Stimulus-Response Variable Extraction from Electronic Health Records
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting patient outcomes from electronic health records are limited by their reliance on a rigid set of structured and unstructured variables, failing to identify and utilize more advanced covariates that could enhance the performance of machine learning algorithms and provide more accurate predictions.
Innovation Solution
The method involves extracting and analyzing events from electronic health records using cognitive models to identify stimulus-response variables, which are then integrated into predictive models to generate more accurate patient outcome predictions by leveraging behavioral covariates and associations within the data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current methods use a limited set of structured and unstructured variables, then the system is simple to operate, but the prediction accuracy is insufficient
Solution Approach 1:
The patent segments the variable extraction process into multiple stages: initial structured variable extraction, unstructured variable extraction from clinical notes, and advanced covariate identification through cognitive modeling. This segmentation allows the system to progressively enhance prediction accuracy while managing complexity through modular processing steps.
Solution Approach 2:
The patent introduces cognitive models as intermediary components that process and analyze extracted variables to identify advanced covariates. These cognitive models act as mediators between raw data and prediction algorithms, enabling the system to derive meaningful patterns and relationships that improve prediction accuracy without directly increasing the complexity of the core prediction engine.
2Productivity
If a wider range of advanced covariates is identified and used, then the prediction performance is enhanced, but the data processing complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining templates and cognitive models that guide the identification of advanced covariates. These pre-configured structures enable the system to systematically process diverse data types and identify relevant patterns before feeding them into prediction algorithms, thereby enhancing prediction performance while managing processing complexity through structured approaches.
Solution Approach 2:
The patent utilizes parameter changes by adjusting the depth and scope of variable extraction based on prediction needs. The system can dynamically modify which types of covariates are extracted and analyzed, allowing it to optimize prediction performance for different clinical scenarios while controlling processing complexity by selectively applying advanced analysis only where beneficial.
3Measurement precision
If cognitive models are used to identify stimulus-response variables, then more accurate behavioral covariates are obtained, but the computational requirements increase
Solution Approach 1:
The patent applies partial action by using cognitive models selectively to identify stimulus-response variables only for specific data types and clinical scenarios where such analysis provides the most value. Rather than applying complex cognitive modeling to all data uniformly, the system targets specific extraction tasks, thereby achieving high covariate accuracy for critical variables while conserving computational resources on less critical data processing.
Data Source
AI summary
A plurality of events are extracted from a plurality of electronic health records associated with a first patient. The extracted plurality of events are analyzed to identify a plurality of stimulus events and a plurality of response events. An association between a first stimulus event and a first response event is determined. A stimulus-response (SR) variable is generated for the first patient based at least in part on the determined association, and the generated SR variable is integrated into one or more predictive cognitive models.