# US 2024/0053307 A1 · Identifying Repetitive Portions of Clinical Notes and Generating Summaries Pertinent to Treatment of a Patient Based on the Identified Repetitive Portions

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

- Number: US 2024/0053307 A1
- Document ID: US20240053307A1
- Status: Published application
- Kind: A1
- Application number: 18/383,180
- Filed: 2023-10-24
- Published: 2024-02-15
- Earliest claimed: 2018-04-30
- Assignees: Merative US L.P.
- Applicants: Merative US L.P.
- Inventors: Eric W. Brown; Maria Eleftheriou; Anca Sailer; Ching-Huei Tsou
- Web page: https://patentgrep.com/patent/US20240053307A1

## CPC

- G01N 30/26: Conditioning of the fluid carrier; Flow patterns
- G01N 30/80: Fraction collectors
- C01B 23/0036: {Physical processing only}
- G01N 30/32: of pressure or speed (G01N30/36 takes precedence)
- G01N 30/6073: {in open tubular form}
- C01B 2210/0031: Helium
- C01B 2210/0045: Oxygen
- C01B 2210/0046: Nitrogen
- C01B 2210/005: Carbon monoxide
- C01B 2210/0051: Carbon dioxide
- G01N 2030/025: {Gas chromatography}
- G01N 2030/324: {speed, flow rate}
- G01N 2030/326: {pumps}
- G01N 2030/328: {valves, e.g. check valves of pumps}
- Y02P 30/00: Technologies relating to oil refining and petrochemical industry
- G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q)
- C01B: NON-METALLIC ELEMENTS; COMPOUNDS THEREOF; {METALLOIDS OR COMPOUNDS THEREOF NOT COVERED BY SUBCLASS C01C}
- Y02P: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS

## Abstract

A mechanism is provided in a data processing system comprising a processor and a memory, the memory comprising instructions that are executed by the processor to specifically configure the processor to implement a repetitive portion identification and weighting engine. A machine learning model is trained for weighting repetitive portions of patient electronic medical records (EMRs). A repetitive portion identification component applies a plurality of templates to clinical notes of a patient EMR to identify one or more candidate portions that match at least one of the plurality of templates. A content analysis component performs content analysis on the one or more candidate portions to determine whether each given candidate portion is relevant. A weighting component assigns a relative weight to each given candidate portion based on relevance. A cognitive summary graphical user interface (GUI) generation component generates cognitive summary reflecting at least a subset of the one or more candidate portions of the patient EMR. The mechanism outputs the cognitive summary in a GUI to a user.

## Claims

### Claim 1

A method, in a data processing system comprising a processor and a memory, the memory comprising instructions that are executed by the processor to specifically configure the processor to implement a repetitive portion identification and weighting engine, the method comprising:
training a machine learning model for weighting repetitive portions of patient electronic medical records (EMRs), wherein training the machine learning model comprises training the machine learning model based on training keyword features and training weights, to output weights for repetitive portions of patient EMRs, wherein the weights for repetitive portions indicate a relevance of corresponding repetitive portions;
applying, by a repetitive portion identification component executing within the repetitive portion identification and weighting engine, a plurality of templates to clinical notes of a patient EMR to identify one or more candidate portions that match at least one of the plurality of templates;
performing, by a content analysis component executing within the repetitive portion identification and weighting engine, content analysis on the one or more candidate portions to determine whether each given candidate portion is relevant;
assigning, by a weighting component executing within the repetitive portion identification and weighting engine, a relative weight to each given candidate portion based on relevance using the trained machine learning model;
generating, by a cognitive summary graphical user interface (GUI) generation component executing within the repetitive portion identification and weighting engine, a cognitive summary reflecting at least a subset of the one or more candidate portions of the patient EMR; and
outputting the cognitive summary in a GUI to a user.

### Claim 2 (depends on 1)

The method of claim 1, wherein performing the content analysis on the one or more candidate portions comprises determining whether each given candidate portion is relevant to the patient's overall medical condition or whether each given candidate portion is relevant to a reason for the patient's scheduled encounter with a medical professional.

### Claim 3

(canceled)

### Claim 4 (depends on 1)

The method of claim 1, further comprising augmenting key aspects of clinical notes with additional features based on external resources.

### Claim 5 (depends on 4)

The method of claim 4, wherein augmenting key aspects of clinical notes with additional features based on external resources comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes; and
extracting keyword features, formatting features, and character features from the labeled clinical notes.

### Claim 6 (depends on 5)

The method of claim 5, wherein augmenting key aspects of clinical notes with additional features based on external resources further comprises generating a keyword and distance features dictionary vector.

### Claim 7 (depends on 6)

The method of claim 6, wherein augmenting key aspects of clinical notes with additional features based on external resources further comprises performing automatic feature expansion based on the keyword features, the keyword and distance features dictionary vector, and the unlabeled clinical notes to generate keyword features and weights,
wherein training the machine learning model comprises training the machine learning model based on the keyword and distance features dictionary vector, the keyword features and weights, the keyword features, the formatting features, and the characters features.

### Claim 8

A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to implement a repetitive portion identification and weighting engine, wherein the computer readable program causes the computing device to:
train a machine learning model for weighting repetitive portions of patient electronic medical records (EMRs)), wherein training the machine learning model comprises training the machine learning model based on training keyword features and training weights, to output weights for repetitive portions of patient EMRs, wherein the weights for repetitive portions indicate a relevance of corresponding repetitive portions;
apply, by a repetitive portion identification component executing within the repetitive portion identification and weighting engine, a plurality of templates to clinical notes of a patient EMR to identify one or more candidate portions that match at least one of the plurality of templates;
perform, by a content analysis component executing within the repetitive portion identification and weighting engine, content analysis on the one or more candidate portions to determine whether each given candidate portion is relevant;
assign, by a weighting component executing within the repetitive portion identification and weighting engine, a relative weight to each given candidate portion based on relevance using the trained machine learning model;
generate, by a cognitive summary graphical user interface (GUI) generation component executing within the repetitive portion identification and weighting engine, a cognitive summary reflecting at least a subset of the one or more candidate portions of the patient EMR; and
output the cognitive summary in a GUI to a user.

### Claim 9 (depends on 8)

The computer program product of claim 8, wherein performing the content analysis on the one or more candidate portions comprises determining whether each given candidate portion is relevant to the patient's overall medical condition or whether each given candidate portion is relevant to a reason for the patient's scheduled encounter with a medical professional.

### Claim 10

(canceled)

### Claim 11 (depends on 8)

The computer program product of claim 8, wherein the computer readable program causes the computing device to augment key aspects of clinical notes with additional features based on external resources.

### Claim 12 (depends on 11)

The computer program product of claim 11, wherein augmenting key aspects of clinical notes with additional features based on external resources comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes; and
extracting keyword features, formatting features, and character features from the labeled clinical notes.

### Claim 13 (depends on 12)

The computer program product of claim 12, wherein augmenting key aspects of clinical notes with additional features based on external resources further comprises generating a keyword and distance features dictionary vector.

### Claim 14 (depends on 13)

The computer program product of claim 13, wherein augmenting key aspects of clinical notes with additional features based on external resources further comprises performing automatic feature expansion based on the keyword features, the keyword and distance features dictionary vector, and the unlabeled clinical notes to generate keyword features and weights,
wherein training the machine learning model comprises training the machine learning model based on the keyword and distance features dictionary vector, the keyword features and weights, the keyword features, the formatting features, and the characters features.

### Claim 15

An apparatus comprising:
at least one processor; and
a memory coupled to the at least one processor, wherein the memory comprises instructions which, when executed by the at least one processor, cause the at least one processor to implement a repetitive portion identification and weighting engine, wherein the instructions cause the at least one processor to:
train a machine learning model for weighting repetitive portions of patient electronic medical records (EMRs)), wherein training the machine learning model comprises training the machine learning model based on training keyword features and training weights, to output weights for repetitive portions of patient EMRs, wherein the weights for repetitive portions indicate a relevance of corresponding repetitive portions;
apply, by a repetitive portion identification component executing within the repetitive portion identification and weighting engine, a plurality of templates to clinical notes of a patient EMR to identify one or more candidate portions that match at least one of the plurality of templates;
perform, by a content analysis component executing within the repetitive portion identification and weighting engine, content analysis on the one or more candidate portions to determine whether each given candidate portion is relevant;
assign, by a weighting component executing within the repetitive portion identification and weighting engine, a relative weight to each given candidate portion based on relevance using the trained machine learning model;
generate, by a cognitive summary graphical user interface (GUI) generation component executing within the repetitive portion identification and weighting engine, a cognitive summary reflecting at least a subset of the one or more candidate portions of the patient EMR; and
output the cognitive summary in a GUI to a user.

### Claim 16 (depends on 15)

The apparatus of claim 15, wherein performing the content analysis on the one or more candidate portions comprises determining whether each given candidate portion is relevant to the patient's overall medical condition or whether each given candidate portion is relevant to a reason for the patient's scheduled encounter with a medical professional.

### Claim 17

(canceled)

### Claim 18 (depends on 15)

The apparatus of claim 15, wherein the computer readable program causes the computing device to augment key aspects of clinical notes with additional features based on external resources.

### Claim 19 (depends on 18)

The apparatus of claim 18, wherein augmenting key aspects of clinical notes with additional features based on external resources comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes; and
extracting keyword features, formatting features, and character features from the labeled clinical notes.

### Claim 20 (depends on 19)

The apparatus of claim 19, wherein augmenting key aspects of clinical notes with additional features based on external resources further comprises:
generating a keyword and distance features dictionary vector, and
performing automatic feature expansion based on the keyword features, the keyword and distance features dictionary vector, and the unlabeled clinical notes to generate keyword features and weights,
wherein training the machine learning model comprises training the machine learning model based on the keyword and distance features dictionary vector, the keyword features and weights, the keyword features, the formatting features, and the characters features.

### Claim 21 (depends on 1)

The method of claim 1, wherein training the machine learning model comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes;
performing feature construction on clusters of labels from the labeled clinical notes to form labeled keyword features;
extracting formatting features and character features from the labeled clinical notes;
performing automatic feature expansion based on the labeled keyword features, the unlabeled clinical notes, and external keyword features from external medical resources to generate training keyword features and weights; and
training the machine learning model based on the training keyword features and weights, the labeled keyword features, the external keyword features, the formatting features, and the characters features to output the weights for repetitive portions.

### Claim 22 (depends on 8)

The computer program product of claim 8, wherein training the machine learning model comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes;
performing feature construction on clusters of labels from the labeled clinical notes to form labeled keyword features;
extracting formatting features and character features from the labeled clinical notes;
performing automatic feature expansion based on the labeled keyword features, the unlabeled clinical notes, and external keyword features from external medical resources to generate training keyword features and weights; and
training the machine learning model based on the training keyword features and weights, the labeled keyword features, the external keyword features, the formatting features, and the characters features to output the weights for repetitive portions.

### Claim 23 (depends on 15)

The apparatus of claim 15, wherein training the machine learning model comprises:
dividing the patient EMRs into labeled clinical notes and unlabeled clinical notes;
performing feature construction on clusters of labels from the labeled clinical notes to form labeled keyword features;
extracting formatting features and character features from the labeled clinical notes;
performing automatic feature expansion based on the labeled keyword features, the unlabeled clinical notes, and external keyword features from external medical resources to generate training keyword features and weights; and
training the machine learning model based on the training keyword features and weights, the labeled keyword features, the external keyword features, the formatting features, and the characters features to output the weights for repetitive portions.
