Core Patient Selection in Clusters for Accurate ML Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models struggle to efficiently process multi-modal patient data to identify representative patient clusters and classify new patients accurately, leading to high computational demands and potential overfitting.
Innovation Solution
A machine learning system that selects core patients for each cluster based on centrality scores, using encoder and classification models to generate embeddings and classify patients, focusing on a limited number of features to reduce computational complexity and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing machine learning models process all patient data to identify representative patient clusters, then classification accuracy may be improved, but computational demands increase significantly
Solution Approach 1:
The patent extracts and identifies a smaller subset of core patients from each patient cluster based on centrality scores. Instead of processing all patient data for training, the system selects only the most representative core patients, thereby reducing computational demands while maintaining classification accuracy.
Solution Approach 2:
The patent applies local quality by differentiating between core patients and non-core patients within clusters. Core patients are identified as having higher centrality scores and are selected for training, giving them special importance in the classification process while reducing the need to process all patient data equally.
2Reliability
If existing machine learning models use all patient data for training, then model completeness is maintained, but overfitting risk increases
Solution Approach 1:
The patent extracts a representative subset of core patients from each cluster for training purposes. By selecting only the most central and representative patients, the model learns from diverse yet concise examples, reducing overfitting risk while maintaining model completeness and generalizability.
Solution Approach 2:
The patent changes the parameter of training data selection from including all patients to including only core patients with high centrality scores. This parameter change optimizes the training set composition, improving model generalizability and reducing overfitting while maintaining reliability.
3Loss of information
If existing machine learning models process multi-modal patient data comprehensively, then analysis thoroughness is improved, but computational complexity increases
Solution Approach 1:
The patent extracts core patients from each multi-modal patient cluster based on centrality scores computed from their embeddings. This extraction approach maintains analysis thoroughness by preserving the most representative patients while reducing computational complexity in subsequent training and classification operations.
Solution Approach 2:
The patent applies local quality by identifying and focusing computational resources on core patients within each cluster. These core patients are determined to have the highest quality representation of their respective clusters, allowing thorough analysis of critical cases while reducing overall computational complexity.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing biomedical data of a plurality of patients. The system selects, for each patient cluster in a set of patient clusters, a proper subset of a plurality of patients included in the patient cluster as core patients based on centrality scores of the patients in the patient cluster. The system outputs data identifying: (i) the set of patient clusters, and (ii) the core patients for each patient cluster.