Hybrid Selection Engine Using Prediction Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of selecting high-performing hybrids from large pools of potential hybrids in plant breeding is significant, given the numerous possible combinations of male and female lines, making it difficult to accurately identify hybrids with desirable traits for commercialization.
Innovation Solution
A system and method that employs a selection engine using prediction scores and algorithms to identify sets of hybrids based on phenotypic data, trait distribution, heterotic diversity, and market segmentation, reducing the complexity of hybrid selection while maintaining accuracy and genetic diversity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques are used to evaluate and select hybrids from large pools, then comprehensive evaluation can be performed, but the complexity and time required for selection increases significantly
Solution Approach 1:
The patent segments the hybrid selection process into distinct phases: initial filtering based on parental genetic data, intermediate evaluation of hybrid performance metrics, and final selection based on commercial criteria. This segmentation reduces the complexity of evaluating all possible hybrids by dividing the task into manageable stages, allowing breeders to focus computational and analytical resources on the most promising candidates at each phase.
Solution Approach 2:
The patent applies preliminary action by pre-evaluating parental lines and predicting hybrid performance before actual hybrid creation and testing. Genetic algorithms and predictive models are used to estimate which parental combinations are most likely to produce desirable hybrids, allowing breeders to prioritize resource allocation to the most promising crosses before investing in physical breeding and field testing.
2Productivity
If the number of hybrids to be tested is reduced to manage complexity, then the selection process becomes more efficient, but the risk of missing high-performing hybrids increases
Solution Approach 1:
The patent implements feedback mechanisms where actual hybrid performance data from field tests is continuously fed back into the predictive models and genetic algorithms. This feedback loop allows the system to learn from real-world results and improve its predictions, ensuring that reducing the number of tested hybrids does not compromise the reliability of selection. The system adapts its selection criteria based on accumulated performance data, maintaining high reliability even with reduced testing scope.
Solution Approach 2:
The patent replaces traditional mechanical breeding evaluation methods with computational models and genetic algorithms. Instead of relying solely on physical testing of numerous hybrids, the system uses in silico predictions, statistical models, and machine learning algorithms to identify high-performing hybrids. This substitution allows for more efficient screening while maintaining or improving selection reliability through the power of computational analysis.
3Adaptability or versatility
If more hybrids are advanced in the breeding pipeline, then the chances of finding commercially successful hybrids increase, but the resources and time required for testing and evaluation increase
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting selection criteria, weighting factors, and evaluation thresholds based on breeding objectives, available resources, and performance data. The system can modify parameters such as the number of hybrids to advance, the strictness of selection criteria, and the allocation of testing resources to optimize the balance between commercial success probability and time investment. This flexibility allows the breeding program to adapt to changing conditions and resource availability.
Solution Approach 2:
The patent uses preliminary action through predictive modeling and genetic algorithms to identify and prioritize the most promising hybrid combinations before they are created and tested. By pre-calculating which parental crosses are most likely to succeed based on genetic data and performance metrics, the system reduces the need for extensive testing of low-potential hybrids, thereby shortening the overall breeding pipeline duration while maintaining high success rates.
Data Source
AI summary
Exemplary systems for identifying hybrids for use in a plant breeding pipeline are disclosed. One exemplary system includes a computing device configured to access phenotypic data related to a pool of hybrids from a data structure and determine a prediction score for each of the hybrids in the pool of hybrids based on the accessed phenotypic data. The prediction score is indicative of a probability of selection and/or a probability of success of the hybrid based on historical data. The computing device is also configured to select a group of hybrids from the pool of hybrids based on the prediction score, identify a set of hybrids, from the selected group of hybrids, based on one or more factors associated with the hybrids, and then direct the set of hybrids to a validation phase of the plant breeding pipeline for planting and/or testing.