# US 2025/0166726 A1 · Stratication using multi-modal predictive features

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0166726 A1
- Document ID: US20250166726A1
- Status: Published application
- Kind: A1
- Application number: 18/514,303
- Filed: 2023-11-20
- Published: 2025-05-22
- Earliest claimed: 2023-11-20
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Pablo Meyer Rojas; Daniel Enoch Platt; Akira Koseki; Toshiya Iwamori; Sanjoy Dey; Michiharu Kudo; Simona Rabinovici-Cohen; Itai Guez
- Web page: https://patentgrep.com/patent/US20250166726A1

## CPC

- G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
- G16B 20/40: Population genetics; Linkage disequilibrium
- G16B 40/00: ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G16B 40/20: Supervised data analysis
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

Stratification using multi-modal predictive features constructs unimodal models, where a unimodal model is trained based on a mode of data that is different from another mode of data used to train another unimodal model. The unimodal models are trained to extract features that are predictive of a health condition. The unimodal models are run that extract sets of features, where each set of the sets of features are associated with a mode of data. For each set of the sets of features, the features in the set are ranked, stratification of a population according to a top ranked feature defines a subpopulation, genome-wide association study (GWAS) using genomic data associated with the subpopulation is performed, and at least one genomic variant associated with the health condition is identified based on the performed GWAS.

## Claims

### Claim 1

A computer-implemented method comprising:
constructing a plurality of unimodal models, a unimodal model of the plurality of unimodal models trained based on a mode of data that is different from another mode of data used to train another one of the plurality of unimodal models, the plurality of unimodal models being trained to extract features that are predictive of a health condition;
running the plurality of unimodal models, wherein sets of features are extracted, each set of the sets of features being associated with a mode of data; and
for each set of the sets of features:
ranking the features in the set;
performing a stratification of a population into a subpopulation according to a top ranked feature, wherein the top ranked feature is used as a special trait that has predictive association with the health condition;
performing a genome-wide association study (GWAS) using genomic data associated with the subpopulation; and
identifying, based on the performed GWAS, at least one genomic variant associated with the health condition.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, further including:
retraining each of the plurality of unimodal models based on the respective subpopulation defined during the stratification; and
repeating the running, the ranking, the performing of the stratification, the performing of the GWAS, and the identifying steps.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2, further including iteratively performing the retraining and the repeating until the subpopulation meets a threshold population size.

### Claim 4 (depends on 1)

The computer-implemented method of claim 1, wherein the performing of the stratification, the performing of the GWAS, and the identifying steps are performed for a threshold number of next top ranked features as the top ranked feature.

### Claim 5 (depends on 1)

The computer-implemented method of claim 1, further including performing gene ontology enrichment analysis to discover biological functions associated to single nucleotide polymorphism (SNPs) using data associated with the subpopulation.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1, wherein multiple modes of data, each of which is used to train a respective different unimodal model in the plurality of unimodal models, include at least image data and clinical data.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1, wherein the health condition is a disease.

### Claim 8 (depends on 1)

The computer-implemented method of claim 1, wherein the plurality of unimodal models are combined as a multimodal model, wherein the running the plurality of unimodal models includes running the multimodal model.

### Claim 9

A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions readable by a device to cause the device to:
construct a plurality of unimodal models, a unimodal model of the plurality of unimodal models trained based on a mode of data that is different from another mode of data used to train another one of the plurality of unimodal models, the plurality of unimodal models being trained to extract features that are predictive of a health condition;
run the plurality of unimodal models, wherein sets of features are extracted, each set of the sets of features being associated with a mode of data; and
for each set of the sets of features:
rank the features in the set;
perform a stratification of a population into a subpopulation according to a top ranked feature, wherein the top ranked feature is used as a special trait that has predictive association with the health condition;
perform a genome-wide association study (GWAS) using genomic data associated with the subpopulation; and
identify, based on the performed GWAS, at least one genomic variant associated with the health condition.

### Claim 10 (depends on 9)

The computer program product of claim 9, wherein the device is further caused to:
retrain each of the plurality of unimodal models based on the respective subpopulation defined during the stratification; and
repeat running of the plurality of unimodal models, ranking of the features, performing of the stratification, performing of the GWAS study, and identifying of the at least one genomic variant associated with the health condition.

### Claim 11 (depends on 10)

The computer program product of claim 10, wherein the device is further caused to iteratively perform retraining of the plurality of unimodal models and repeating of the running, ranking, performing of the stratification, until the subpopulation meets a threshold population size.

### Claim 12 (depends on 9)

The computer program product of claim 9, wherein the device is further caused to perform the stratification, perform the GWAS, and identify at least one genomic variant, for a threshold number of next top ranked features as the top ranked feature.

### Claim 13 (depends on 9)

The computer program product of claim 9, wherein the device is further caused to perform gene ontology enrichment analysis to discover biological functions associated to single nucleotide polymorphism (SNPs) using data associated with the subpopulation.

### Claim 14 (depends on 9)

The computer program product of claim 9, wherein multiple modes of data, each of which is used to train a respective different unimodal model in the plurality of unimodal models, include at least image data and clinical data.

### Claim 15 (depends on 9)

The computer program product of claim 9, wherein the health condition is a disease.

### Claim 16 (depends on 9)

The computer program product of claim 9, wherein the plurality of unimodal models are combined as a multimodal model, wherein the device caused to run the plurality of unimodal models includes the device caused to run the multimodal model.

### Claim 17

A system comprising:
at least one memory device; and
at least one computer processor configured to at least:
construct a plurality of unimodal models, a unimodal model of the plurality of unimodal models trained based on a mode of data that is different from another mode of data used to train another one of the plurality of unimodal models, the plurality of unimodal models being trained to extract features that are predictive of a health condition;
run the plurality of unimodal models, wherein sets of features are extracted, each set of the sets of features being associated with a mode of data; and
for each set of the sets of features:
rank the features in the set;
perform a stratification of a population into a subpopulation according to a top ranked feature, wherein the top ranked feature is used as a special trait that has predictive association with the health condition;
perform a genome-wide association study (GWAS) using genomic data associated with the subpopulation; and
identify, based on the performed GWAS, at least one genomic variant associated with the health condition.

### Claim 18 (depends on 17)

The system of claim 17, wherein the at least one computer processor is further configured to:
retrain each of the plurality of unimodal models based on the respective subpopulation defined during the stratification; and
repeat running of the plurality of unimodal models, ranking of the features, performing of the stratification, performing of the GWAS study, and identifying of the at least one genomic variant associated with the health condition.

### Claim 19 (depends on 18)

The system of claim 18, wherein the at least one computer processor is further configured to iteratively perform retraining of the plurality of unimodal models and repeating of the running, the ranking, the performing of the stratification, until the subpopulation meets a threshold population size.

### Claim 20 (depends on 19)

The system of claim 19, wherein the at least one computer processor is further configured to perform the stratification, perform the GWAS, and identify at least one genomic variant, for a threshold number of next top ranked features as the top ranked feature.
