Predictive assignments that relate to genetic information and leverage machine learning models
Inventors
Trunck, Ryan P. • Glode, Christopher M. • Powers, Rani K. • Lescallett, Jennifer L.
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11699069-B2
Publication Date
2023-07-11
Expiration Date
Abstract
Systems and methods are provided for performing predictive assignments pertaining to genetic information. One embodiment is a system that includes a genetic prediction server. The genetic prediction server includes an interface that acquires records that each indicate one or more genetic variants determined to exist within an individual, and a controller. The controller selects one or more machine learning models that utilize the genetic variants as input, and loads the machine learning models. For each individual in the records: the controller predictively assigns at least one characteristic to that individual by operating the machine learning models based on at least one genetic variant indicated in the records for that individual. The controller also generates a report indicating at least one predictively assigned characteristic for at least one individual, and transmits a command via the interface for presenting the report at a display.
Core Innovation
A system and method use a genetic prediction server to predictively assign genetic variants to individuals based on records indicating one or more characteristics. The characteristics comprise at least one item selected from phenotypes and behaviors of the individual. The server acquires the records via an interface and operates one or more machine learning models selected by a controller, where at least one model utilizes a combination of multiple ones of the characteristics as input.
For each individual in the records, the controller predicts at least one genetic variant by operating the selected machine learning models based on multiple ones of the characteristics indicated in the records. The at least one genetic variant comprises a variation in a nucleotide sequence. The controller generates a report indicating the predictively assigned genetic variant(s) for at least one individual and transmits a command for presenting the report at a display.
The description includes refinements in which the machine learning models are updated using a cost function and may output confidence values for genetic variants. It further specifies neural network architectures in which multi-layer neural networks use weighted connections with top-layer nodes corresponding to characteristics and bottom-layer nodes corresponding to genetic variants. In embodiments, dimensional coordinates are assigned to input characteristics and the characteristics are convolved in a convolutional layer based on the dimensional coordinates, and outputs across models may be aggregated, including via a weighted average of confidences.
Claims Coverage
The independent claims identified are clm-00001, clm-00008, and clm-00015, each directed to predicting and reporting genetic variants from records of individual characteristics using one or more machine learning models, with display presentation. Across these independent claims, the main inventive features include selecting/loading models, predictively assigning nucleotide-sequence variants from combined characteristics, generating/transmitting a report for display, and refinements such as confidence-threshold gating and cost-function-based model revision.
Acquiring individual characteristic records for phenotypes and behaviors
An interface acquires records that each indicate one or more characteristics determined for an individual, where the characteristics comprise at least one item selected from phenotypes and behaviors of the individual.
Selecting and loading machine learning models that combine multiple characteristics
A controller selects one or more machine learning models at least one of which utilizes a combination of multiple ones of the characteristics as input, loads the one or more machine learning models, and operates the one or more machine learning models based on multiple ones of the characteristics indicated in the records for that individual.
Predictively assigning nucleotide sequence genetic variants from combined characteristics
For each individual in the records, the controller predictively assigns at least one genetic variant by operating the one or more machine learning models based on multiple ones of the characteristics indicated in the records, where the at least one genetic variant comprises a variation in a nucleotide sequence.
Generating and transmitting a report for display presentation
The controller generates a report indicating at least one predictively assigned genetic variant for at least one individual and transmits a command via the interface for presenting the report at a display.
Predictively assigning genetic variants in a method
A method comprising acquiring records indicating characteristics selected from phenotypes and behaviors, selecting and loading one or more machine learning models using a combination of multiple characteristics as input, and for each individual predictively assigning at least one genetic variant that comprises a variation in a nucleotide sequence.
Non-transitory computer readable medium embodying programmed instructions for predictive assignment and reporting
A non-transitory computer readable medium embodying programmed instructions which, when executed by a processor, perform a method that acquires records indicating characteristics selected from phenotypes and behaviors, selects and loads one or more machine learning models, for each individual predictively assigns at least one genetic variant comprising a variation in a nucleotide sequence, generates a report indicating the assigned genetic variant(s), and transmits a command for presenting the report at a display.
Across the independent claims, the core coverage centers on predicting nucleotide-sequence genetic variants for individuals using one or more machine learning models that take combined individual characteristics, including phenotypes and behaviors, as input, then generating and transmitting a report for display. Dependent-claim refinements in the families add mechanisms such as cost-function-based model revision and confidence-threshold gating for variant assignment, as well as specified neural-network structural mappings and coordinate-based convolution.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?