Multiple Myeloma Mortality Prediction Model Using Variable Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational models for predicting mortality in multiple myeloma patients are constrained by limited numerical variables and predictors, resulting in crude, qualitative predictions rather than specific numerical forecasts.
Innovation Solution
The development of advanced computer models that generate specific numerical predictions by narrowing the universe of variables to the most predictive ones, using multiple regression analyses to determine weighting factors and improve model training efficiency, allowing for faster processing with reduced computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If numerical computer models use limited variables and predictors, then model complexity is reduced, but prediction precision and accuracy deteriorate
Solution Approach 1:
The patent performs preliminary variable selection through multiple regression analyses before building the final prediction model. Candidate variables are screened and ranked based on their predictive power for mortality outcomes, allowing the model to focus only on the most relevant predictors. This preliminary action enables the model to achieve high prediction precision with a limited set of carefully selected variables rather than using all available data.
Solution Approach 2:
The patent transforms the prediction approach by changing from using all available variables to using a optimized subset of variables with specific weighting factors. The model dynamically adjusts parameter importance through regression coefficients, changing the effective parameters from a static complete set to a dynamic weighted subset that maximizes prediction accuracy while minimizing complexity.
2Loss of information
If all available variables are used in the model, then prediction completeness is improved, but computational resources and training time increase
Solution Approach 1:
The patent extracts and removes irrelevant or less predictive variables from the complete set of available data through multiple regression analyses. By taking out only the most predictive variables and their associated weighting factors, the model maintains prediction completeness for mortality outcomes while significantly reducing computational resource requirements and training time.
Solution Approach 2:
The patent applies partial action by using only the necessary subset of variables required for accurate mortality prediction rather than processing all available variables. This selective approach provides sufficient predictive power for clinical decision-making without the excessive computational burden of analyzing every available data point.
3Measurement precision
If more variables are included in the model, then prediction accuracy may improve, but model training speed decreases
Solution Approach 1:
The patent performs preliminary screening and ranking of candidate variables using multiple regression analyses before final model training. This preliminary action identifies and prioritizes the most predictive variables, allowing the training process to focus only on these high-value variables. Consequently, the model achieves high prediction accuracy with faster training speed compared to training on all available variables.
Data Source
AI summary
Computer-implemented systems and methods are provided for constructing a numerical model to generate a probability that a patient diagnosed with multiple myeloma will die within a predetermined period of time. Reference data for a plurality of patients diagnosed with multiple myeloma is received. First variables selected from the reference data are deemed predictive of mortality for a first predetermined period of time, and second variables selected from the reference data are deemed predictive of mortality for a second predetermined period of time. A first computer model comprising a combination of variables of the first selected variables and first weighting factors is generated. A second computer model comprising a combination of variables of the second selected variables and second weighting factors is generated. The first computer model and the second computer model are trained using the reference data to determine numerical values for the respective first and second weighting factors.