to skill-based filtering. By introducing skill level as a selection parameter, the system can quality-filter data while maintaining adequate quantity through stratified sampling or weighted inclusion of data from determiners with different skill levels
3Reliability
If determination results from all determiners are used, then comprehensive coverage is achieved, but the reliability of agitation detection decreases
Solution Approach 1:
The system performs preliminary skill assessment of determiners before incorporating their determination results into the learning model. This pre-filtering action ensures that only determiners meeting minimum reliability criteria are included, thereby maintaining high reliability while still allowing multiple determiners to contribute to the overall versatility of the system
Solution Approach 2:
The system segments the data sources by determiner skill levels, creating distinct groups of determiners based on their reliability metrics. This segmentation allows the system to selectively include data from high-reliability determiners while maintaining the ability to incorporate data from diverse sources, thus balancing reliability with versatility through structured data organization
Data Source
AI summary
An information processing device 100 of the present disclosure includes a data acquisition unit 121 and a data selection unit 122. The data acquisition unit 121 acquires biological data measured from a person whose state related to agitation is to be determined, and a determination result of the state by a determiner with respect to the person. The data selection unit 122 selects learning data from the biological data on the basis of a skill value and the determination result. The skill value represents the ability of determining the state and is set for the determiner. Selection of the learning data is optimized by the data selection unit 122 of the present disclosure.