Synthesizing Risk Prediction Models via Multi-Task Weight Adjustment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing risk prediction models for diseases are challenging to synthesize quantitatively due to diverse populations, algorithms, and paper qualities, resulting in varied risk factors and weights, making it difficult to generate a generalized risk prediction model that accounts for multiple aspects.

Innovation Solution

A computer-implemented method and system that retrieve and analyze multiple literatures to extract study features and risk factor weights, calculating adjusted weights using a multi-task model and Matrix Variate Normal distribution to form an adjusted risk prediction model, enabling quantitative synthesis across different papers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple literatures with diverse populations, algorithms, and paper qualities are synthesized, then the comprehensiveness of the risk prediction model is improved, but the difficulty of quantitative synthesis increases

Engineering Contradiction:

Improvecomprehensiveness of risk prediction modelVSAvoiddifficulty of quantitative synthesis

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent transforms qualitative literature data into quantitative parameters by extracting study features (sample size, population characteristics, algorithm types) and risk factor weights from multiple literatures. This parameter transformation enables mathematical synthesis through the multi-task model and Matrix Variate Normal distribution, resolving the contradiction between comprehensiveness and synthesis difficulty

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces a multi-task model as an intermediary computational framework that bridges diverse literature sources and the final synthesized risk prediction model. This intermediary model standardizes heterogeneous data from different populations and algorithms into a unified quantitative format, making synthesis tractable

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If risk factor weights are extracted from multiple literatures, then the accuracy of individual risk factors is improved, but the variation in weights across studies increases

Engineering Contradiction:

Improveaccuracy of risk factor weightsVSAvoidconsistency of risk factor weights

Core Design Contradiction:

Measurement precisionVSStability of the object's composition

Solution Approach 1:

The patent applies the Matrix Variate Normal distribution to create an equipotential framework where risk factor weights from different studies are normalized to a common statistical baseline. This distributional approach adjusts weights to account for variations in study quality and population characteristics, achieving consistency while preserving individual study contributions

Inventive Principle:

Principle #12Equipotentiality

Solution Approach 2:

The synthesized risk prediction model serves multiple functions simultaneously: it maintains the specific weight variations from individual studies while providing an adjusted unified model that accounts for cross-study differences. This multi-functionality allows the system to preserve measurement precision from individual studies while achieving stability in the aggregate model

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Adaptability or versatility

If a generalized risk prediction model is generated from multiple studies, then the applicability to different populations is improved, but the time required for model synthesis increases

Engineering Contradiction:

Improveapplicability to different populationsVSAvoidtime required for model synthesis

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent performs preliminary automated extraction of study features and risk factor weights from multiple literatures before synthesis. By pre-processing and organizing data from diverse populations and algorithms into standardized formats, the system reduces the time required for the actual synthesis process while maintaining comprehensive population coverage

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces manual qualitative synthesis with automated computational methods including multi-task models and Matrix Variate Normal distributions. This substitution of mechanical manual analysis with algorithmic processing dramatically reduces synthesis time while enabling the generation of generalized models applicable to multiple populations

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20220359075A1Synthesis for risk prediction models

Publication Date: 2022.11.10 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

This disclosure relates to a method, a system and a computer program product for synthesizing risk prediction models to generate a generalized risk prediction model for a particular disease. The method comprises retrieving a plurality of literatures from one or more databases. Each of the plurality of literatures defines a risk prediction model for a same disease. The method further comprises extracting study features from each of the plurality of literatures. The method further comprises extracting weights of risk factors in the risk prediction model defined by each of the plurality of literatures from the plurality of literatures. The method further comprises calculating adjusted weights of risk factors based on the extracted study features and the extracted weights of risk factors, to form an adjusted risk prediction model.