patent.fyi
Applications  /  US 2026/0354698
United States Patent Application Publication

US 2026/0354698 A1

Published
Nov 12, 2026
Filed
May 6, 2025
Earliest date on the face
May 6, 2025
Classification
A61B 5/02427
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.

1.1Broadband lamp with filters

What ifWhat if one broadband lamp supplies all illumination bands?
Claim language“a plurality of light emitting diode (LED) emitters”

A tungsten-halogen lamp sends light through a rotating filter wheel with passbands centered at 660, 810, and 940 nm. A photodiode measures diffuse return during each filter interval, and the controller groups the samples by wheel position to recover a pulse waveform for each band.

1.2Separately powered detector module

What ifWhat if the photodetector operates in a separately powered module with no electrical connection to the emitter board?
Claim language“electronically connected to the printed circuit board”

The photodiode connects by package leads to a transimpedance amplifier and digitizer in a battery-powered detector module. A plastic optical fiber carries digitized samples to the controller, and a second fiber carries the emitter board's timing markers to the detector module so it can label each sample.

1.3Analog pulse timing extraction

What ifWhat if the controller receives only beat intervals calculated by a separate analog timing stage?
Claim language“receive light signal data from each of the plurality”

An analog switch routes the amplified photodiode signal into a separate sample-and-hold circuit for each illumination time slot. Band-pass filters and hysteretic comparators generate beat edges, and a counter sends channel-tagged intervals to the controller for conversion to pulse rates.

1.3Waveform fusion before estimation

What ifWhat if the controller combines all optical waveforms before calculating a single pulse rate?
Claim language“calculate a plurality of health parameters”

The controller subtracts dark samples and divides each wavelength's pulsatile signal by its slowly varying baseline. It adds the normalized waveforms sample by sample and derives one pulse rate from the dominant frequency of the combined waveform.

1.3Pulse rate by channel selection

What ifWhat if the controller reports the pulse rate from the channel with the highest signal quality score?
Claim language“an average or a weighted average”

The controller estimates a pulse rate for each wavelength and scores each channel by the prominence of its pulse-frequency peak relative to the surrounding spectrum. A three-axis accelerometer identifies dominant motion frequencies, and the controller excludes channels whose pulse peaks coincide with those frequencies before selecting the highest-scoring remaining channel.

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

Reflectance pulse sensing for cattle

A spring clip holds a circuit board with 660 nm, 810 nm, and 940 nm LEDs and a silicon photodiode against the inner surface of a cow’s ear. A black elastomer rim blocks ambient light and distributes contact pressure around the optical window. The controller pulses the LEDs in sequence and subtracts a reading taken with all LEDs off from each illuminated reading. It estimates pulse rate independently for each wavelength and averages the three rates for transmission to a livestock monitor.

Component substitution

Digital light sensor with integrated readout

A digital ambient light sensor with a broadband silicon photodiode and an integrated analog-to-digital converter replaces the discrete photodiode and external analog readout circuit. The sensor mounts beside 660 nm and 850 nm LEDs on the same board, with an opaque divider blocking direct light from the emitters. The controller triggers a sensor conversion during each LED pulse and a third conversion with both LEDs off, then retrieves the readings over a two-wire serial bus. It subtracts the dark reading from each channel, extracts a pulse rate from each resulting waveform, and averages the two rates.

Parameter or scale variation

Two detector spacings on one board

A reflectance board carries 660 nm, 810 nm, and 940 nm LEDs in a central cluster and two photodiodes at nominal center-to-center distances of 3 mm and 7 mm from the cluster. The controller repeats a 10 ms frame containing three 1 ms illumination slots, one 1 ms dark slot, and 6 ms with all emitters off. Both photodiodes are sampled in each slot, and the controller subtracts each photodiode’s dark reading from its illuminated readings. For each wavelength, it selects the detector with the stronger pulse-band signal relative to broadband noise and averages the three wavelength-specific pulse rates.

Automated or data-driven variant

Motion informed weighting of pulse estimates

A wrist sensor board carries three wavelength channels, a shared photodiode, and a triaxial accelerometer sampled on the same time base. The controller estimates a pulse rate for each wavelength from the dominant spectral peak between 0.7 Hz and 3.5 Hz in a rolling 12-second window. It scores each peak by its power relative to the rest of the pulse band and reduces the score in proportion to the optical waveform’s coherence with accelerometer data near that frequency. It normalizes the scores to sum to one and uses them to compute a weighted mean of the three pulse rates.

Automatic exposure control for each wavelength

A reflectance head uses three LEDs with separate programmable current settings and a shared photodiode connected to an analog-to-digital converter. The controller pulses one LED at a time and samples the converter during illumination and during a dark slot in each frame. Over successive one-second windows, it raises or lowers each LED current independently to keep the median illuminated reading near midscale, within the LED’s rated pulsed current. It calculates a pulse rate for each wavelength from dark-subtracted samples collected at a settled current and averages the rates from channels with unsaturated readings.

Specific version of a part

One wavelength at multiple positions

Three nominally 850 nm emitters sit 3, 5, and 7 mm from a common photodiode and pulse in separate time slots. The controller identifies return signals by time slot and compares the pulse amplitude relative to the steady return at each spacing.

Three wavelengths on one board

A rigid glass-epoxy printed circuit board carries surface-mount emitters with nominal wavelengths of 525, 660, and 940 nm. A constant-current driver pulses each emitter in a separate time slot so the detector records the returning light from one wavelength at a time.

Silicon photodiode with optical barrier

A broadband silicon photodiode is soldered to the printed circuit board and connected to a transimpedance amplifier. An opaque partition between the emitters and the photodiode blocks direct illumination while leaving a path for light scattered back from the tissue.

Arithmetic mean of pulse rates

The controller subtracts an emitter-off reading from each illuminated reading and demultiplexes the corrected samples by wavelength. It estimates a pulse rate from the beat intervals in each wavelength channel and calculates the arithmetic mean of those rates.

Signal quality weighted pulse rate

The controller estimates a pulse rate separately for each wavelength over a common eight-second window. It assigns each estimate a weight proportional to its pulse-band signal-to-noise ratio on a linear scale and normalizes the weights to sum to one before calculating the weighted pulse rate.

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 2017/0325698 A1
Published
Nov 16, 2017
Filed
May 10, 2017
Class
A61B 5/0261
Document BUS 2021/0330285 A1
Published
Oct 28, 2021
Filed
Apr 27, 2021
Class
A61B 8/0883
Claim 1
Document A
Document B
1.1a circuit comprising a plurality of light emitting diode (LED) emitters connected to a printed circuit board, the circuit configured to determine a volumetric variation of blood circulation of a patient, each of the plurality of LED emitters configured to emit electromagnetic radiation having a nominal wavelength, the nominal wavelength for each of the plurality of LED emitters being different from a remaining of the plurality of LED emitters;
For example, light emitter 306 can be configured to emit light in the wavelength range of 495-570 nm, and signal 350a can represent the pulsatile blood volume changes of the user. Light emitter 308 can be configured to emit light in the wavelength range of 750-1400 nm, and the reflected light (e.g., light 325) can represent noise.
A · US 2017/0325698 A1 · ¶ 74
1.2one or more photodetectors, each of the one or more photodetectors electronically connected to the printed circuit board, wherein the one or more photodetectors is configured to capture light signals from each of the plurality of LED emitters; and
In some examples, light sensor 507 can be configured to detect wavelengths of light emitted by light emitter 506 and wavelengths of light emitted by light emitter 505.
A · US 2017/0325698 A1 · ¶ 86
1.3a controller in electronic communication with the plurality of LED emitters and the one or more photodetectors, the controller configured to: receive light signal data from each of the plurality of LED emitters through the one or more photodetectors; calculate a plurality of health parameters, each of the plurality of health parameters associated with one of the plurality of LED emitters; and calculate a predicted health parameter based on one of an average or a weighted average of the plurality of health parameters.
Controller 509 can receive one or more signals that include pulsatile blood volume changes (e.g., signals, such as signal 550, from light sensor 504) and one or more signals that includes little-to-no pulsatile blood information (e.g., signals, such as signal 555, from light sensor 507) and can apply one or more algorithms to determine the user's physiological signal.
A · US 2017/0325698 A1 · ¶ 84
The machine learning circuitry is configured to: receive a sequence of ultrasound images of a patient; receive prior knowledge associated with an anatomical structure; and automatically estimate one or more physiological parameters associated with a structure observed in the sequence of the ultrasound images of a patient based at least in part on the prior knowledge.
B · US 2021/0330285 A1 · ¶ 7

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

Statements in these documents
Given that a user's motion may not be limited to gross, periodic motion, a PPG system capable of differentiating pulsatile blood volume changes from anatomical motion can be desired.
A · US 2017/0325698 A1 · ¶ 4
Motion artifacts included in the signals can be canceled or reduced by measuring non-pulsatile blood volume changes and adjusting the signal to account for the non-pulsatile blood information.
A · US 2017/0325698 A1 · ¶ 5
However, ultrasound image sequences typically contain temporal information that is relevant to the desired measurement that is not utilized by ML algorithms or ultrasound users.
B · US 2021/0330285 A1 · ¶ 4
Feeding an ultrasound image sequence into the machine learning circuitry or ML-based workflows allows for incorporating prior information, such as periodicity for cardiac ultrasound data, which facilitates significant improvements in the performance of machine learning-based workflows.
B · US 2021/0330285 A1 · ¶ 5
Cite aspatent.fyi, US 2026/0354698 A1, claim 1, first published 2026-11-19T02:03:35-05:00, https://patent.fyi/US20260354698/, SHA-256 4f761ec7655388876c194878f4a31480c0414c514e08c11ca92d829a1f85c90c
First published
Permanent addresshttps://patent.fyi/US20260354698/
Content fingerprint (SHA-256)4f761ec7655388876c194878f4a31480c0414c514e08c11ca92d829a1f85c90c
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi