Pan-cancer platinum response predictor
Inventors
Abraham, Jim • Korn, Wolfgang Michael • Spetzler, David
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12731666-B2
Publication Date
2026-09-08
Expiration Date
Abstract
Comprehensive molecular profiling provides a wealth of data concerning the molecular status of patient samples. Such data can be compared to patient response to treatments to identify biomarker signatures that predict response or non-response to such treatments. This approach has been applied to identify biomarker signatures that correlate with response of cancer patients to platinum-based chemotherapy. Described herein are data structures, data processing, and machine learning models to predict a probability of benefit of a treatment for a disease or disorder of a subject having a particular set of biomarkers, as well as an exemplary application of such a model to precision medicine, e.g., to methods for selecting a treatment based on a molecular profile, e.g., a treatment comprising platinum therapy.
Core Innovation
The invention relates to classification of an entity for responding to a treatment that includes a platinum therapy. An entity is assessed by performing nucleic acid sequencing of a biological sample to obtain input data representing the entity, where the input data includes data from at least five biomarkers selected from a specified list. The classification targets multiple candidate entity classes that include a responsive class and a non-responsive class.
For each machine learning model of a plurality of machine learning models, input data is provided to a trained model that has been trained to determine a classification among the multiple candidate entity classes. Each model processes the input data to generate output data representing an initial entity class. Output data from each of the plurality of machine learning models is then used to determine an actual entity class for the entity, where the actual entity class is either the responsive class or the non-responsive class.
The invention also encompasses system and software implementations that obtain the sequencing-derived input data, apply multiple trained machine learning models to generate initial entity classes, and determine the actual responsive versus non-responsive classification based on the models’ output data. In additional embodiments, the input data includes data measured using nucleic acid sequencing for biomarkers from the specified list, and may further include one or more biomarkers referenced from Tables 2-8.
Claims Coverage
The provided material identifies three independent claims covering a method, a system, and a non-transitory computer-readable medium. Across these independent claims, the core claim coverage centers on nucleic acid sequencing-derived biomarker input data, multiple trained machine learning models that each output an initial class, and determination of an actual responsive versus non-responsive class for platinum therapy response.
Platinum therapy response classification using sequencing-based biomarker input
Obtaining input data that represents an entity by performing nucleic acid sequencing of a biological sample, where the input data includes data from at least five biomarkers selected from a specified list.
Multiple candidate class training for responsive vs non-responsive outcomes
Training a machine learning model using training data to determine a class of multiple candidate entity classes including a responsive class of responding to a treatment that includes a platinum therapy and a non-responsive class of not responding to the treatment that includes the platinum therapy.
Ensemble-style use of multiple machine learning models to determine an actual class
Providing output data obtained for each of a plurality of machine learning models, where the output data includes data representing an initial entity class determined by each of the plurality of machine learning models; and determining, based on the output data for each of the plurality of machine learning models, an actual entity class that is the responsive class or the non-responsive class.
System implementation of the classification operations
A system comprising one or more computers and one or more storage media storing instructions executable to perform operations that obtain sequencing-based input data, apply the plurality of machine learning models, generate initial entity class output data, and determine the actual entity class as responsive or non-responsive.
Non-transitory computer-readable medium implementation of the classification operations
A non-transitory computer-readable medium storing software comprising instructions executable by one or more computers to perform operations including obtaining sequencing-based input data, applying the plurality of machine learning models, generating initial entity class output data, and determining the actual entity class as responsive or non-responsive.
Across the method, system, and non-transitory computer-readable medium claims, the central inventive coverage is the combination of nucleic-acid-sequencing biomarker input data with multiple trained machine learning models that each output an initial entity class, followed by determining the actual responsive versus non-responsive entity class for a treatment that includes platinum therapy.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?