Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12260945-B2
Publication Date
2025-03-25
Expiration Date
Abstract
The present disclosure relates to a discovery platform including machine-learning techniques for using medical imaging data to study a phenotype of interest, such as complex diseases with weak or unknown genetic drivers. An exemplary method identifying a covariant of interest with respect to drug response phenotype (DRP) of a treatment is disclosed.
Core Innovation
The invention provides a system and method for identifying a covariant of interest with respect to drug response phenotype (DRP) of a treatment using a group of clinical subjects. The approach receives covariant information of a covariate class obtained from the group, and receives a plurality of baseline medical images and a plurality of follow-up medical images from the group. The system obtains a plurality of progression embeddings based on the baseline medical images and the follow-up medical images.
The progression embeddings are input into a trained classification model to obtain a plurality of classification results indicative of DRP values of the group of clinical subjects. The classification results are then used together with the covariant information and one or more machine learning models. Based on the covariant information, the classification results, and the one or more machine learning models, the system determines an association between each candidate covariant and the DRP values to identify the covariant of interest.
In the described implementations, progression embeddings can be obtained using embeddings derived from baseline and follow-up medical images and progression embeddings computed from baseline and follow-up latent-space representations. Downstream use includes identifying clinically relevant patient subgroups and linking patient cluster identity to covariates, including treatment progression and adverse side effects, using association testing and subgroup-specific models with thresholded P-value metrics.
Claims Coverage
The document includes three independent claims (a system, a method, and a non-transitory computer-readable storage medium) that share the core workflow of receiving covariant information and baseline/follow-up medical images, producing progression embeddings, obtaining DRP-indicative classification results, and determining associations between candidate covariates and DRP values to identify a covariant of interest. The inventive features are organized around the image-to-embedding-to-classification pathway followed by association determination using machine learning models and covariant information.
Progression embeddings from baseline and follow-up medical images
receiving a plurality of baseline medical images and a plurality of follow-up medical images from the group of clinical subjects; obtaining a plurality of progression embeddings based on the plurality of baseline medical images and the plurality of follow-up medical images
DRP-indicative classification results from progression embeddings
inputting the plurality of progression embeddings into a trained classification model to obtain a plurality of classification results indicative of DRP values of the group of clinical subjects
Association between candidate covariants and DRP values using machine learning models
determining, based on the covariant information for the group of clinical subjects, the plurality of classification results, and one or more machine learning models, an association between each candidate covariant of a plurality of candidate covariants and the DRP values to identify the covariant of interest
Receiving covariant information of a covariate class
receiving covariant information of a covariate class obtained from a group of clinical subjects
Across the independent claims, the same inventive structure is present: covariant information is received, baseline and follow-up medical images are received to obtain progression embeddings, the progression embeddings are input into a trained classification model to obtain DRP-indicative classification results, and then an association between each candidate covariant and the DRP values is determined using covariant information and one or more machine learning models to identify the covariant of interest.
Stated Advantages
Improved power of imputed DRP analyses vs small-trial true-treatment analyses.
Documented Applications
Evaluating treatment efficacy using predicted continuous medical diagnosis scores and deriving placebo vs treatment progression scores for association with treatment using correlation metrics such as P values.
Identifying clinically relevant patient subgroups by clustering latent-space embeddings from unsupervised/contrastive image models and performing association testing to link clusters to covariates including treatment progression and adverse side effects.
Use cases discussed include liver biopsy/H&E embeddings, gene-expression-based clustering, biomarker discovery, and comparative results illustrating improved power of imputed DRP analyses.
Interested in licensing this patent?