# US 2026/0018306 A1 · Apparatus and method for generating a preoperative data structure using a pre-operative panel

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

- Number: US 2026/0018306 A1
- Document ID: US20260018306A1
- Status: Published application
- Kind: A1
- Application number: 18/773,195
- Filed: 2024-07-15
- Published: 2026-01-15
- Earliest claimed: 2024-07-15
- Assignees: Anumana, Inc.
- Applicants: Anumana, Inc.
- Inventors: Samir Awasthi; Shahir Asfahan; Shashi Kant; Rakesh Barve; Lars Hegstrom
- Web page: https://patentgrep.com/patent/US20260018306A1

## CPC

- G16H 70/20: relating to practices or guidelines
- G16H 10/60: for patient-specific data, e.g. for electronic patient records
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

An apparatus and method for generating a preoperative data structure using a pre-operative panel are disclosed. The apparatus includes a memory containing instructions configuring at least a processor to receive subject data including ECG data, generate a plurality of panel outputs as a function of the subject data using a pre-operative panel machine-learning module including a plurality of panel machine-learning models, wherein each of the plurality of panel machine-learning models is configured to generate one panel output for one panel focus, wherein generating the plurality of panel outputs includes generating a plurality of sets of panel training data, training each of the plurality of panel machine-learning models using each of the plurality of sets of panel training data and generating the plurality of panel outputs using the plurality of trained panel machine-learning models and generate a pre-operative data structure as a function of the plurality of panel outputs.

## Claims

### Claim 1

An apparatus for generating a preoperative data structure using a pre-operative panel, the apparatus comprising:
at least a processor; and
a memory communicatively connected to the at least a processor, wherein the memory contains instructions configuring the at least a processor to:
receive subject data, wherein the subject data comprises electrocardiogram (ECG) data;
generate a plurality of panel outputs as a function of the subject data using a pre-operative panel machine-learning module;
wherein the pre-operative panel machine-learning module comprises a plurality of panel machine-learning models and a large language model (LLM), wherein each of the plurality of panel machine-learning models and the large language model are configured to generate one panel output for one panel focus as a function of the subject data;
wherein the large language model comprises a transformer architecture that employs self-attention and positional encoding, to receive the subject data as input, analyze the subject data, and generate an output;
wherein generating the plurality of panel outputs comprises:
generating a plurality of sets of panel training data, wherein the plurality of sets of panel training data comprises correlations between exemplary subject data, exemplary panel focuses and exemplary panel outputs, wherein generating the plurality of sets of panel training data comprises:
sanitizing the plurality of sets of panel training data using a dedicated hardware unit comprising circuitry configured to perform signal processing operations, wherein sanitizing the plurality of sets of panel training data comprises:
determining by the dedicated hardware unit that at least one training data entry of the plurality of sets of panel training data has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the plurality of sets of panel training data to create a sanitized plurality of sets of panel training data;
training each of the plurality of panel machine-learning models using each of the sanitized plurality of sets of panel training data; and
generating the plurality of panel outputs using the plurality of trained panel machine-learning models; and
generate a pre-operative data structure as a function of the plurality of panel outputs.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein generating the plurality of panel outputs comprises:
determining at least an ECG feature as a function of the ECG data; and
determining the plurality of panel outputs as a function of the ECG feature.

### Claim 3 (depends on 2)

The apparatus of claim 2, wherein determining the at least an ECG feature further comprises:
generating ECG feature training data, wherein the ECG feature training data comprises correlations between exemplary ECG data and exemplary ECG features;
training an ECG feature machine-learning model using the ECG feature training data; and
determining the at least an ECG feature using the trained ECG feature machine-learning model.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein the plurality of panel machine-learning models comprises a first panel machine-learning model comprising a first panel focus related to coronary heart disease, wherein the first panel machine-learning model is configured to generate a first panel output related to the coronary heart disease as a function of the subject data.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein the plurality of panel machine-learning models comprises a second panel machine-learning model comprising a second panel focus related to pulmonary hypertension, wherein the second panel machine-learning model is configured to generate a second panel output related to the pulmonary hypertension as a function of the subject data.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein the plurality of panel machine-learning models comprises a third panel machine-learning model comprising a third panel focus related to atrial fibrillation, wherein the third panel machine-learning model is configured to generate a third panel output related to the atrial fibrillation as a function of the subject data.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein the plurality of panel machine-learning models comprises a fourth panel machine-learning model comprising a fourth panel focus related to ejection fraction, wherein the fourth panel machine-learning model is configured to generate a fourth panel output related to the ejection fraction as a function of the subject data.

### Claim 8 (depends on 1)

The apparatus of claim 1, wherein the memory contains instructions further configuring the at least a processor to:
generate cohort training data, wherein the cohort training data comprises correlations between exemplary subject data and exemplary subject cohorts;
train a cohort classifier using the cohort training data; and
classify the subject data to one or more subject cohorts using the trained cohort classifier.

### Claim 9 (depends on 8)

The apparatus of claim 8, wherein the memory contains instructions further configuring the at least a processor to update the panel training data as a function of an output of the cohort classifier.

### Claim 10 (depends on 1)

The apparatus of claim 1, wherein the plurality of panel outputs comprises a pre-operative optimization output.

### Claim 11

A method for generating a preoperative data structure using a pre-operative panel, the method comprising:
receiving, using at least a processor, subject data, wherein the subject data comprises electrocardiogram (ECG) data;
generating, using the at least a processor, a plurality of panel outputs as a function of the subject data using a pre-operative panel machine-learning module;
wherein the pre-operative panel machine-learning module comprises a plurality of panel machine-learning models and a large language model (LLM), wherein each of the plurality of panel machine-learning models and the large language model are configured to generate one panel output for one panel focus as a function of the subject data;
wherein the large language model comprises a transformer architecture that employs self-attention and positional encoding, to receive the subject data as input, analyze the subject data, and generate an output;
wherein generating the plurality of panel outputs comprises:
generating a plurality of sets of panel training data, wherein the plurality of sets of panel training data comprises correlations between exemplary subject data, exemplary panel focuses and exemplary panel outputs, wherein generating the plurality of sets of panel training data comprises:
sanitizing the plurality of sets of panel training data using a dedicated hardware unit comprising circuitry configured to perform signal processing operations, wherein sanitizing the plurality of sets of panel training data comprises:
determining by the dedicated hardware unit that at least one training data entry of the plurality of sets of panel training data has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the plurality of sets of panel training data to create a sanitized plurality of sets of panel training data;
training each of the plurality of panel machine-learning models using each of the sanitized plurality of sets of panel training data; and
generating the plurality of panel outputs using the plurality of trained panel machine-learning models; and
generating, using the at least a processor, a pre-operative data structure as a function of the plurality of panel outputs.

### Claim 12 (depends on 11)

The method of claim 11, wherein generating the plurality of panel outputs comprises:
determining, using the at least a processor, at least an ECG feature as a function of the ECG data; and
determining, using the at least a processor, the plurality of panel outputs as a function of the ECG feature.

### Claim 13 (depends on 12)

The method of claim 12, wherein determining the at least an ECG feature further comprises:
generating, using the at least a processor, ECG feature training data, wherein the ECG feature training data comprises correlations between exemplary ECG data and exemplary ECG features;
training, using the at least a processor, an ECG feature machine-learning model using the ECG feature training data; and
determining, using the at least a processor, the at least an ECG feature using the trained ECG feature machine-learning model.

### Claim 14 (depends on 11)

The method of claim 11, wherein the plurality of panel machine-learning models comprises a first panel machine-learning model comprising a first panel focus related to coronary heart disease, wherein the first panel machine-learning model is configured to generate a first panel output related to the coronary heart disease as a function of the subject data.

### Claim 15 (depends on 11)

The method of claim 11, wherein the plurality of panel machine-learning models comprises a second panel machine-learning model comprising a second panel focus related to pulmonary hypertension, wherein the second panel machine-learning model is configured to generate a second panel output related to the pulmonary hypertension as a function of the subject data.

### Claim 16 (depends on 11)

The method of claim 11, wherein the plurality of panel machine-learning models comprises a third panel machine-learning model comprising a third panel focus related to atrial fibrillation, wherein the third panel machine-learning model is configured to generate a third panel output related to the atrial fibrillation as a function of the subject data.

### Claim 17 (depends on 11)

The method of claim 11, wherein the plurality of panel machine-learning models comprises a fourth panel machine-learning model comprising a fourth panel focus related to ejection fraction, wherein the fourth panel machine-learning model is configured to generate a fourth panel output related to the ejection fraction as a function of the subject data.

### Claim 18 (depends on 11)

The method of claim 11, further comprising:
generating, using the at least a processor, cohort training data, wherein the cohort training data comprises correlations between exemplary subject data and exemplary subject cohorts;
training, using the at least a processor, a cohort classifier using the cohort training data; and
classifying, using the at least a processor, the subject data to one or more subject cohorts using the trained cohort classifier.

### Claim 19 (depends on 18)

The method of claim 18, further comprising:
updating, using the at least a processor, the panel training data as a function of an output of the cohort classifier.

### Claim 20 (depends on 11)

The method of claim 11, wherein the plurality of panel outputs comprises a pre-operative optimization output.
