Skew-Mitigated Prediction Model Using Capped and Outlier Submodels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for predicting service costs, such as those for healthcare and content streaming services, face challenges in accuracy due to varying individual costs and the impact of extreme values, leading to skewed predictions based on averages, which can result in inefficient pricing strategies and profitability issues for service providers.
Innovation Solution
A method involving the formation of a claims risk model that includes a capped submodel and an outlier submodel, trained on modified datasets, and a splice model to interpolate predictions, allowing for more accurate per-member-per-month cost estimation by redistributing outcomes across cohorts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional average-based prediction methods are used, then the prediction process is simple, but the prediction accuracy deteriorates due to skewed distributions and extreme values
Solution Approach 1:
The patent divides the prediction model into multiple specialized submodels (capped submodel for typical values, outlier submodel for extreme values, splice model for integration) to handle different data characteristics. This segmentation allows each submodel to focus on specific aspects of the skewed distribution, improving overall prediction accuracy while managing complexity through modular design.
Solution Approach 2:
The patent introduces a splice model as an intermediary component that integrates predictions from the capped submodel and outlier submodel. This intermediary layer reconciles the different prediction outcomes and combines them into a final accurate prediction, resolving the contradiction by adding a coordinating layer rather than directly complicating the core prediction logic.
2Reliability
If fixed fee levels are set based on overall costs, then pricing strategy is simple, but profitability deteriorates due to varying individual subscriber costs
Solution Approach 1:
The patent applies local quality by transitioning from uniform fixed fee levels to individualized dynamic pricing based on each subscriber's specific cost characteristics. The prediction model analyzes individual subscriber data to determine appropriate fee levels, ensuring that each subscriber pays according to their actual resource consumption and risk profile, thereby improving profitability while maintaining pricing simplicity through automated assessment.
3Measurement precision
If more data and information are used for cost prediction, then prediction accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent extracts and separates extreme values (outliers) from the main dataset and processes them through a specialized outlier submodel. This extraction allows the capped submodel to handle typical values efficiently while the outlier submodel专门 processes extreme cases, reducing the burden on the main processing system and improving overall accuracy without proportionally increasing complexity.
Data Source
AI summary
Methods, computer program products, and systems are presented for service cost prediction using machine learning. A claims risk model is formed for predicting a per member per month cost based on variables trained in early prediction models and variables of claim record data from data sources. A training dataset for the claims risk model is modified based on a preconfigured cap value and two distinctive datasets are generated, which trains the claims risk model into a capped submodel and an outlier submodel, respectively. Each submodel makes prediction of the per member per month cost and a slice model interpolates a cost predicted by the capped submodel and another predicted by the outlier submodel. The splice model redistributes the outcomes of the capped submodel and the outlier submodel.