Clinical Trial Patient Selection Using Missing Data Complementation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing clinical trial systems face inefficiencies in selecting candidates due to missing data, particularly in conditions with few cases, limiting the number of suitable patients available for trials.
Innovation Solution
A clinical trial support device and method that acquires patient treatment data, complements missing data using a graph-based approach with machine learning models like GPT-2, GPT-3, and BERT to generate feature vectors, and selects patients based on the complemented data for clinical trials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If graph-based machine learning models (GPT-2, GPT-3, BERT) are used to complement missing patient data, then the pool of eligible patients expands and selection accuracy improves, but the system complexity and computational resources required increase
Solution Approach 1:
The patent introduces a graph-based machine learning model as an intermediary component that bridges the gap between incomplete patient data and the requirements for accurate clinical trial candidate selection. The model acts as a mediator that infers missing data points by analyzing relationships within the patient data graph, thereby improving selection accuracy without requiring complete original data
Solution Approach 2:
The system performs preliminary data complementation using graph-based machine learning before the patient selection process. By pre-filling missing data points and generating feature vectors in advance, the system prepares enriched patient profiles that can be directly used for selection, eliminating the need for manual data collection and preprocessing during the selection phase
2Productivity
If graph-based machine learning models are used to complement missing data, then more patients can be selected for clinical trials, but the computational time and processing resources increase
Solution Approach 1:
The graph-based machine learning model performs data complementation and feature vector generation as a preliminary step before selection. By pre-processing and enriching patient data in advance, the system reduces the computational burden during the actual selection process, enabling faster identification of eligible candidates when needed
Solution Approach 2:
The system creates simplified copies or representations of patient data in the form of feature vectors that capture essential characteristics. These compressed representations can be quickly processed and compared during selection, reducing the time required to evaluate each patient while maintaining selection accuracy
3Reliability
If comprehensive patient data is required for selection, then selection accuracy is maintained, but the number of patients with complete data is limited
Solution Approach 1:
The graph-based machine learning model serves as an intermediary that bridges incomplete patient data with selection requirements. By inferring missing information through graph relationships and pattern recognition, the model enables patients with partial data to be evaluated with the same reliability as those with complete data, thereby expanding the pool of eligible candidates
Solution Approach 2:
The system transforms the selection approach by changing from requiring complete original data parameters to using inferred and complemented data parameters. The graph-based model generates new data representations that capture patient characteristics even when original data is missing, fundamentally altering how eligibility is determined and expanding the eligible patient population
Data Source
AI summary
A clinical trial support device includes an acquisition unit, a complementing unit, a selection unit, and an output unit. The acquisition unit acquires data regarding a treatment of a patient. The complementing unit complements missing data in the data regarding a treatment among data used to select a patient to be clinically tested. The selection unit selects a patient to be clinically tested based on the complemented data. The output unit outputs information about the selected patient to be clinically tested. With such a configuration, the clinical trial support device can support decision-making regarding selection of a patient to be clinically tested.