# US 2025/0375141 A1 · Apparatus and method for adaptive noise detection in wearable devices

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

- Number: US 2025/0375141 A1
- Document ID: US20250375141A1
- Status: Published application
- Kind: A1
- Application number: 18/736,186
- Filed: 2024-06-06
- Published: 2025-12-11
- Earliest claimed: 2024-06-06
- Assignees: Anumana, Inc.
- Applicants: Anumana, Inc.
- Inventors: Sairam Bade; Rakesh Barve; Shashi Kant; Shiva Verma; Shayan Ghosh
- Web page: https://patentgrep.com/patent/US20250375141A1

## CPC

- A61B 5/318: Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B 5/7221: {Determining signal validity, reliability or quality (preventing, reducing or removing noise induced by motion artefacts A61B5/7207; noise originating from a therapeutic or surgical apparatus A61B5/7217)}
- A61B 5/256: Wearable electrodes, e.g. having straps or bands
- A61B 5/7203: {for noise prevention, reduction or removal}
- A61B 5/725: {using specific filters therefor, e.g. Kalman or adaptive filters (specific diagnostics methods using using bioelectric or biomagnetic signals A61B5/316)}
- A61B 5/7264: {Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems}
- A61B 5/7267: {involving training the classification device}
- A61B 5/742: {using visual displays (displays for heart-related electrical signals, e.g. ECG, A61B5/339)}
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)

## Abstract

An apparatus and method for adaptive noise detection in wearable devices. The apparatus includes at least a physiological signal input channel configured to receive a physiological signal from a subject. The apparatus for adaptive noise detection in wearable devices further includes an adaptive noise detector communicatively connected to the at least a physiological signal input channel, wherein the adaptive noise detector further includes a signal characteristic model configured to generate a signal characteristic profile based on the physiological signal using profile training data, a signal output datapath, and a decision block.

## Claims

### Claim 1

An apparatus for adaptive noise detection in wearable devices,
wherein the apparatus comprises:
a sensor, wherein the sensor is configured to receive a physiological signal from a subject;
at least a processor;
a memory communicatively connected to the at least a processor, wherein the memory contains instructions configuring the at least a processor to receive the physiological signal and a profile training data set, wherein the profile training data set comprises outputs correlated to inputs, wherein the inputs comprise a plurality of physiological signal data and the outputs comprise a plurality of signal characteristic profile data;
at least a physiological signal input channel configured to receive the physiological signal;
a dedicated hardware unit, communicatively connected to the at least a processor, configured to sanitize the profile training data set, wherein the dedicated hardware unit comprises circuitry configured to perform signal processing operations, wherein sanitizing the profile training data set comprises:
determining by the dedicated hardware unit that at least one training data entry of the profile training data set has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the profile training data set to create a sanitized profile training data set;
an adaptive noise detector communicatively connected to the at least a physiological signal input channel, wherein the adaptive noise detector comprises a signal characteristic model which is configured to:
receive the sanitized profile training data set
train, iteratively, the signal characteristic model using the sanitized profile training data set, wherein training the signal characteristic model includes retraining the signal characteristic model with previous results of the signal characteristic model; and
generate a signal characteristic profile as a function of the physiological signal using the trained signal characteristic model;
a signal output datapath, wherein the signal output datapath receives the signal characteristic profile and is displayed through a graphical user interface, wherein a user may interact with the signal characteristic profile; and
a decision block operably connected to the at least a processor, wherein the decision block in combination with the at least a processor is configured to:
determine, based on the signal characteristic profile, that the physiological signal is within a quality tolerance, wherein the quality tolerance is a probability threshold based on a distribution of a plurality of physiological signals; and
transmit a result of whether the physiological signal is within the quality tolerance using the signal output datapath.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the sensor is communicatively connected to the at least a physiological signal input channel.

### Claim 3 (depends on 1)

The apparatus of claim 1, the apparatus is further configured to determine a frequency profile and then generate the signal characteristic profile using the frequency profile of the physiological signal.

### Claim 4 (depends on 1)

The apparatus of claim 1, the apparatus is further configured to:
detect a first signal from the at least a physiological signal input channel, wherein the first signal is not within the quality tolerance;
detect a second signal from the at least a physiological signal input channel, wherein the second signal is within the quality tolerance.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein the apparatus further comprises an initial signal processing module and a 12-lead database to filter high frequencies from the physiological signal using a low pass filter.

### Claim 6 (depends on 5)

The apparatus of claim 5, wherein the low pass filter removes a predictable noise element.

### Claim 7

(canceled)

### Claim 8 (depends on 1)

The apparatus of claim 1, wherein the signal characteristic model comprises a statistical model.

### Claim 9 (depends on 8)

The apparatus of claim 8, wherein the statistical model comprises a machine learning model.

### Claim 10 (depends on 1)

The apparatus of claim 1, wherein the at least a processor is configured to conditionally display, using the graphical user interface, a notification as a function of the signal characteristic profile.

### Claim 11

A method for adaptive noise detection in wearable devices, wherein the method comprises:
receiving, using a sensor, a physiological signal from a subject;
receiving, using a processor, the physiological signal and a profile training data set, wherein the profile training data set comprises outputs correlated to inputs, wherein the inputs comprise a plurality of physiological signal data and the outputs comprise a plurality of signal characteristic profile data;
sanitizing, using a dedicated hardware unit communicatively connected to the processor, the profile training data set, wherein the dedicated hardware unit comprises circuitry configured to perform signal processing operations, wherein sanitizing the profile training data set comprises:
determining by the dedicated hardware unit that at least one training data entry of the profile training data set has a signal to noise ratio below a threshold value; and
removing the at least one training data entry from the profile training data set to create a sanitized profile training data set;
receiving, using at least a physiological signal input channel, the physiological signal from the subject;
generating, using an adaptive noise detector, a signal characteristic profile based on the physiological signal, wherein the adaptive noise detector comprises a signal characteristic model, wherein generating the signal characteristic profile comprises:
receiving the sanitized profile training data set;
training, iteratively, the signal characteristic model using the sanitized profile training data set, wherein training the signal characteristic model includes retraining the signal characteristic model with previous results of the signal characteristic model; and
generating a signal characteristic profile as a function of the physiological signal using the trained signal characteristic model;
receiving, at a signal output datapath, the signal characteristic profile;
displaying, through a graphical user interface, the signal output datapath, wherein a user may interact with the signal characteristic profile;
determining, using a decision block communicatively connected to the processor, that the physiological signal is within a quality tolerance, wherein the quality tolerance is a probability threshold based on a distribution of a plurality of physiological signals; and
transmitting, using the signal output datapath, a result of whether the physiological signal is within the quality tolerance.

### Claim 12 (depends on 11)

The method of claim 11, further comprising communicatively connecting the sensor to the at least a physiological signal input channel, the physiological signal.

### Claim 13 (depends on 11)

The method of claim 11, further comprising determining a frequency profile and then generating the signal characteristic profile using the frequency profile of the physiological signal.

### Claim 14 (depends on 11)

The method of claim 11, further comprising:
detecting a first signal from the at least a physiological signal input channel, wherein the first signal is not within the quality tolerance;
detecting a second signal from the at least a physiological signal input channel, wherein the second signal is within the quality tolerance.

### Claim 15 (depends on 11)

The method of claim 11, further comprising filtering, using an initial signal processing module and a 12-lead database, high frequencies from the physiological signal using a low pass filter.

### Claim 16 (depends on 15)

The method of claim 15, wherein the low pass filter removes a predictable noise element.

### Claim 17

(canceled)

### Claim 18 (depends on 11)

The method of claim 11, wherein the signal characteristic model comprises a statistical model.

### Claim 19 (depends on 18)

The method of claim 18, wherein the statistical model comprises a machine learning model.

### Claim 20 (depends on 11)

The method of claim 11, wherein the method further comprises conditionally displaying, using the processor and the graphical user interface, a notification as a function of the signal characteristic profile.
