# US 2023/0144374 A1 · Machine learning models for genomic predictive data analysis

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

- Number: US 2023/0144374 A1
- Document ID: US20230144374A1
- Status: Published application
- Kind: A1
- Application number: 17/453,705
- Filed: 2021-11-05
- Published: 2023-05-11
- Earliest claimed: 2021-11-05
- Assignees: Optum, Inc.
- Applicants: Optum, Inc.
- Inventors: Rakhee Srivastava; Vivek Kumar Yadav; Pranav Mahajan; Lalit Singla
- Web page: https://patentgrep.com/patent/US20230144374A1

## CPC

- G16B 5/20: Probabilistic models
- G16B 40/20: Supervised data analysis
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/086: using evolutionary algorithms, e.g. genetic algorithms or genetic programming
- G16B 30/10: Sequence alignment; Homology search
- 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 40/30: Unsupervised data analysis
- G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing genomic predictive data analysis operations. For example, certain embodiments of the present invention utilize systems, methods, and computer program products that perform genomic predictive data analysis operations by using at least one of viral genomic processing machine learning models and bacterial genomic processing machine learning models.

## Claims

### Claim 1

A computer-implemented method for performing predictive genomic analysis, the computer-implemented method comprising:
identifying, using a processor, a first genomic sequence set;
for each first genomic sequence in the first genomic sequence set:
determining, using the processor and a frequency-based k-mer extraction layer of a viral genome processing machine learning model, and based at least in part on the first genomic sequence, one or more frequent k-mers of the first genomic sequence; and
determining, using the processor and a one-dimensional convolutional neural network layer of the viral genome processing machine learning model, and based at least in part on the frequency-based refined representation, a viral replication origin k-mer of the one or more frequent k-mers; and
performing, using the processor, one or more prediction-based actions based at least in part on each viral replication origin k-mer.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, further comprising:
identifying, using the processor, a plurality of viral replication origin k-mers, wherein the plurality of viral replication origin k-mers comprise each viral replication origin k-mer for the first genomic sequence set; and
determining, using the processor, one or more viral genome clusters based at least in part on the plurality of viral replication origin k-mers.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2, further comprising:
identifying, using the processor, an input viral genomic sequence;
for each viral genome clusters, determining, using the processor, a viral cluster similarity measure with respect to the input viral genomic sequence; and
determining, using the processor, a viral strain prediction for the input viral genomic sequence based at least in part on each viral cluster similarity measure.

### Claim 4 (depends on 1)

The computer-implemented method of claim 1, wherein determining the one or more frequent k-mers of a particular first genomic sequence comprises:
determining one or more detected k-mers of the particular first genomic sequence;
for each detected k-mer, determining an occurrence frequency score within the particular first genomic sequence; and
determining the one or more frequent k-mers based at least in part on each occurrence frequency score.

### Claim 5 (depends on 4)

The computer-implemented method of claim 4, wherein determining the one or more frequent k-mers based at least in part on each occurrence frequency score comprises:
identifying a hypothesis k-mer of the one or more frequent k-mers; and
for each detected k-mer, determining that the detected k-mer is one of the one or more frequent k-mers if the occurrence frequency score for the detected k-mer exceeds the occurrence frequency score for the hypothesis k-mer.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1, further comprising:
identifying, using the processor, a second genomic sequence set;
determining, using the processor and a two-dimensional convolutional neural network layer of a bacterial genome processing machine learning model, and based at least in part on the second genomic sequence set, one or more resistant bacterial segments of the second genomic sequence set; and
performing, using the processor, one or more second prediction-based actions based at least in part on the one or more resistant bacterial segments.

### Claim 7 (depends on 6)

The computer-implemented method of claim 6, wherein the two-dimensional convolutional neural network layer is further configured to determine a non-resistant antibacterial recommendation for each resistant bacterial segment.

### Claim 8 (depends on 6)

The computer-implemented method of claim 6, further comprising:
for each resistant bacterial segment, identifying, using the processor, a location-wise frequency measure for the resistant bacterial segment within genomic sequence data for a location-wide designation of the second genomic sequence set; and
generating, using the processor, a location-wise bacterial spread prediction for the location-wide designation based at least in part on each location-wise frequency measure.

### Claim 9 (depends on 6)

The computer-implemented method of claim 6, further comprising:
for each resistant bacterial segment, identifying, using the processor, a demographic frequency measure for the resistant bacterial segment within genomic sequence data for a location-wide designation of the second genomic sequence set;
generating, using the processor, a demographic bacterial spread prediction for the demographic designation based at least in part on each demographic frequency measure.

### Claim 10

An apparatus for performing predictive genomic analysis, the apparatus comprising at least one processor and at least one memory including program code, the at least one memory and the program code configured to, with the processor, cause the apparatus to at least:
identify a first genomic sequence set;
for each first genomic sequence in the first genomic sequence set:
determine, using a frequency-based k-mer extraction layer of a viral genome processing machine learning model and based at least in part on the first genomic sequence, one or more frequent k-mers of the first genomic sequence; and
determine, using a one-dimensional convolutional neural network layer of the viral genome processing machine learning model and based at least in part on the frequency-based refined representation, a viral replication origin k-mer of the one or more frequent k-mers; and
perform one or more prediction-based actions based at least in part on each viral replication origin k-mer.

### Claim 11 (depends on 10)

The apparatus of claim 10, wherein the at least one memory and the program code are further configured to, with the processor, cause the apparatus to at least:
identify a plurality of viral replication origin k-mers, wherein the plurality of viral replication origin k-mers comprise each viral replication origin k-mer for the first genomic sequence set; and
determine one or more viral genome clusters based at least in part on the plurality of viral replication origin k-mers.

### Claim 12 (depends on 11)

The apparatus of claim 11, wherein the at least one memory and the program code are further configured to, with the processor, cause the apparatus to at least:
identify an input viral genomic sequence;
for each viral genome clusters, determine a viral cluster similarity measure with respect to the input viral genomic sequence; and
determine a viral strain prediction for the input viral genomic sequence based at least in part on each viral cluster similarity measure.

### Claim 13 (depends on 10)

The apparatus of claim 10, wherein determining the one or more frequent k-mers of a particular first genomic sequence comprises:
determining one or more detected k-mers of the particular first genomic sequence;
for each detected k-mer, determining an occurrence frequency score within the particular first genomic sequence; and
determining the one or more frequent k-mers based at least in part on each occurrence frequency score.

### Claim 14 (depends on 13)

The apparatus of claim 13, wherein determining the one or more frequent k-mers based at least in part on each occurrence frequency score comprises:
identifying a hypothesis k-mer of the one or more frequent k-mers; and
for each detected k-mer, determine that the detected k-mer is one of the one or more frequent k-mers if the occurrence frequency score for the detected k-mer exceeds the occurrence frequency score for the hypothesis k-mer.

### Claim 15 (depends on 10)

The apparatus of claim 10, wherein the at least one memory and the program code are further configured to, with the processor, cause the apparatus to at least:
identify a second genomic sequence set;
determine, using a two-dimensional convolutional neural network layer of a bacterial genome processing machine learning model and based at least in part on the second genomic sequence set, one or more resistant bacterial segments of the second genomic sequence set; and
perform one or more second prediction-based actions based at least in part on the one or more resistant bacterial segments.

### Claim 16 (depends on 15)

The apparatus of claim 15, wherein the two-dimensional convolutional neural network layer is further configured to determine a non-resistant antibacterial recommendation for each resistant bacterial segment.

### Claim 17 (depends on 15)

The apparatus of claim 15, wherein the at least one memory and the program code are further configured to, with the processor, cause the apparatus to at least:
for each resistant bacterial segment, identify a location-wise frequency measure for the resistant bacterial segment within genomic sequence data for a location-wide designation of the second genomic sequence set; and
generate a location-wise bacterial spread prediction for the location-wide designation based at least in part on each location-wise frequency measure.

### Claim 18 (depends on 15)

The apparatus of claim 15, wherein the at least one memory and the program code are further configured to, with the processor, cause the apparatus to at least:
for each resistant bacterial segment, identify a demographic frequency measure for the resistant bacterial segment within genomic sequence data for a location-wide designation of the second genomic sequence set;
generate a demographic bacterial spread prediction for the demographic designation based at least in part on each demographic frequency measure.

### Claim 19

A computer program product for performing predictive genomic analysis, the computer program product comprising at least one non-transitory computer readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions configured to:
identify a first genomic sequence set;
for each first genomic sequence in the first genomic sequence set:
determine, using a frequency-based k-mer extraction layer of a viral genome processing machine learning model and based at least in part on the first genomic sequence, one or more frequent k-mers of the first genomic sequence; and
determine, using a one-dimensional convolutional neural network layer of the viral genome processing machine learning model and based at least in part on the frequency-based refined representation, a viral replication origin k-mer of the one or more frequent k-mers; and
perform one or more prediction-based actions based at least in part on each viral replication origin k-mer.

### Claim 20 (depends on 19)

The computer program product of claim 19, wherein the computer-readable program code portions are further configured to:
identify a plurality of viral replication origin k-mers, wherein the plurality of viral replication origin k-mers comprise each viral replication origin k-mer for the first genomic sequence set; and
determine one or more viral genome clusters based at least in part on the plurality of viral replication origin k-mers.
