Biological Entity Pair Ranking Using Structured and Unstructured Evidence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to effectively identify biological entity pairs with intermediate levels of evidence for association, hindering drug discovery research by prioritizing less promising avenues.
Innovation Solution
A computer-implemented method that mines both structured and unstructured datasets to generate scores for entity pairs, using a classifier to determine a predicted ranking based on these scores, optimizing the classifier with machine learning or rule-based systems to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated data mining is performed on medical andscientific datasets to determine evidence levels for biological entity pairs, then productivity in identifying research avenues is improved, but device complexity increases due to multiple data mining processes and classifiers
Solution Approach 1:
The system segments the data mining process into two independent modules: first data mining on unstructured datasets and second data mining on structured datasets. Each module generates separate scores that are then combined by a classifier, allowing independent optimization and maintenance of each component while achieving comprehensive evidence assessment.
Solution Approach 2:
The system merges results from two separate data mining processes (unstructured and structured datasets) into a unified evidence ranking. The classifier integrates the first scores and second scores to produce a combined predicted ranking, leveraging complementary information from both data types to improve overall assessment accuracy.
2Measurement precision
If multiple data mining processes are used to generate comprehensive evidence scores, then measurement precision of evidence levels is improved, but loss of time increases due to processing multiple datasets
Solution Approach 1:
The system performs preliminary data mining operations on both unstructured and structured datasets in parallel or pre-processed manner, generating first scores and second scores before the final classification stage. This preliminary processing allows the classifier to work with pre-computed evidence metrics, reducing the time burden of the final integration step.
Solution Approach 2:
The system implements a multi-stage scoring process where not all datasets need to be fully processed for every query. The classifier can work with partial score sets or use threshold-based filtering to avoid complete processing of all datasets when lower precision is acceptable, balancing time consumption against measurement precision.
Data Source
AI summary
A computer-implemented method of electronically mining medical and scientific datasets to determine a ranking indicating a level of evidence for an association between two entities is disclosed. The method comprises receiving a representation of an entity pair, performing first data mining on one or more unstructured datasets to generate one or more first scores each representing an extent of association between the entities of the entity pair, and performing second data mining on one or more structured datasets to generate one or more second scores each representing an extent of association between the entities of the entity pair. The method also comprises using a classifier to determine a predicted ranking for the entity pair using the one or more first scores and the one or more second scores, and providing the predicted ranking to a user as an indication of the strength of evidence for an association between the entities of the entity pair.