Dynamic Model Selection for Accurate Importance Derivation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for deriving the degree of importance for specific targets using machine learning models often apply inappropriate models, leading to inaccurate results due to model mismatch with the target state.
Innovation Solution
An information processing apparatus and method that selects an appropriate model from a plurality of models based on input data to derive the degree of importance for each item, using a selection model that is chosen based on the output results from either a mortality prediction model or a long-term hospitalization prediction model, depending on the threshold value of the mortality probability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single machine learning model is used to derive the degree of importance for all targets, then the device complexity is reduced, but the measurement precision and reliability of the derived importance degree deteriorate due to model mismatch with different target states
Solution Approach 1:
The patent divides the machine learning model into multiple specialized models (first prediction model and second prediction model) that handle different target states separately. Each model is optimized for specific conditions, ensuring high measurement precision for the degree of importance derivation while maintaining manageable system complexity through structured organization.
Solution Approach 2:
The patent changes the parameter of model selection based on the state of the specific target. By determining the target state and selecting the appropriate model accordingly, the system adapts to different conditions, thereby improving the reliability and precision of the degree of importance derivation without requiring an overly complex universal model.
2Reliability
If multiple specialized models are used to handle different target states, then the measurement precision and reliability of the degree of importance are improved, but the device complexity increases due to model selection requirements
Solution Approach 1:
The patent performs preliminary determination of the target state before selecting the prediction model. This advance preparation allows the system to choose the most appropriate model based on pre-assessed conditions, ensuring high reliability in degree of importance derivation while managing complexity through a systematic pre-selection process rather than ad-hoc model switching.
Solution Approach 2:
The system automatically determines the target state and selects the appropriate model without requiring manual intervention. This self-service mechanism improves reliability by consistently applying the correct model for each target state while avoiding the complexity of manual model selection processes.
3Device complexity
If an inappropriate machine learning model is applied to a specific target, then the device complexity is reduced, but the loss of information increases as the derived degree of importance becomes inaccurate
Solution Approach 1:
The patent implements a dynamic model selection mechanism that adapts to the specific state of each target. By determining the target state and selecting the most appropriate model dynamically, the system prevents information loss that would occur with static or inappropriate model application, while maintaining manageable complexity through automated adaptation.
Solution Approach 2:
The system uses feedback from target state determination to guide model selection. By continuously assessing the target state and selecting models accordingly, the system ensures accurate degree of importance derivation, minimizing information loss while maintaining efficient model management through feedback-driven decisions.
Data Source
AI summary
An information processing apparatus including: plural models including at least a first model that uses first data as input to carry out a first prediction task and a second model that uses second data as input to carry out a second prediction task different from the first prediction task, and at least one processor, in which the processor is configured to: acquire information data representing information related to a specific target; select a selection model to be used for prediction from among the plural models based on the information data; and derive a degree of importance to the specific target for each item included in the information data by using the selection model.