Statistical Combination Therapy Discovery System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods struggle to identify statistically significant combination therapies from clinical data due to the difficulty in formulating reasonable hypotheses and testing numerous candidate combinations, as well as the inability to consider confounding factors and measure outcomes effectively.
Innovation Solution
A computer-implemented method that iteratively combines potential combination therapies using a breadth-first search, performs univariate and multivariate analyses to identify candidate therapies with statistically significant protection against specific outcomes, and prunes therapies based on predefined thresholds, while considering confounding factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all candidate combination therapies are tested to identify effective treatments, then the completeness of therapy evaluation is improved, but the time and resources required become prohibitive due to the sheer number of numerical combinations
Solution Approach 1:
The patent performs univariate analysis as a preliminary step before multivariate analysis. This preliminary action filters out ineffective individual treatments and reduces the search space for combination therapies, allowing the system to evaluate completeness without testing all possible combinations exhaustively
Solution Approach 2:
The patent segments the therapy evaluation process into distinct phases: univariate analysis of individual treatments, followed by multivariate analysis of combinations. This segmentation allows systematic evaluation of completeness while managing time constraints through hierarchical filtering
2Ease of operation
If association rule mining methods are used to discover co-occurrence patterns, then the ease of identifying potential therapies is improved, but the ability to compare effectiveness against single-intervention therapies deteriorates
Solution Approach 1:
The patent merges association rule mining with formal statistical hypothesis testing. It combines the pattern-discovery capability of association rules with the effectiveness comparison capability of univariate and multivariate analyses, achieving both ease of identification and precise measurement of therapeutic benefit
Solution Approach 2:
The patent creates a multi-functional analysis system that performs multiple functions: discovering co-occurrence patterns, comparing combination therapy effectiveness against single interventions, and controlling for confounding factors. This universal approach resolves the limitation of single-purpose methods
3Measurement precision
If the analysis considers multiple confounding factors to ensure statistical validity, then the accuracy of outcome measurement is improved, but the complexity of the analysis increases
Solution Approach 1:
The patent performs univariate analysis with confounding factor control as a preliminary step before multivariate combination analysis. This preliminary action adjusts outcomes for confounding factors early, simplifying subsequent combination analyses while maintaining measurement accuracy
Solution Approach 2:
The patent segments confounding factor adjustment into distinct analytical phases: univariate analysis with confounder control, then multivariate analysis. This segmentation manages complexity by handling confounding factors systematically at each stage rather than all at once
Data Source
AI summary
Combination therapies are automatically discovered that provide statistically significantly greater protection against a specific outcome, compared to therapies with fewer treatments. A dataset is received that associates each of a plurality of respective individual treatments with a corresponding outcome and a corresponding set of confounding factors. A plurality of new potential combination therapies are combined iteratively. The received dataset is used to automatically identify at least one candidate combination therapy of the new potential combination therapies that provides a statistically significantly greater protection against an outcome with reference to each of the plurality of respective individual treatments.