Two-Stage Disease Probability Estimation From Biomarker Expression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for estimating specific diseases, such as cancer, from specimen expression levels face challenges in classification accuracy, especially when the incidence probability is low, leading to degraded estimation performance.
Innovation Solution
An information processing apparatus and method that estimates disease incidence probability in two stages: first estimating a general disease probability based on biomarker expression levels and specimen-related information, then refining this to a specific disease type using learned models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single learned model is used to directly estimate specific disease from specimen, then the device complexity is reduced, but the estimation precision deteriorates
Solution Approach 1:
The patent divides the disease estimation problem into two separate learned models: a first learned model that estimates overall disease incidence probability, and a second learned model that estimates specific disease type probability. This segmentation allows each model to specialize in a specific estimation task, improving overall estimation precision while maintaining manageable complexity through modular architecture.
2Adaptability or versatility
If the incidence probability of specific disease type is low, then the classification problem becomes more difficult, but the estimation performance degrades
Solution Approach 1:
The patent performs preliminary estimation of overall disease incidence probability using the first learned model before estimating specific disease type probability with the second learned model. This preliminary action provides contextual information that helps the second model handle low-incidence cases more effectively, improving reliability even when specific disease types are rare.
Solution Approach 2:
The first disease incidence probability acts as an intermediary that bridges the gap between raw specimen data and specific disease type classification. This intermediate estimation provides additional contextual information that enhances the second model's ability to accurately classify specific disease types, especially those with low incidence probabilities.
Data Source
AI summary
An information processing apparatus according to an embodiment includes a hardware processor connected to a memory. The hardware processor estimates a first disease incidence probability of a disease based on first information about expression levels of one or more types of biomarkers of a specimen. The hardware processor estimates a second disease incidence probability of a disease type based on the first information and the first disease incidence probability.