# US 12,118,468 B2 · Methods and systems for predicting prescription directions using machine learning algorithm

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

- Number: US 12,118,468 B2
- Document ID: US12118468B2
- Status: Granted
- Kind: B2
- Application number: 18/242,098
- Filed: 2023-09-05
- Published: 2023-12-21
- Granted: 2024-10-15
- Earliest claimed: 2019-07-31
- Assignees: Express Scripts Strategic Development, Inc.
- Applicants: Express Scripts Strategic Development, Inc.
- Inventors: Sudipto Dey; Pulla Reddy P. Yeduru
- Web page: https://patentgrep.com/patent/US12118468B2

## CPC

- G06N 3/08: Learning methods
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0499: Feedforward networks
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G06Q 40/08: Insurance
- G16H 20/10: relating to drugs or medications, e.g. for ensuring correct administration to patients
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

Methods and systems for predicting prescription drug directions are described. In one embodiment, a drug direction prediction subsystem receives and pre-processes values of a plurality of required pharmacy elements for a corresponding prescription of a plurality of prescriptions, generates respective weights for the values of the plurality of required pharmacy elements of the prescription based on one or more of the values of the plurality required pharmacy elements of the prescription, creates a machine learning model to be used by the applicable one of the plurality of machine learning algorithms in predicting drug directions of the prescription, the machine learning model using the values of the plurality of required pharmacy elements of the prescription and the respective weights, and predicts a plurality of drug directions of a new prescription by executing the machine learning model using weighted values of the plurality of required pharmacy elements of the new prescription.

## Claims

### Claim 1

A system for automatically processing prescriptions, the system comprising:
a memory component, configured to store a plurality of prescriptions;
at least one processor communicatively coupled to the memory component, the at least one processor configured to:
receive and pre-process values of a plurality of required pharmacy elements for a corresponding prescription of the plurality of prescriptions;
generate respective weights for the values of the plurality of required pharmacy elements of the prescription based on one or more of the values of the plurality required pharmacy elements of the prescription;
create a machine learning model to be used by the applicable one of the plurality of machine learning algorithms in predicting drug directions of the prescription, the machine learning model using the values of the plurality of required pharmacy elements of the prescription and the respective weights; and
predict a plurality of drug directions of a new prescription by executing the machine learning model using weighted values of the plurality of required pharmacy elements of the new prescription.

### Claim 2 (depends on 1)

The system of claim 1, wherein the at least one processor is further configured to:
obtain records, to create obtained records;
train a first one of a plurality of machine learning algorithms stored in the memory component using a predetermined percentage of the obtained records, to create a trained first algorithm, wherein the obtained records include the predetermined percentage and a remainder;
implement the trained first algorithm on the remainder of the obtained records;
determine a first success rate of the trained first algorithm, based on implementing the trained first algorithm on the remainder;
train a second one of the plurality of machine learning algorithms stored in the memory component using the predetermined percentage of the obtained records, to create a trained second algorithm;
implement the trained second algorithm on the remainder of the obtained records;
determine a second success rate of the trained second algorithm;
perform a comparison of the first success rate and the second success rate; and
select either the trained first algorithm or the trained second algorithm based on the comparison.

### Claim 3 (depends on 2)

The system of claim 2, wherein the at least one processor trains the first one of the plurality of machine learning algorithms by being further configured to:
use the first one of the plurality of machine learning algorithms to perform an analysis of each of the predetermined percentage of the obtained records; and
determine known values for required pharmacy elements and relationships between the known values and drug directions for the predetermined percentage, based on the analysis.

### Claim 4 (depends on 2)

The system of claim 2, wherein the first success rate indicates a frequency of the first trained algorithm correctly predicting drug directions for an individual prescription of the remainder of the obtained records.

### Claim 5 (depends on 2)

The system of claim 2, wherein the at least one processor is further configured to:
identify a highest success rate, based on the comparison of the first success rate and the second success rate; and
use the highest success rate to select either the trained first algorithm or the trained second algorithm.

### Claim 6 (depends on 1)

The system of claim 1, wherein the at least one processor tests the machine learning model using current data, by being further configured to:
determine a success rate associated with the machine learning model.

### Claim 7 (depends on 1)

The system of claim 1, wherein the at least one processor is further configured to:
test the machine learning model according to a timed interval schedule; and
determine whether to replace or retrain the machine learning module on an ongoing basis, according to the timed interval schedule.

### Claim 8 (depends on 1)

The system of claim 1, wherein the plurality of machine learning algorithms comprise neural network machine learning algorithms.

### Claim 9

A method comprising:
a drug direction prediction subsystem receiving and pre-processing values of a plurality of required pharmacy elements for a corresponding prescription of a plurality of prescriptions;
the drug direction prediction subsystem generating respective weights for the values of the plurality of required pharmacy elements of the prescription based on one or more of the values of the plurality required pharmacy elements of the prescription;
the drug direction prediction subsystem creating a machine learning model to be used by the applicable one of the plurality of machine learning algorithms in predicting drug directions of the prescription, the machine learning model using the values of the plurality of required pharmacy elements of the prescription and the respective weights; and
the drug direction prediction subsystem predicting a plurality of drug directions of a new prescription by executing the machine learning model using weighted values of the plurality of required pharmacy elements of the new prescription.

### Claim 10 (depends on 9)

The method of claim 9, further comprising:
obtaining records, to create obtained records;
training a first one of a plurality of machine learning algorithms stored in the memory component using a predetermined percentage of the obtained records, to create a trained first algorithm, wherein the obtained records include the predetermined percentage and a remainder;
implementing the trained first algorithm on the remainder of the obtained records;
determining a first success rate of the trained first algorithm, based on implementing the trained first algorithm on the remainder;
training a second one of the plurality of machine learning algorithms stored in the memory component using the predetermined percentage of the obtained records, to create a trained second algorithm;
implementing the trained second algorithm on the remainder of the obtained records;
determining a second success rate of the trained second algorithm;
performing a comparison of the first success rate and the second success rate; and
select either the trained first algorithm or the trained second algorithm based on the comparison.

### Claim 11 (depends on 10)

The method of claim 10, wherein training the first one of the plurality of machine learning algorithms further comprises:
using the first one of the plurality of machine learning algorithms to perform an analysis of each of the predetermined percentage of the obtained records; and
determining known values for required pharmacy elements and relationships between the known values and drug directions for the predetermined percentage, based on the analysis.

### Claim 12 (depends on 10)

The method of claim 10, wherein the first success rate indicates a frequency of the first trained algorithm correctly predicting drug directions for an individual prescription of the remainder of the obtained records.

### Claim 13 (depends on 10)

The method of claim 10, further comprising:
identifying a highest success rate, based on the comparison of the first success rate and the second success rate; and
using the highest success rate to select either the trained first algorithm or the trained second algorithm.

### Claim 14 (depends on 9)

The method of claim 9, wherein testing the subset of the plurality of machine learning algorithms using current data and previously obtained data, further comprises:
determining a success rate associated with the machine learning model.

### Claim 15 (depends on 9)

The method of claim 9, further comprising:
testing the machine learning model according to a timed interval schedule; and
determining whether to replace or retrain the machine learning module on an ongoing basis, according to the timed interval schedule.

### Claim 16 (depends on 9)

The method of claim 9, wherein the plurality of machine learning algorithms comprise neural network machine learning algorithms.

### Claim 17 (depends on 9)

The method of claim 9, further comprising:
displaying the plurality of predicted drug directions of the new prescription;
receiving a user selection of one of the plurality of displayed predicted drug directions, to create a selected drug direction; and
populating a form with the selected drug direction.

### Claim 18

A non-transitory machine-readable medium comprising instructions, which, when executed by one or more processors, cause the one or more processors to perform the following operations:
receive and pre-process values of a plurality of required pharmacy elements for a corresponding prescription of the plurality of prescriptions;
generate respective weights for the values of the plurality of required pharmacy elements of the prescription based on one or more of the values of the plurality required pharmacy elements of the prescription;
create a machine learning model to be used by the applicable one of the plurality of machine learning algorithms in predicting drug directions of the prescription, the machine learning model using the values of the plurality of required pharmacy elements of the prescription and the respective weights; and
predict a plurality of drug directions of a new prescription by executing the machine learning model using weighted values of the plurality of required pharmacy elements of the new prescription.

### Claim 19 (depends on 18)

The non-transitory, computer-readable medium of claim 18, further causes the one or more processors to perform the following operations:
obtain records, to create obtained records;
train a first one of a plurality of machine learning algorithms stored in the memory component using a predetermined percentage of the obtained records, to create a trained first algorithm, wherein the obtained records include the predetermined percentage and a remainder;
implement the trained first algorithm on the remainder of the obtained records;
determine a first success rate of the trained first algorithm, based on implementing the trained first algorithm on the remainder;
train a second one of the plurality of machine learning algorithms stored in the memory component using the predetermined percentage of the obtained records, to create a trained second algorithm;
implement the trained second algorithm on the remainder of the obtained records;
determine a second success rate of the trained second algorithm;
perform a comparison of the first success rate and the second success rate; and
select either the trained first algorithm or the trained second algorithm based on the comparison.

### Claim 20 (depends on 19)

The non-transitory, computer-readable medium of claim 19, further causes the one or more processors to perform the following operations:
use the first one of the plurality of machine learning algorithms to perform an analysis of each of the predetermined percentage of the obtained records; and
determine known values for required pharmacy elements and relationships between the known values and drug directions for the predetermined percentage, based on the analysis.