Predictive assignments that relate to genetic information and leverage machine learning models
Inventors
Glode, Christopher M. • Trunck, Ryan P. • 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-9922285-B1
Publication Date
2018-03-20
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 genetic prediction server acquires records that indicate one or more genetic variants determined to exist within an individual. A controller selects one or more machine learning models that utilize genetic variants as input, loads the one or more machine learning models, and assigns a dimensional coordinate to each genetic variant used as an input. For each individual, the controller predictively assigns at least one characteristic by operating the one or more machine learning models using at least one genetic variant indicated in the records as input.
Each machine learning model comprises a multi-layer neural network with layers of nodes coupled via weighted connections. Nodes in an input layer correspond with genetic variants and nodes in an output layer correspond with a characteristic. Each neural network includes a convolutional layer that convolves about genetic variants based on the dimensional coordinates of the genetic variants.
After predicting, the controller generates a report indicating at least one predictively assigned characteristic and transmits a command via the interface for presenting the report at a display. The controller analyzes input indicating accuracy of a predictively assigned characteristic, determines a score for one of the one or more machine learning models based on the analyzed input and a cost function, and revises the weighted connections based on the cost function for each neural network.
Claims Coverage
The document includes three independent claims that cover a genetic prediction server system, a corresponding method, and a non-transitory computer readable medium. The inventive features consist of dimensional-coordinated genetic-variant inputs to convolutional multi-layer neural networks, predictively assigning characteristics to individuals, reporting those assignments for display, and updating weighted connections based on accuracy-related input and a cost function.
Dimensional coordinates for genetic-variant inputs to convolutional multi-layer neural networks
A controller selects one or more machine learning models that utilize genetic variants as input, loads the one or more machine learning models, and assigns a dimensional coordinate to each genetic variant used as an input, where each machine learning model comprises a multi-layer neural network with weighted connections and includes a convolutional layer that convolves about genetic variants based on the dimensional coordinates of genetic variants.
Predictively assign characteristics to individuals using genetic variants as input
For each individual in the records, the controller predictively assigns at least one characteristic to that individual by operating the one or more machine learning models, utilizing at least one genetic variant indicated in the records for that individual as input.
Generate and present a report for predictively assigned characteristics
The controller 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.
Score model performance from analyzed accuracy input using a cost function and revise weighted connections
The controller analyzes input indicating accuracy of a predictively assigned characteristic, determines a score for one of the one or more machine learning models based on the analyzed input and a cost function, and revises the weighted connections based on the cost function for each neural network.
Programmed instructions on a non-transitory computer readable medium to perform predictive assignment and weighted-connection revision
A non-transitory computer readable medium embodies programmed instructions which, when executed by a processor, are operable for acquiring records of genetic variants, selecting and loading one or more machine learning models, assigning dimensional coordinates to genetic variants, predictively assigning at least one characteristic using convolutional multi-layer neural networks based on the dimensional coordinates, generating and transmitting a report for presentation at a display, analyzing accuracy-related input, determining a score based on a cost function, and revising weighted connections based on the cost function.
The claims cover genetic prediction via multi-layer neural networks with weighted connections and a convolutional layer that operates on genetic variants using dimensional coordinates, including generation and presentation of a report of predictively assigned characteristics and revision of weighted connections based on a cost function derived from analyzed accuracy input.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?