Data Threat Evaluation System Using Regression Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The vast amount of data collected in the healthcare and pharmaceutical sectors makes it challenging to detect fraud, waste, or abuse effectively, as existing methods struggle to derive meaningful insights from the overwhelming data to identify potential nefarious acts in the pharmaceutical supply chain.
Innovation Solution
A data threat evaluation system that uses regression models to identify predictive features indicative of fraud, waste, or abuse, reducing the data types to a subset that indicate threats, and calculates variable threat values to flag potential issues, which can trigger investigations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all collected data is analyzed to detect fraud, waste or abuse, then detection completeness is improved, but processing time and computational resources increase significantly
Solution Approach 1:
The patent segments the overwhelming data into different categories (claims data, enrollment data, provider data, pharmacy data) and applies different analytical approaches to each segment. Regression models are used to identify and focus on specific high-risk segments rather than analyzing all data uniformly, thus reducing processing time while maintaining detection completeness.
Solution Approach 2:
The patent extracts and isolates specific data types and patterns that are most indicative of fraud, waste or abuse using regression analysis. By taking out only the relevant predictive features from the vast dataset and focusing analysis on these extracted elements, the system achieves effective detection without processing the entire dataset.
2Measurement precision
If regression models are used to identify predictive features, then detection precision is improved, but model development complexity increases
Solution Approach 1:
The patent changes the parameters of analysis by using regression models to transform raw data into predictive features with specific statistical properties. By adjusting and optimizing model parameters such as coefficients, significance levels, and feature selection criteria, the system achieves high detection precision while managing development complexity through systematic parameter tuning.
3Productivity
If data is reduced to a subset indicating threats, then processing efficiency is improved, but risk of missing fraudulent patterns increases
Solution Approach 1:
The patent performs preliminary action by using regression models to pre-identify and flag data patterns that are predictive of fraud, waste or abuse before the main detection process. This preliminary filtering based on statistically significant features allows the system to focus subsequent analysis on high-risk cases, improving processing efficiency while maintaining detection accuracy through pre-screening.
Solution Approach 2:
The patent implements feedback mechanisms where the results of initial analysis are used to refine and update the regression models continuously. By feeding back detected patterns and outcomes into model retraining, the system adapts to new fraudulent patterns while maintaining focus on the most relevant data subset, thus balancing processing efficiency with detection accuracy.
Data Source
AI summary
Data threat evaluation systems and methods are described. A data model structure includes a root object query that, when executed, returns a third data subset from the plurality of data types that predate a known threat, the third data subset including data types in both the first data subset and the second data subset; and a model schema to extract, from the third data subset, data types of the first subset that predicate and indicate the threat, the model schema to produce at least an individualized data threat regression model, a script originator regression model, and a script filler data threat regression model using the extracted data types. The system may use the individualized data threat regression model, the script originator regression model, and the script filler data threat regression model back on the data set to identify potential threats. The system can be applied as a fraud, waste or abuse detector.