patent.fyi
Applications  /  US 2026/0354733
United States Patent Application Publication

US 2026/0354733 A1

Published
Nov 12, 2026
Filed
Nov 10, 2023
Earliest date on the face
Nov 11, 2022
Classification
A61B 5/364
First published here
Nov 19, 2026

Generated automatically. Not reviewed by a person. Not legal advice or an opinion.

Section 1 · machine-generated

What if?

Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.

1.1Synthetic waveforms for bench testing

What ifWhat if the device used synthetic electrocardiograms to test rhythm classification?
Claim language“obtaining biometric-data of a user”

A waveform generator creates sampled traces from parametric P-wave, QRS-complex, and T-wave shapes and sets the timing of each beat. The device classifies each trace and compares the output with the rhythm label assigned by the generator.

1.2Rhythm scoring with fixed rules

What ifWhat if fixed rhythm rules generated the first information?
Claim language“a pre-trained first neural network model”

A QRS detector locates ventricular beats and computes the intervals between successive detections. A rule engine returns an atrial fibrillation indicator when interval variability exceeds a configured threshold and a waveform check finds no consistent P waves.

1.2Text input for first inference

What ifWhat if the first network inferred the rhythm category from a clinician’s written interpretation?
Claim language“inputting the obtained biometric-data”

A tokenizer converts a narrative rhythm interpretation into token identifiers. A text classification network maps those identifiers to an atrial fibrillation score.

1.3Rhythm scoring with logistic regression

What ifWhat if a logistic regression model generated the second information?
Claim language“a pre-trained second neural network model”

A signal processor extracts beat interval variability and P-wave detection frequency from the electrocardiogram. The regression model multiplies those features by fitted coefficients and converts the sum to an atrial fibrillation probability with a logistic function.

1.4Independent physician rhythm interpretation

What ifWhat if the device used a physician’s independent reading to set the diagnosis result?
Claim language“based on the first and second information”

A review screen presents the electrocardiogram with time and voltage scales. The physician selects a rhythm category, and the device records that selection as the diagnosis result.

Section 2 · machine-generated

Variations

Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.

Other field of use

Equine rhythm assessment during recovery

A veterinary telemetry unit records a single-lead electrocardiogram from a horse during recovery after exercise. A processor feeds each recording window to two neural networks trained on equine recordings, one producing beat timing markers and the other marking intervals with organized atrial activity. A decision module checks for sustained irregular beat spacing during intervals with no detected organized atrial activity and reports an atrial fibrillation category for veterinary review.

Component substitution

Wavelet features with boosted tree classification

A gradient-boosted decision tree ensemble replaces the second neural network. The first neural network processes the electrocardiogram and produces a rhythm classification for the window. A wavelet transform extracts features from the same window, and the tree ensemble uses them to assign a binary label for organized atrial activity. A lookup table maps the two categorical outputs to an atrial fibrillation result or a review result.

Parameter or scale variation

Rhythm analysis with overlapping windows

A patch records one electrocardiogram lead at 250 samples per second and passes 30-second windows to two pretrained networks every 10 seconds. The first network labels rhythm in each window, and the second identifies intervals affected by motion artifact or electrode contact loss. The processor reports an atrial fibrillation segment when three consecutive windows carry that rhythm label and each contains at least 24 seconds of usable signal. The processor stores the start of the first window and the end of the third window as the segment bounds.

Combination with existing systems

Bedside telemetry with waveform review

A central cardiac telemetry station receives continuous electrocardiogram samples from bedside monitors and buffers the samples by bed identifier. Two pretrained networks process the same buffered segment, with one marking ventricular beats and the other marking runs of atrial activity. The station combines the annotations to classify a segment as suspected atrial flutter when regular atrial activity persists across a slower ventricular rhythm. It adds the result and the corresponding waveform strip to the station's review queue.

Ambulatory recording with symptom event markers

An ambulatory electrocardiogram recorder stores a continuous waveform and inserts a timestamp when the wearer presses a symptom button. At upload, an analysis workstation extracts a waveform interval around each marker and sends the same interval to two pretrained networks that output premature ventricular beat locations and ventricular run boundaries, respectively. The workstation counts consecutive marked beats inside each run and assigns a ventricular tachycardia category when the run meets configured beat count and rate criteria. It stores the run times and their offset from the symptom marker in the review record.

Automated or data-driven variant

Automatic recording extension for rhythm assessment

A wearable recorder sends each electrocardiogram window to two pretrained networks that output a rhythm label and a sample-level signal quality mask. A controller counts the usable seconds carrying an atrial fibrillation label and commands the recorder to capture another window if the total is below 30 seconds and recording time remains. The controller stops at 30 usable seconds with the same rhythm label or at a 120-second recording limit. It returns the atrial fibrillation category when the duration threshold is met and an indeterminate result when usable evidence remains insufficient at the limit.

Specific version of a part

Single lead chest recording

Two adhesive chest electrodes feed a differential amplifier and a reference electrode stabilizes the input common-mode voltage. An analog-to-digital converter samples the resulting electrocardiogram at 250 samples per second for 30 seconds.

Atrial fibrillation convolutional scoring

A pre-trained one-dimensional convolutional network receives a 10-second electrocardiogram segment containing 2,500 samples. The output layer returns an atrial fibrillation probability for the segment.

Full trace attention model

A separately pre-trained attention network receives the full 30-second electrocardiogram as successive one-second waveform blocks. A classification head combines the block representations to produce a second atrial fibrillation probability for the recording.

Dual probability thresholding

The processor checks both atrial fibrillation probabilities against their respective thresholds. It assigns the positive category when both meet their thresholds, the negative category when both fall below their thresholds, and an indeterminate category when the decisions differ.

Section 3 · quoted from USPTO records

Other work

Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.

Document AUS 2008/0242954 A1
Published
Oct 2, 2008
Filed
Mar 27, 2008
Class
A61B 5/00
Document BUS 2022/0354407 A1
Published
Nov 10, 2022
Filed
Nov 13, 2019
Class
A61B 5/301
Claim 1
Document A
Document B
1.1obtaining biometric-data of a user;
An image presentation apparatus is provided that includes a display section that sequentially displays a plurality of images, a biosignal acquisition section that acquires a signal value of a biosignal indicating a biological response of an observer watching the image displayed on the display section, and a display controlling section that controls a timing at which an image subsequent to the image being displayed on the display section is displayed on the display section bas
A · US 2008/0242954 A1 · abstract
The apparatus includes an analog front end for biosignal acquisition.
B · US 2022/0354407 A1 · abstract
1.2generating first information about a first disease of the user by inputting the obtained biometric-data to a pre-trained first neural network model;
1.3generating second information about the first disease of the user by inputting the obtained biometric-data to a pre-trained second neural network model; and
1.4generating a diagnosis result for the user regarding a first category of the first disease based on the first and second information.

An empty cell means only that no passage was reproduced there.

Statements in these documents
The present invention relates to an image presentation apparatus, an image presentation method, and a program and, more particularly, the present invention relates to an image presentation apparatus, an image presentation method, and a program for presenting an image to an observer observing the image to control a mental state of the observer.
A · US 2008/0242954 A1 · ¶ 3
Accordingly, there is a case where presentation of the images and sounds appropriate for treating the anxiety of the patient cannot automatically correspond to the psychological condition and anxiety of the patient.
A · US 2008/0242954 A1 · ¶ 6
Advances in wearable health monitoring technology have spurred a growing demand for sensing and acquisition of multiple physiological biosignals.
B · US 2022/0354407 A1 · ¶ 4
In some embodiments of the apparatus, the biosignal is selected from the group including electrocardiogram (ECG), electromyogram (EMG) and electroencephalogram (EEG).
B · US 2022/0354407 A1 · ¶ 7
Cite aspatent.fyi, US 2026/0354733 A1, claim 1, first published 2026-11-19T02:47:19-05:00, https://patent.fyi/US20260354733/, SHA-256 8af8ae1ec0689aa1c696a7c55cc7bc06162dff9c21359289dfd21f03de1e554f
First published
Permanent addresshttps://patent.fyi/US20260354733/
Content fingerprint (SHA-256)8af8ae1ec0689aa1c696a7c55cc7bc06162dff9c21359289dfd21f03de1e554f
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi