Classifying an entity for FOLFOX treatment
Inventors
Abraham, Jim • Spetzler, David • Helmstetter, Anthony • Korn, Wolfgang Michael • Magee, Daniel
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12165759-B2
Publication Date
2024-12-10
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 strongly correlate with response of colorectal cancer patients to FOLFOX. Described herein are data structures, data processing, and machine learning models to predict effectiveness 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 administration of 5-fluorouracil/leucovorin combined with oxaliplatin (FOLFOX) or with irinotecan (FOLFIRI).
Core Innovation
The invention relates to classification of a test entity for a treatment by obtaining data that represents the test entity, wherein the obtained data includes data from at least one biomarker selected from a specified list of biomarkers. The treatment classes are defined as a responsive class for a test entity responding to FOLFOX (5-fluorouracil and leucovorin combined with oxaliplatin) and a non-responsive class for a test entity not responding to FOLFOX.
The method provides the obtained biomarker-derived data as input to one or more machine learning models that have each been trained on the same set of training data to determine a particular class of one or more training entities from multiple different entity classes. For each machine learning model, processing of the provided data through each layer of the machine learning model generates output data indicating the particular class as an initial classification for the test entity.
The invention then obtains the output data generated by each of the one or more machine learning models, where the provided output data includes data representing a determination of the initial classification by each machine learning model. Based on the provided output data, a most likely entity class is determined, the most likely entity class being the responsive class or the non-responsive class. The invention is implemented as a method, a system, and non-transitory computer-readable storage media.
Claims Coverage
The independent claims are directed to biomarker-based classification of a test entity into a responsive versus non-responsive class for FOLFOX treatment using one or more machine learning models. Three inventive features are covered across the independent claims: biomarker-driven input, ensemble machine learning initial classification, and determination of a most likely entity class from the model outputs.
Biomarker-driven test entity data for FOLFOX response classification
Obtaining data that represents the test entity, wherein the obtained data includes data from at least one biomarker selected from a specified list; the classification outputs include a responsive class for a test entity responding to FOLFOX (5-fluorouracil and leucovorin combined with oxaliplatin) and a non-responsive class for a test entity not responding to FOLFOX.
Ensemble machine learning for responsive vs non-responsive initial classification
One or more machine learning models are each trained on the same set of training data, receive the obtained data as input, process the data through each layer, and generate output data indicating the particular class as an initial classification for the test entity.
Most likely entity class determination from multiple model outputs
The output data obtained for each machine learning model is used to determine a most likely entity class for the test entity, the most likely entity class being the responsive class or the non-responsive class.
The independent claims collectively require biomarker-selected test-entity data, per-model machine learning processing producing initial classifications for responsive vs non-responsive to FOLFOX, and a final most likely entity class determination based on the multiple model outputs.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?