Decision Tree Data Analysis with Split-Set Effect Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional medical data analysis systems face challenges in predicting treatment effects due to overfitting and lack of interpretability in decision trees, particularly when dealing with individual patient variability.
Innovation Solution
A computer system that partitions data into subsets using a random forest of decision trees, evaluates intervention effects, and calculates scores for each branch to enhance prediction accuracy and interpretability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a decision tree is generated by partitioning a data set into two data sets with different applications to prevent overfitting, then prediction reliability is improved, but the structure becomes different with each training session leading to lack of consistency
Solution Approach 1:
The patent applies segmentation by dividing the data set into multiple subsets (first data set and second data set) for different purposes: one for searching branching conditions and another for evaluating intervention effects. This segmentation allows the system to maintain structure consistency while improving prediction reliability through rigorous validation.
Solution Approach 2:
The patent implements feedback by evaluating intervention effects using the second data set after generating branching conditions with the first data set. This feedback mechanism ensures that the decision tree structure remains consistent across training sessions while maintaining high prediction reliability through iterative validation.
2Measurement precision
If a random forest using a plurality of decision trees is used to predict treatment effect, then prediction accuracy is improved, but interpretability of prediction result is not ensured
Solution Approach 1:
The patent extracts and presents only the most important branching conditions from the random forest model. By selecting and displaying key branches that contribute most to prediction accuracy, the system maintains high prediction accuracy while restoring interpretability by showing users the critical decision paths.
Solution Approach 2:
The patent uses visual presentation techniques to highlight important branches, making the complex random forest model interpretable. By visually distinguishing key branching conditions from less important ones, the system maintains prediction accuracy while enabling users to understand the model's decision-making process.
Data Source
AI summary
A computer system is accessibly connected to a database that stores data including values of a plurality of factors. The computer system repeatedly executes: first processing of partitioning an analysis data set including a plurality of pieces of data into a first data set and a second data set; second processing of searching, using the first data set, for a branching condition for partitioning the analysis data set into two groups, evaluating an intervention effect using the second data set, determining the branching condition to be used, and generating a decision tree that includes at least one branching condition and is used to predict an event; and third processing of calculating a score indicating quality of a branch of the decision tree for each of a plurality of decision trees. The computer system generates information for displaying the plurality of decision trees and the score.