Systems and methods for vascular health assessment using acoustic analysis and machine learning

Abstract

A method is provided. The method comprises receiving acoustic data from at least one transducer device positioned proximate to an artery. The method comprises processing the received acoustic data using signal analysis techniques. The processing comprises generating frequency domain representations of the acoustic data. The method comprises partitioning the frequency domain representations into a plurality of frequency bins. The method comprises analyzing the frequency domain representations using at least one machine learning model. The analyzing comprises determining a degree of arterial stenosis based on the frequency domain representations. The method comprises outputting an indication corresponding to the determined degree of arterial stenosis. The frequency domain representations may be generated using a frequency domain analysis of the acoustic data. The at least one machine learning model may be trained using a dataset comprising acoustic recordings associated with known stenosis levels.

Description

WSGR Docket No. 69145-701.601SYSTEMS AND METHODS FOR VASCULAR HEALTH ASSESSMENT USING ACOUSTIC ANALYSIS AND MACHINE LEARNING CROSS-REFERENCE

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 747,509, filed January 21, 2025, and Polish Patent Application No. P.450991, filed January 20, 2025, of which are incorporated herein by reference.BACKGROUND

[0002] Strokes are often identified only after they happen, meaning many individuals first learn of their underlying cerebrovascular disease during the actual stroke event. According to the Centers for Disease Control and Prevention (CDC), up to 80% of these events could be prevented if their early warning signs were detected. Unfortunately, the diagnostic tools currently used to assess stroke risk - such as carotid ultrasounds or more advanced imaging techniques (e.g., Magnetic Resonance Angiography or Computed Tomography Angiography) - lack the specificity needed to catch subtle changes in plaque development and tend to be available only in hospital or specialized clinical settings. As a result, there may be lengthy intervals between assessments. Patients deemed to have moderate stroke risk may simply be told to “watch and wait,” returning for another check-up in six months or a year. This prolonged period of uncertainty may be highly stressful and leaves a critical gap in monitoring the progression of cerebrovascular disease.

[0003] Atherosclerotic plaque buildup is central to stroke risk, yet this process typically goes undetected until it is far advanced. In many cases, symptoms do not appear until the carotid artery is significantly blocked - often around 80% occlusion. Moreover, plaque accumulation is both systemic and incremental, unfolding over months or years, and it may “remodel” itself in ways unique to each individual. Softer, lipid-rich plaques have a higher likelihood of rupturing than harder, calcified plaques, which underscores the importance of understanding the plaque’s structure and composition. However, standard imaging methods often struggle to make these distinctions. Without more frequent or detailed assessments, many patients remain in the dark about the real-time progression of their disease, further magnifying the psychological and clinical challenges of living with moderate stroke risk.WSGR Docket No. 69145-701.601SUMMARY

[0004] Recognized herein is a need for robust and accessible systems and methods for personalized stroke risk monitoring. Current practices do not sufficiently address the incremental nature of atherosclerotic plaque development in the carotid arteries, nor do they offer timely detection and assessment outside of clinical settings. As a result, individuals with moderate or evolving stroke risk remain underserved, which may lead to missed opportunities for early intervention and heightened patient anxiety.

[0005] The present disclosure provides systems and methods for non-invasive vascular health assessment that utilize acoustic analysis, signal processing, and machine learning to monitor plaque development in the carotid arteries. In an aspect, the present disclosure provides a method for monitoring arterial health. In some cases, the method may comprise receiving physiological sound data from an mechanical transducer device positioned proximate to an artery. In some cases, the method may comprise processing the physiological sound data to generate acoustic frequency or time domain data. In some cases, the method may comprise analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a measure of stenosis in the artery. In some cases, the method may comprise outputting an indication of the determined measure of stenosis in the artery. In some cases, the physiological sound data is infrasound or audible range acoustic vibration data. In some cases, acoustic analysis, signal processing, and machine learning may be used to generate a plaque development risk assessment. In some cases, the method may further comprise analyzing a plurality of changes in the acoustic frequency or time domain data over a period of time. In some cases, the method may further comprise predicting future stenosis progression based on the plurality of changes over the period of time. In some cases, the method may further comprise generating one or more risk assessments based on the predicted future stenosis progression. In some cases, the method may further comprise detecting systolic cycles within the acoustic frequency or time domain data. In some cases, the method may further comprise windowing the acoustic frequency or time domain data prior to generating the acoustic frequency or time domain data. In some cases, the method may further comprise receiving environmental noise data from at least one secondary mechanical transducer device. In some cases, the method may further comprise applying noise reduction to the acoustic frequency or time domain data based on the environmental noise data. In some cases, the method may further comprise receiving positioning data indicating a position of the mechanical transducer device relative to the artery. In some cases, the method may further comprise providing feedback for adjusting the position of the mechanical transducer device based on the positioning data. In someWSGR Docket No. 69145-701.601cases, the method may further comprise analyzing amplitude, frequency content, and noise levels of the physiological sound data to determine a signal quality. In some cases, when the signal quality falls outside a specified threshold, the method may comprise transmitting pressure fluctuations to the mechanical transducer device using a pressure sensor. In some cases, the method is performed at home or outside of a healthcare environment.

[0006] In an aspect, the present disclosure provides a system for monitoring arterial health. In some cases, the system may comprise a housing, and at least one mechanical transducer device configured to capture physiological sound signals. In some cases, the system is configured to be placed proximate to an artery of a subject. In some cases, the system further comprises a computer-readable medium storing instructions that, when executed by a computer, cause it to perform: (a) processing the physiological sound signals to generate acoustic frequency or time domain data; (b) analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a degree of stenosis in the artery; and (c) outputting an indication of a determined degree of arterial stenosis. In some cases, the at least one mechanical transducer device collects infrasound or audible range acoustic vibration data. In some cases, the housing comprises an acoustic membrane configured to contact the subject’s skin. In some cases, the acoustic membrane is configured to capture infrasonic frequency vibrations below 20 Hz. In some cases, the acoustic membrane is configured to capture audible range acoustic vibrations. In some cases, the system further comprises a supplementary sensor assembly configured to detect applied contact pressure. In some cases, the system further comprises a digital signal processing (DSP) unit configured to extract time or frequency information indicative of vascular flow characteristics from the physiological sound signals. In some cases, the system further comprises an interface module configured to display real-time visualizations of the physiological sound signals. In some cases, the system further comprises a data integration platform. In some cases, the data integration platform is configured to synchronize the analyzed acoustic frequency or time domain data with electronic health records (EHRs). In some cases, the system further comprises a server architecture. In some cases, the server architecture comprises pre-trained machine learning or deep learning models configured to perform stenosis analysis. In some cases, the system further comprises a layered hardware architecture configured to allow upgrades or replacements of modules without affecting the entire system. In some cases, the system further comprises a data validation system configured to flag poor-quality measurements. In some cases, the system further comprises a data security assembly configured to perform encryption of subject health information. In some cases, the system further comprises a telemedicine interfaceWSGR Docket No. 69145-701.601configured to allow real-time consultation with a healthcare provider based on the determined measure of arterial stenosis. In some cases, the system is handheld. In some cases, the system is portable.

[0007] In some embodiments, the systems and methods disclosed herein may comprise an at-home, self-testing device configured to capture infrasound and low-frequency audio signals from the carotid region using at least one mechanical transducer. The at-home, self-testing device may be hand-held. The at-home, self-testing device may be portable. In some cases, the systems and methods disclosed herein may include a multi-sensor array with a primary transducer for bloodflow detection and supplemental sensors (e.g., pressure sensors, noise-cancellation microphones) to optimize data collection. In some instances, the systems and methods disclosed herein may encompass feedback mechanisms (visual or haptic) to guide users on correct device placement. As an example, the systems and methods disclosed herein may comprise data-processing software employing a frequency domain analysis (e.g., Fast Fourier Transform (FFT), Wavelet transform, Goertzel Algorithm, etc.) to decompose the recorded signals into distinct frequency bins, generating “fingerprints” correlated with varying degrees of carotid stenosis. In some cases, time domain analysis may be used.

[0008] In some embodiments, the systems and methods disclosed herein may comprise a processing module configured to preprocess acquired physiological sound signals from the carotid artery. Information indicative of vascular conditions may be extracted from the physiological signals. The systems and methods may further comprise a classification module, which may comprise a machine-learning or deep-learning model trained to associate the extracted information with one or more atherosclerotic plaque levels. In some cases, the classification model may be trained to map said “fingerprints” to specific levels of atherosclerotic plaque buildup. In some cases, the systems and methods disclosed herein may include real-time or near real-time classification of plaque severity, providing users and clinicians with immediate risk assessments. In some instances, the systems and methods disclosed herein may implement longitudinal data tracking to observe how plaque morphology changes over months or years, thereby offering a more comprehensive stroke-risk profile. As an example, the systems and methods disclosed herein may incorporate cloud-based analytics that aggregate, store, and update user data to generate personalized risk trends.

[0009] In some embodiments, the systems and methods disclosed herein may include noise mitigation techniques, such as active noise cancellation or various signal filtering methods (e.g. spectral subtraction), to enhance signal quality in clinical or non-clinical environments. In some cases, the systems and methods disclosed herein may provide a user interface (e.g., via aWSGR Docket No. 69145-701.601smartphone application or dedicated software platform) for visualizing an output indicative of vascular health, such as plaque levels or progression, and receiving personalized recommendations. In some instances, the systems and methods disclosed herein may be configured to integrate additional biometric data, such as heart rate, blood pressure, or cholesterol levels, to refine stroke-risk algorithms. As an example, the systems and methods disclosed herein may incorporate remote clinical collaboration features, allowing healthcare professionals to review real-time or stored patient data and offer timely interventions. As another example, systems and methods herein may be able to detect bruits from various locations around the body. The methods herein may be capable of identifying locations of bruits around the body and the corresponding clinical meaning such as bruits of carotid arteries, cervical veins or arteriovenous (AV) connections, aorta, renal arteries, iliac arteries, hepatic artery, and splenic artery.INCORPORATION BY REFERENCE

[0010] All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference. To the extent publications and patents or patent applications incorporated by reference contradict the disclosure contained in the specification, the specification is intended to supersede and / or take precedence over any such contradictory material.BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings (also “Figure” and “FIG.” herein), of which:

[0012] FIG. 1 shows an example of a system for monitoring arterial health, in accordance with some embodiments.

[0013] FIG.2 shows an example of an apparatus for monitoring arterial health, in accordance with some embodiments.

[0014] FIG.3 shows an example of a device for monitoring arterial health, in accordance with some embodiments.

[0015] FIG.4 shows an example of data processing workflow for monitoring arterial health, in accordance with some embodiments.

[0016] FIG. 5 shows a computer system, in accordance with some embodiments.WSGR Docket No. 69145-701.601DETAILED DESCRIPTION

[0017] While various embodiments of the invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions may occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed.SYSTEM

[0018] The present disclosure provides systems and methods for monitoring arterial health. In some embodiments, the system comprises at least one housing. In some cases, the system comprises at least one mechanical transducer device configured to capture physiological sound signals. In some cases, the system is configured to be placed proximate to an artery of a subject. In some cases, the system further comprises a computer-readable medium storing instructions that, when executed by a computer, cause it to perform: (a) processing the physiological sound signals to generate acoustic frequency or time domain data; (b) analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a degree of stenosis in the artery; and (c) outputting an indication of a determined degree of arterial stenosis. In some cases, the at least one housing comprises at least one acoustic membrane configured to contact a patient’s skin. In some cases, there may be a plurality of membranes optimized for specific frequencies. Alternatively, the device or system may not include a membrane. In some instances, the acoustic membrane is optimized to capture low-amplitude vibrations associated with blood flow. In some cases, the at least one housing comprises a detachable membrane assembly. In some instances, the detachable membrane assembly may be configured for replacement or sterilization. For example, the at least one acoustic membrane may be configured to capture infrasonic frequency vibrations below 20 Hz. The acoustic membrane may be configured to capture audible range acoustic vibrations. As an example, the membrane is configured to enhance signal quality by transmitting subtle pressure fluctuations to the system’s transducers. As an example, the housing may include an acoustic shield to help reduce environmental noise.Transducers and Infrasonic Detection

[0019] In some embodiments, the system comprises at least one transducer assembly. In some cases, the at least one transducer assembly comprises a plurality of transducers configured to detect acoustic signals in different frequency ranges. In some instances, the transducer assembly includes at least one infrasonic transducer for capturing sub-20 Hz signals and at least one acoustic transducer for capturing audible-range frequencies. In some cases, the plurality ofWSGR Docket No. 69145-701.601transducers is coupled to signal conditioning circuitry. In some instances, the signal conditioning circuitry is configured to amplify, filter, or otherwise process the electrical signals. For example, the system may be configured to convert mechanical vibrations into digital data via at least one low-noise analog-to-digital converter. In some cases, the A / D converter is sampling a transducer with multiple channels simultaneously to achieve a higher signal to noise ratio. As an example, these transducers may be arrayed in a multi-channel audio capture system, where one or more of the channels focus on blood flow signals and another on ambient noise.Noise Reduction

[0020] In some embodiments, the system comprises at least one noise reduction system. In some cases, the at least one noise reduction system comprises at least one environmental noise sensor and an acoustic shield surrounding the transducers. In some instances, the noise reduction system is configured to implement active noise cancellation using phase-inversion techniques or passive noise cancellation through acoustic shielding. In some cases, the system comprises signal post processing techniques. The signal post processing techniques may comprise, for example, spectral subtraction modules and other filtering methods. In some instances, these modules are configured to remove ambient or mechanical noise from the recorded signals. For example, the system may be configured to use a secondary microphone to capture environmental noise separately from primary blood flow signals. In secondary microphone may be a set of microphones. As an example, adaptive filtering may be configured to respond to changing noise environments and improve signal-to-noise ratio.Positioning and User Feedback

[0021] In some embodiments, the system comprises at least one positioning system. In some cases, the positioning system comprises proximity sensors or orientation sensors for guiding device placement on a carotid artery. In some cases, the sensor for guiding placement may be any suitable type of sensor such as optical sensors or built-in cameras for positioning. In some instances, the positioning system comprises at least one force sensor to detect contact force between the membrane and the patient’s skin. In some cases, the system comprises at least one user feedback system. In some instances, the user feedback system includes haptic or visual indicators. For example, the system may be configured to provide real-time vibration cues or onscreen prompts to help position the device properly. The prompts or feedback may be provided via a user interface of a mobile application or a software application. As an example, this proprietary positioning approach is configured to ensure consistent device alignment over repeated measurements.Supplementary SensorsWSGR Docket No. 69145-701.601

[0022] In some embodiments, the system comprises at least one supplementary sensor assembly. In some cases, the at least one supplementary sensor assembly comprises at least one force sensor for detecting applied contact pressure. The pressure measurement can be used to ensure reliable measurement conditions. In some instances, the assembly comprises at least one position sensor for detecting angular orientation or device tilt. In some cases, the supplementary sensor assembly may be configured to optimize measurement conditions by communicating calibration data to the processing unit. In some instances, these sensors may work in tandem with the positioning system to minimize external factors and ensure consistent measurements. For example, the supplementary sensors may be configured to provide real-time feedback regarding suboptimal device placement. As an example, the system may automatically adjust data-collection parameters (e.g., signal amplifier gain) based on sensor readings.Signal Acquisition and Digital Signal Processing

[0023] In some embodiments, the system comprises at least one signal acquisition assembly. In some cases, the at least one signal acquisition assembly comprises a transducer array coupled to at least one analog-to-digital converter. In some instances, the signal acquisition assembly is configured to capture pressure fluctuations from the membrane and convert them into digitized signals. In some cases, the system comprises at least one digital signal processing (DSP) unit. In some instances, the DSP unit may be configured to perform DSP operations such as filtering and ratification reduction. The DSP unit may be configured to extract time and / or frequency information indicative of vascular flow characteristics from the physiological signals. In some cases, the DSP may be based on a chain of methods applied to the recorded signal to maximize the amount and quality of useful information that can be used for prediction. For example, the DSP unit may focus on systolic cycles of the heartbeat for more precise plaque assessment. As an example, the DSP unit may implement adaptive noise filtering to generate more accurate data for stenosis characterization.Machine Learning and Processing

[0024] In some embodiments, the system comprises one or more processors. In some cases, the one or more processors may be configured to receive acoustic data from multiple transducers. The acoustic data may comprise time or frequency-domain representations. In some instances, the processor comprises at least one machine learning or deep learning model configured to analyze extracted signal components to generate an indication of arterial stenosis. . In some cases, some of the data processing and analysis may be performed by a processor located on the device while others are performed on a cloud computing environment. For example, the device's microcontroller may capture the signal, and the ML-based prediction may be either performed onWSGR Docket No. 69145-701.601cloud by receiving the processed data transmitted from the device or performed on the device (e.g., edge-computing architecture). In some cases, the model is configured to integrate additional patient-specific metrics (e.g., heart rate, blood pressure, BMI, cholesterol levels, or cardiac event history). In some instances, the system may achieve improved accuracy using a deep learning network optimized with an optimization algorithm, for example, the Adam algorithm. For example, the processor may be configured to map newly captured fingerprints to known stenosis levels. As an example, the machine learning model may be updated over time via a secure server to improve prediction accuracy.Communication Modules

[0025] In some embodiments, the system comprises at least one communication module. In some cases, the communication module comprises wireless capabilities (e.g., Wi-Fi, Bluetooth) to transmit acoustic data to external systems. In some instances, the communication module is configured to receive patient-metric data from wearable devices or external health record systems. In some cases, the communication module may be configured to synchronize analysis results with electronic health records (EHRs). In some instances, data is transmitted in encrypted form to a secure cloud-based analysis platform. For example, the communication module may be configured to maintain a log of transmissions for auditability. As an example, the system may queue data for retransmission in case of connectivity issues.Interface Modules

[0026] In some embodiments, the system comprises at least one interface module. In some cases, the interface module is configured to display real-time visualizations of acoustic measurements. In some instances, the interface module may include multi-level access controls, providing different views for patients and healthcare providers. In some cases, the interface module may present trend analyses of stenosis measurements over time. In some instances, the module may guide the user through interactive tutorials for proper device operation. For example, the interface module may generate on-screen instructions for device positioning. As an example, the interface may display comparative analyses between a patient’s current and past measurements. In some cases, the interface module may comprise a computer or mobile application.Redundancy and Backup

[0027] In some embodiments, the system comprises at least one redundant processing pathway. In some cases, the system includes local storage (e.g., non-volatile internal memory) configured to store acoustic data. In some instances, the system comprises backup power circuitry for uninterrupted operation. In some cases, the system may be configured to detect and recover from pipeline failures through error recovery procedures. In some instances, the redundancy ensuresWSGR Docket No. 69145-701.601data integrity and continuity if a primary processing unit malfunctions. For example, the system may automatically switch to a secondary DSP or processor. As an example, the system logs all failover events for later diagnostics.Software Management

[0028] In some embodiments, the system comprises at least one software management system. In some cases, the software management system is configured to receive and verify secure software updates using cryptographic signatures. In some instances, the system maintains version control of all processing algorithms. In some cases, the system implements rollback procedures in the event of a failed update. In some instances, the system may dynamically adjust processing parameters based on real-time signal -quality metrics. In some cases, the system (e.g., software application) may modify noise-filtering thresholds or suggest taking the measurements in a more suitable acoustic environment if it detects excessive ambient noise. As an example, the system logs update histories for regulatory or quality-assurance purposes.Data Integration and Server Architecture

[0029] In some embodiments, the system comprises at least one data integration platform. In some cases, the platform is configured to synchronize analysis results with electronic health records (EHRs) and support telemedicine interfaces. In some instances, the platform supports automated backups to maintain data integrity. In some cases, the system comprises at least one server architecture. In some instances, the server architecture includes pre-trained machine learning or deep learning models for advanced stenosis analysis. For example, the server may be configured to perform distributed processing of multiple patient recordings simultaneously. As an example, the server architecture may continually update a proprietary database of plaque development patterns.Calibration

[0030] In some embodiments, the system comprises at least one calibration subsystem. In some cases, the calibration subsystem is configured to perform periodic self-tests of the transducers. In some instances, it may adjust gain settings based on environmental conditions. In some cases, the calibration subsystem includes an internal signal generator for sensitivity checks. In some instances, the subsystem may compensate for temperature variations affecting the transducers. For example, the system may automatically recalibrate if the ambient temperature exceeds a predefined range. As an example, calibration logs may be maintained for quality assurance. Data Collection Device

[0031] In some embodiments, the system comprises at least one data collection device. In some cases, the data collection device includes an acoustic assembly configured to non-invasivelyWSGR Docket No. 69145-701.601capture audio signals from the carotid arteries. In some instances, the data collection device houses microphones, noise reduction subsystems, and force sensors and / or other sensors. In some cases, the data collection device may be modular, allowing components to be swapped (for their variants optimized for specific types of measurements) or upgraded. In some instances, the device captures measurements for real-time processing or offline storage. For example, the device may store raw recordings until they are transferred to a host application. As an example, the data collection device may include a power management module for portable operation. Data Storage, Transmission, and Management

[0032] In some embodiments, the system comprises at least one data storage and transmission component. In some cases, the data storage component comprises internal memory for buffering raw acoustic recordings and data from supplementary sensors. In some instances, the data storage and transmission component may implement secure wireless transmission protocols to a mobile or desktop application. In some cases, the system comprises at least one data management assembly configured to build and maintain a database of frequency-based stenosis fingerprints. In some instances, the data management assembly is configured to integrate complementary patient metrics such as BMI or cholesterol levels. For example, the system may automatically append patient history data to each new acoustic recording. As an example, the data management assembly may generate reports or analytics for clinicians.

[0033] In some embodiments, data generated or utilized by the system may be stored locally on a user device, on a remote computing system, or in a distributed computing environment accessible via a network. Data stored on a remote or distributed computing system may be managed using data protection mechanisms, such as encryption and anonymization, to safeguard user privacy and data integrity.Longitudinal Monitoring

[0034] In some embodiments, the system comprises at least one longitudinal monitoring system. In some cases, this system includes at least one database for storing historical measurements of acoustic signals. In some instances, the database is configured to track changes in acoustic data over time. In some cases, the system may model plaque development patterns and predict future progression. In some instances, these longitudinal data points are used to alert clinicians when a patient’s risk profile changes significantly. For example, the system may compare newly captured signals with historical baselines to identify accelerating stenosis. As an example, trend analysis results may be displayed to both the patient and healthcare provider via the interface module. The algorithm may use previously learned information from other patients' journeys to generate advice.WSGR Docket No. 69145-701.601Power Management

[0035] In some embodiments, the system comprises at least one power management system. In some cases, the power management system includes at least one power supply unit configured to provide stable operation. In some instances, the system may incorporate a rechargeable battery or adopt a power monitoring module to display remaining battery life. In some cases, the power management system is configured to implement power optimization protocols for extended device operation. In some cases, the device may only sample when the device is in measurement mode such as upon a switch / press of a button on the device / in the application. An example of power management may include shutting of all currently unused peripherals and subsystems. In some instances, the system may enter a low-power state when idle. As an example, the power management system may send alerts for low battery conditions.Layered Hardware Architecture

[0036] In some embodiments, the system comprises at least one layered hardware architecture. In some cases, the layered hardware architecture employs a modular design that separates functionalities (e.g., transducer interface, signal processing, network communication) for efficient maintenance. In some instances, a microcontroller unit may manage data flow between layers. In some cases, this architecture is configured to allow upgrades or replacements of specific modules without affecting the entire system. In some instances, each hardware layer includes standardized interfaces for rapid prototyping. For example, the transducer layer may connect to the DSP layer via a well-defined, standardized communication bus. As an example, this modular architecture simplifies regulatory approvals by isolating new features.Data Validation, Diagnostics, and Predictive Analytics

[0037] In some embodiments, the system comprises at least one data validation system. In some cases, this system includes a quality assessment module configured to flag poor-quality measurements. In some instances, the system may prompt the user to recapture data if signal quality does not meet a preset threshold. In some cases, the system comprises at least one diagnostic assembly configured to perform self-checks on recording, positioning, data acquisition and supplementary sensor functionality. In some cases, the data validation system may employ classification neural network models to assess the quality of the captured signal. In some instances, the diagnostic assembly may provide real-time feedback regarding measurement validity. For example, the system may detect improper device alignment and prompt a repositioning message.Data Security and AccessibilityWSGR Docket No. 69145-701.601

[0038] In some embodiments, the system comprises at least one data security assembly. In some cases, the data security assembly implements encryption for all transmissions of patient health information. In some instances, it supports multi-factor authentication and audit logging of access attempts. In some cases, the system is configured to be HIPAA-compliant or to adhere to equivalent data protection standards. In some instances, the data security assembly comprises a data integrity verification module to detect unauthorized modifications. For example, the system may implement cryptographic checksums for each recorded file. As an example, security updates may be distributed through the software management system.

[0039] In some embodiments, the system comprises at least one accessibility assembly. In some cases, it includes a multilingual interface or adaptive feedback module for users with varying abilities. In some instances, it may provide configurable text sizes, contrast settings, or audible prompts. For example, the user may configure voice guidance for device placement if visual cues are insufficient. As an example, the accessibility assembly may integrate with screen readers or other assistive devices.Remote Monitoring and Customization

[0040] In some embodiments, the system comprises at least one remote monitoring assembly. In some cases, this assembly includes a telemedicine interface configured to permit real-time consultation with healthcare providers. In some instances, the remote monitoring assembly may implement alert management to notify providers of critical findings. In some cases, data streaming modules are configured to transmit diagnostic information securely in asynchronous or synchronous modes. In some instances, the system may configure direct feedback loops between the patient and clinician via online platforms.

[0041] In some embodiments, the system comprises at least one customization assembly. In some cases, the customization assembly includes a threshold configuration module for adjusting diagnostic parameters. In some instances, the customization assembly comprises a patient profile module to store individual measurement preferences. For example, certain clinical protocols may override default settings for patients with unique risk factors. As an example, these custom settings may be consistently applied to ensure reproducible measurements across sessions.Error Handling and Training Data

[0042] In some embodiments, the system comprises at least one error handling assembly. In some cases, this assembly includes an error classification module to categorize anomalies. In some instances, it may include a user notification module that provides context-specific troubleshooting guidance. In some cases, the assembly comprises an error recovery moduleWSGR Docket No. 69145-701.601configured to restore system stability after encountering issues. In some instances, the system maintains detailed error logs for retrospective analysis and improvement.

[0043] In some embodiments, the system comprises at least one training data assembly. In some cases, the training data assembly includes a data collection protocol module to standardize how patient recordings are gathered. In some instances, it comprises a consent management module for handling patient permissions. In some cases, the assembly includes a data anonymization module to remove personally identifiable information. In some instances, the training data assembly ensures demographic diversity in the dataset for robust machine learning model performance. For example, the system may integrate phantom artery data (with known stenosis levels) and real-patient data to train or retrain the ML models.

[0044] In some embodiments, the system comprises at least one expansion capability system. In some cases, the expansion capability system includes an adaptable measurement module for other large vessel occlusions, such as femoral arteries. In some instances, the system may be configured for monitoring diabetic limb conditions where vascular blockage is a concern. In some cases, the system may be adaptable to assess coronary artery health with certain hardware or software modifications. In some instances, these expansion modules remain subject to feasibility and specialized calibration. For example, the system may employ different transducer geometries or flow models for coronary arteries. As an example, the data processing pipeline may be tailored to detect relevant signatures in those vessels.

[0045] FIG. 1 illustrates an example stroke-risk assessment system 100 that integrates both data-acquisition and data-processing components in a cohesive workflow. In particular, a device 101 is configured to non-invasively acquire acoustic data from a patient’s arteries by capturing low-amplitude vibrations associated with blood flow. This device may include multiple transducers — such as microphones for capturing different frequency ranges and background noise — along with supplementary sensors (for example, contact-pressure monitoring) to optimize signal quality and ensure patient safety.

[0046] A mobile or desktop application 102 manages data collection, user interaction, and partial signal processing. Within this application, several functional modules may be implemented. A Patient Metrics Acquisition module 103 gathers relevant parameters, such as blood pressure, cholesterol, and BMI, to supplement the recorded acoustic data. A Data Compilation and Prediction Request module 104 aggregates these signals and patient metrics, then transmits them to remote servers for advanced analysis. Meanwhile, a Measurement Assistance module 105 provides real-time guidance — using graphical or haptic feedback — to help operators or patientsWSGR Docket No. 69145-701.601optimize device placement. The results of the stenosis or plaque-buildup assessment are then displayed in the application by a Results Presentation module 106.

[0047] A server infrastructure 107, 108 performs the bulk of the signal processing and hosts one or more pre-trained machine learning models. By comparing incoming acoustic data and complementary patient metrics against known patterns of vascular stenosis, these servers may generate stenosis or plaque-buildup predictions and relay them back to the application 102 in real time or near-real time. In operation, system 100 coordinates the multi-sensor device 101, the application 102, and the servers 107, 108 to facilitate non-invasive arterial health assessments. This integration of acoustic signal capture, patient metrics, and machine-learning predictions provides a comprehensive way to track plaque accumulation and stroke risk over time.APPARATUS

[0048] The systems, methods and devices provided in the present disclosure may comprise an apparatus configured for monitoring arterial health.

[0049] In some embodiments, the apparatus comprises at least one acoustic monitoring device configured for at-home vascular assessment or inside of a clinical setting, for example to speed up time to diagnosis. In some cases, the at least one acoustic monitoring device comprises at least one bell-shaped enclosure having a modular configuration. In some instances, the at least one acoustic monitoring device comprises at least one primary microphone positioned to receive arterial blood flow signals. In some cases, the at least one primary microphone may be configured to detect signals transmitted through at least one membrane contacting a patient's skin surface. Alternatively, the device may not comprise a membrane.

[0050] In some embodiments, the apparatus may comprise at least one secondary microphone. In some cases, the at least one secondary microphone may be configured to capture ambient acoustic signals for noise reduction processing. In some instances, the apparatus may comprise at least one force sensor. In some instances, the at least one force sensor may be configured to measure a contact pressure value between the at least one membrane and the patient's skin surface.

[0051] In some embodiments, the apparatus may comprise at least one stacked electronics board assembly. In some cases, the at least one stacked electronics board assembly may comprise at least one wireless transceiver. In some instances, the at least one wireless transceiver may be configured to transmit digitized audio signals and other complementary data assisting with the measurements to at least one remote processing system.

[0052] In some embodiments, the apparatus may comprise at least one removable membrane sub-assembly. In some cases, the at least one removable membrane sub-assembly may beWSGR Docket No. 69145-701.601configured to permit replacement, cleaning, or sterilization. In some instances, the at least one removable membrane sub-assembly may be configured to maintain hygiene standards during repeated at-home usage.

[0053] In some embodiments, the apparatus may comprise at least one user feedback indicator. In some cases, the at least one user feedback indicator may be configured to provide at least one of a visual indication or an auditory indication. In some instances, the at least one user feedback indicator may be configured to activate when the contact pressure measured by the at least one force sensor deviates from at least one predefined threshold value, permitting consistent placement and data quality assessment.

[0054] FIG. 2 illustrates an example of an apparatus architecture 200 in which multiple subsystems are arrayed around a contact membrane 201 that lies against the patient’s skin 210 to capture vibrations from blood flow. In some cases, supplementary sensors 202 support usage assistance, calibration, and device positioning, helping ensure that the device is accurately aligned with the artery of interest. A noise reduction system 203 may actively or passively mitigate ambient sound, thereby improving the clarity of signals fed into an audio signal acquisition subsystem 204. In some instances, to further refine measurements, the apparatus includes a device positioning system 205 that locates the correct anatomical site and a user feedback system 206 that delivers real-time visual, auditory, or haptic cues. In some instances, data storage and transmission 207 components may record acquired signals locally, while connectivity 208 allows uploading data to a mobile / desktop application or other external platforms for advanced processing or remote review. Finally, external systems 209 may host machine learning algorithms, manage long-term data archives, or facilitate additional clinical evaluations. Altogether, this integrated architecture 200 combines carefully tuned sensing elements, positioning aids, noise reduction, and user-friendly feedback into a non-invasive device that supports accurate, real-time acoustic data collection and streamlined workflows for both patients and medical professionals.

[0055] FIG.3 shows an example of an apparatus 300 that may implement various acoustic sensing and environmental noise-reduction features. In some embodiments, the apparatus comprises a stethoscope bell and membrane 303 configured to capture low-amplitude vibrations from arterial blood flow. One or more main microphones 304 may be placed to detect these primary signals, while a separate microphone 305 may be used to record environmental noise for subsequent cancellation or filtering. In certain implementations, the apparatus may also include force sensing capabilities to ensure correct contact pressure against a patient’s skin. ThisWSGR Docket No. 69145-701.601arrangement may configure precise acoustic data collection, real-time noise handling, and consistent device positioning — facilitating a non-invasive approach to arterial health assessment.METHODS AND ALGORITHMS

[0056] The systems, methods and devices provided in the present disclosure may comprise methods configured for monitoring arterial health. In an aspect, the method may comprise receiving physiological sound data from a mechanical transducer device positioned proximate to an artery. In some cases, the method comprises processing the physiological sound data to generate acoustic frequency or time domain data. In some cases, the method comprises analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a measure of stenosis in the artery. In some cases, the method comprises outputting an indication of the determined measure of stenosis in the artery. In some cases, the physiological sound data is infrasound or audible range acoustic vibration data.

[0057] In some embodiments, the method may comprise recording vascular sounds from a patient using an acoustic monitoring device. In some cases, the method may comprise positioning the acoustic monitoring device on a patient's carotid artery to capture raw audio signals. In some instances, the method may comprise transmitting the captured raw audio signals to a remote processing system through a software application interface. In some instances, the method may comprise processing the raw audio signals through multiple stages of analysis. As an example, the method may comprise applying a windowing technique to isolate specific portions of the cardiac cycle within the recorded signals. For example, the method may comprise converting the windowed signals from a time-based representation to a frequency -based representation. As an example, the frequency domain signal data may be input into a machine learning model configured to assess vascular health parameters.

[0058] In some cases, the method further comprises analyzing a plurality of changes in the acoustic frequency or time domain data over a period of time. In some cases, the method further comprises predicting future stenosis progression based on the plurality of changes over the period of time. In some cases, the method further comprises generating one or more risk assessments based on the predicted future stenosis progression.

[0059] In some embodiments, the method may comprise signal conditioning. In some cases, the method may comprise adjusting the signal components to account for variations in recording conditions. In some instances, the method may comprise applying compensation factors to minimize the impact of ambient noise on the recorded signals. In some instances, the methodWSGR Docket No. 69145-701.601may comprise generating standardized signal representations suitable for machine learning analysis.

[0060] In some embodiments, the method may comprise real-time monitoring of device positioning parameters. In some cases, the method may comprise measuring the contact force between the acoustic monitoring device and the patient's skin surface. In some instances, the method may comprise analyzing the measured force values to determine if they fall within acceptable ranges for accurate signal acquisition. In some instances, the method may comprise generating immediate feedback through the software application when positioning adjustments are needed. As an example, the method may comprise providing guidance to the user to achieve improved device placement.

[0061] In some embodiments, the method may comprise receiving acoustic data from a transducer device positioned proximate to an artery. In some cases, the method may comprise processing the acoustic data to generate time and frequency components, where the processing may include partitioning the data into frequency bins. In some instances, the method may comprise analyzing the time and frequency components using at least one machine learning or deep learning model to determine a degree of arterial stenosis. In some instances, the method may comprise outputting an indication of the determined degree of arterial stenosis.

[0062] In some embodiments, the method may comprise detecting systolic cycles within the acoustic data. In some cases, the method may comprise windowing the acoustic data based on the detected systolic cycles prior to generating the time and frequency components. In some instances, the method may comprise normalizing the acoustic data and performing DSP operations such as filtering and ratification reduction on the normalized acoustic data.

[0063] In some embodiments, the method may comprise receiving environmental noise data from at least one secondary transducer. In some cases, the method may comprise applying noise reduction to the acoustic data based on the environmental noise data, where the noise reduction may include receiving noise measurements from an acoustic shield surrounding the transducer device and receiving noise measurements from an environmental microphone.

[0064] In some embodiments, the method may comprise receiving positioning data indicating a position of the transducer device relative to the artery. In some cases, the method may comprise providing feedback for adjusting the position of the transducer device based on the positioning data, where the feedback may include generating haptic feedback, displaying visual guidance, and providing audio cues confirming proper positioning.

[0065] In some embodiments, the method may comprise storing the acoustic data and determined degree of arterial stenosis. In some cases, the method may comprise receiving subsequentWSGR Docket No. 69145-701.601acoustic data at a later time and determining a subsequent degree of arterial stenosis. In some instances, the method may comprise generating a trend analysis based on changes between the determined degrees of arterial stenosis over time.

[0066] In some embodiments, the method may comprise receiving signals from multiple transducers configured to detect different frequency ranges. In some cases, the method may comprise combining the signals from the multiple transducers to generate composite acoustic data. In some instances, the method may comprise analyzing temporal changes in the acoustic data and predicting future stenosis progression based on the temporal changes.

[0067] In some embodiments, the method may comprise generating a personalized monitoring profile based on patient-specific characteristics. In some cases, the method may comprise adjusting analysis parameters of the machine learning model according to the monitoring profile. In some instances, the method may comprise dynamically updating the monitoring profile based on observed acoustic patterns, where updating may include analyzing historical acoustic data patterns and identifying patient-specific frequency characteristics.

[0068] In some embodiments, the method may comprise transmitting the acoustic data to a cloud-based analysis platform. In some cases, the method may comprise allowing collaborative review of the acoustic data by multiple authorized healthcare providers and receiving consolidated feedback from the healthcare providers. In some instances, the method may comprise implementing secure data transmission protocols and maintaining an audit trail of data access and analysis, to support data integrity, accountability, and user privacy.

[0069] In some embodiments, the method may comprise extracting feature sets corresponding to different plaque characteristics. In some cases, the method may comprise determining plaque morphology based on the extracted features and classifying plaque stability based on the determined morphology. In some instances, the method may comprise tracking changes in plaque morphology over time and generating risk assessments based on morphological changes.

[0070] In some embodiments, the method may comprise generating periodic reports of arterial health status. In some cases, the method may comprise adapting report frequency based on detected changes in stenosis levels and customizing report content based on recipient type. In some instances, the method may comprise implementing redundant data storage across multiple secure locations and verifying data integrity through periodic checksums.

[0071] In some embodiments, the method may comprise detecting signal quality metrics during data acquisition. In some cases, the method may comprise validating the acoustic data against predetermined quality thresholds and requesting repeated measurements when quality metrics fall below thresholds. In some instances, the method may comprise detecting anomalies in theWSGR Docket No. 69145-701.601acoustic data during processing and classifying detected anomalies based on predetermined error patterns.

[0072] In some embodiments, the method may comprise receiving lifestyle data comprising diet, exercise, and medication information. In some cases, the method may comprise correlating the lifestyle data with temporal changes in arterial stenosis and adjusting monitoring frequency based on identified lifestyle-related patterns. In some instances, the method may comprise identifying high-risk periods based on lifestyle patterns and increasing measurement frequency during identified high-risk periods.

[0073] In some embodiments, the method may comprise validating measurement procedures against regulatory standards and maintaining audit trails of all measurements and analyses. In some cases, the method may comprise generating compliance reports for regulatory submissions and implementing version control for algorithm updates. In some instances, the method may comprise implementing multiple independent analysis pathways and comparing results between analysis pathways for consistency.

[0074] In some embodiments, the method may comprise detecting changes in environmental conditions during measurement. In some cases, the method may comprise assessing the impact of environmental changes on measurement quality and adjusting processing parameters to compensate for environmental effects. In some instances, the method may comprise maintaining a quality assurance database of measurement conditions and correlating measurement quality with environmental and patient factors.Positioning and Acoustic Data Acquisition

[0075] In some embodiments, the method may comprise positioning a sensing device on a patient's skin proximate to a carotid artery. In some cases, the method may comprise receiving positioning feedback from at least one supplementary sensor to guide proper device placement. In some instances, the positioning feedback may comprise visual indicators, haptic feedback, or combinations thereof.

[0076] In some embodiments, the method may comprise acquiring acoustic data from the carotid artery. In some cases, the method may comprise recording acoustic signals during systolic cycles of the patient's heartbeat. In some instances, the method may comprise detecting environmental noise concurrent with the acoustic data acquisition. In some instances, the method may comprise applying noise reduction techniques to the acquired acoustic data using the detected environmental noise.WSGR Docket No. 69145-701.601Preprocessing the Acoustic Data

[0077] In some embodiments, the method may comprise preprocessing the acoustic data. In some cases, the method may comprise applying noise reduction techniques to the recorded signals. In some instances, the method may comprise filtering environmental noise using a separate noise recording channel. In some instances, the method may comprise calibrating the signal based on device positioning data. As an example, the method may comprise validating the signal quality before processing.

[0078] In some embodiments, the method may comprise processing the acquired acoustic data. In some cases, the method may comprise performing signal normalization on the acoustic data. In some instances, the method may comprise applying a windowing technique to identify systolic cycles within the normalized acoustic data. In some instances, the method may comprise performing DSP operations such as filtering and ratification reduction on the windowed acoustic data to generate frequency domain data. In some instances, the frequency domain data may comprise multiple frequency bins representing different frequency components of the acoustic data.

[0079] In some embodiments, the method may comprise acquiring acoustic data from a patient's arteries. In some cases, the method may comprise capturing composite multichannel binary data which are later processed (e.g., assembled into audio files, mixed, filtered etc.). In some instances, the method may comprise creating multiple audio files corresponding to different microphones. In some instances, the method may comprise processing the recorded signals through a windowing process. As an example, the method may comprise normalizing the signal and converting it from time domain to frequency domain.Examples of Signal Extraction and Statistical Analysis

[0080] In an experiment setup, the method may comprise signal extraction procedures. In some cases, the method may comprise identifying individual carotid pulse cycles. In some instances, the method may comprise manual visual and audible inspection of recordings. In some instances, the method may comprise using a Fast Fourier Transform (FFT) with a selected frequency bin. For example, the frequency bin may be a frequency size of about 1Hz or any other bin size. As an example, the method may comprise analyzing frequency ranges from 0.001-800 Hz.

[0081] In the experiment, the method may comprise statistical analysis procedures. In some instances, the method may comprise using an alpha value and checking for statistically significant differences between frequency bin pairs across all samples. For example, the method may comprise validating that different stenosis levels produce significantly different acoustic signatures.WSGR Docket No. 69145-701.601

[0082] In the experiment, the method may comprise correlation analysis of frequency data. In some cases, the method may comprise creating correlation matrices for each stenosis level. In some instances, the method may comprise correlating frequency bins within single stenosis levels using Spearman correlation with 95% confidence interval. In some instances, the method may comprise excluding non-significant r values (p > 0.05) and outliers. As an example, the method may comprise visualizing correlation matrices by plotting positive correlations as black datapoints and negative correlations as red datapoints. For example, the method may comprise creating unique “fingerprints” for each stenosis level tested.Machine Learning-Based Analysis

[0083] In some embodiments, the method may comprise using an inference module such as a machine learning or deep learning model to analyze processed physiological sound information and generate an output indicative of stenosis levels.

[0084] In some embodiments, the method may comprise analyzing the processed signals. In some cases, the method may comprise breaking down the signals into frequency or time components. In some instances, the method may comprise selecting desired frequency ranges. In some instances, the method may comprise inputting the processed data into a machine learning algorithm. As an example, the method may comprise performing inference on the unique data input. For example, the method may comprise outputting a defined stenosis grading of a patient’s carotid artery.

[0085] In some embodiments, the method may comprise analyzing the processed acoustic data using a machine learning algorithm. In some cases, the method may comprise inputting the frequency or time domain data into a neural network trained to identify stenosis levels. In some instances, the method may comprise supplementing the frequency or time domain data with additional patient metrics comprising at least one of: heart rate, blood pressure, body mass index, cholesterol levels, cardiac event history, or smoking history.

[0086] In some embodiments, the method may comprise employing machine learning-based analysis. In some cases, the method may comprise utilizing neural network models. In some instances, the method may comprise incorporating additional patient metrics such as blood pressure, BMI, and cardiac history. In some instances, the method may comprise generating confidence scores for the stenosis predictions. As an example, the method may comprise tracking changes in stenosis levels over time. For example, the method may comprise generating longitudinal analysis reports of plaque development patterns.Training Algorithm for the Machine Learning ModelWSGR Docket No. 69145-701.601

[0087] In some embodiments, the methods may comprise a training algorithm for training the model utilizing unique training dataset. The training dataset may comprise paired datasets including labeled input features. For example, the training dataset comprises recordings of blood flow sounds from the carotid artery, each associated with a known level of stenosis (labels).

[0088] In some embodiments, the method may comprise receiving patient metric data comprising at least one of: heart rate, blood pressure, body mass index, cholesterol levels, or cardiac event history. In some cases, the method may comprise using the patient metric data as additional input to the machine learning model. In some instances, the method may comprise synthesizing acoustic data for pretraining of the prediction model, based on empirical correlations between the frequency or time domain representations of the signal and parameters of interest. The machine learning algorithm is trained on the acoustic signal dataset, supplementary patient metric data, as well as the ground-truth labels (i.e., maps the signal to the corresponding stenosis levels (temporal evolution)).

[0089] FIG. 4 shows an example of a machine learning workflow and the training algorithm for training a model 400. The workflow 400 may comprise both inference processing and training validation procedures for stenosis detection. In some instances, the workflow 400 may comprise a recording module 401 configured to capture acoustic data from a measurement unit, which may be included along with additional patient metrics (e.g., blood pressure or heart rate). In some instances, the figure may depict a data upload component 402 that may transmit recorded files to a processing pipeline, which may be executed locally or on remote server infrastructure.

[0090] In some instances, the workflow 400 may comprise a binary decomposition module 403 that separates input files into multiple audio streams 404, each corresponding to a distinct sensor channel or microphone. The figure may also depict a windowing processor 405 that may isolate systolic cycle portions of the cardiac waveform, allowing the system to focus on the most relevant segments of the heartbeat.

[0091] In some instances, the workflow 400 may comprise a domain conversion module 406 that transforms the windowed data into frequency components, producing a converted data output 407. For example, the workflow may include a signal processing chain comprising a normalization unit 408, a frequency binning module 409, and a frequency range selector 410. These steps aim to capture pertinent acoustic features of arterial blood flow while filtering out noise or irrelevant frequency ranges.

[0092] The workflow 400 may comprise a machine learning component 411 that processes the frequency-domain data. In some instances, the machine learning component 411 may operate in an inference mode for analyzing individual patient recordings or in a training mode forWSGR Docket No. 69145-701.601developing prediction models using labeled data. The figure may depict an inference engine 412 that may generate stenosis predictions based on the processed data. For example, the workflow may culminate in an output module 413 that produces a defined stenosis grading (e.g., mild, moderate, or severe).

[0093] FIG.4 illustrates how the system processes both individual patient recordings for realtime analysis and multiple labeled recordings for algorithm training. As shown, the same signal processing steps may support both inference and training operations within a unified workflow architecture, thereby leveraging shared data pipelines to ensure consistency across real-time assessments and model development.

[0094] In some cases, the method may comprise utilizing recordings from carotid phantoms with induced stenosis levels. In some instances, the method may comprise incorporating patient recordings validated through medical imaging techniques including Doppler Ultrasound, Magnetic Resonance Angiography (MRA), or Computed Tomography Angiography (CTA). In some instances, the method may comprise rigorous testing with unique data. As an example, the method may comprise evaluating specificity, sensitivity, accuracy, negative predictive value (NPV), and positive predictive value (PPV).Artificial Artery Testing Procedures

[0095] In some embodiments, the method may comprise collecting artificial (non-in vivo) data. In some embodiments, the method may comprise artificial artery testing procedures. In some cases, the method may comprise using CT-scan based models of carotid artery bifurcation. In some instances, the method may comprise 3D printing models using compliant resin. In some instances, the method may comprise creating models with various stenosis levels (0, 30, 40, 50, 60, 70, 74, 78, 88%). The stenosis levels can be any number from 0 to 100%, where the plaque can be made of a homogeneous or morphology mimicking material. As an example, the method may comprise utilizing a blood mimicking fluid comprising a mix of glycerin, water, and xanthan gum. For example, the method may comprise pumping the blood mimicking fluid through the models using a custom pump configured to replicate natural blood flow characteristics.Optimization and Accuracy in Experiment Setup

[0096] In an experiment setup, the method may comprise optimization. For example, the Adam optimization algorithm may be used. In an experiment, a model is trained on datasets of a sufficient size , for example of 30 samples, each sample may comprise 1Hz frequency bins from the range of 0.001-800Hz for each stenosis level. The model is tested on unique databases not previously seen by the model. In the experiment, the model achieves a sufficient accuracy in stenosis level prediction.WSGR Docket No. 69145-701.601Generating a Stenosis Assessment

[0097] In some embodiments, the method may comprise generating a stenosis assessment based on the analysis. In some cases, the method may comprise determining a degree of arterial stenosis present in the carotid artery. In some instances, the method may comprise storing the stenosis assessment in association with a timestamp to configure longitudinal monitoring of stenosis progression.Repeated Measurements and Longitudinal Monitoring

[0098] In some embodiments, the method may comprise performing repeated measurements over time to monitor plaque development. In some cases, the method may comprise comparing sequential stenosis assessments to identify changes in arterial stenosis levels. In some instances, the method may comprise generating alerts when changes in stenosis levels exceed predetermined thresholds.Automated Quality Control and Maintenance Guidance

[0099] In some embodiments, the method may comprise implementing automated quality control protocols with actionable maintenance guidance. In some cases, the method may comprise analyzing calibration history trends to generate specific maintenance recommendations, such as “replace acoustic sensor,” “perform manual recalibration,” or “update noise reduction parameters.” In some instances, the reliability reports may include predictive maintenance schedules based on usage patterns and performance metrics. In some instances, the method may comprise maintaining audit trails of all maintenance activities to ensure regulatory compliance and system reliability.Dynamic Noise Reduction Optimization

[0100] In some embodiments, the method may comprise implementing dynamic noise reduction optimization. In some cases, the method may comprise conducting continuous environmental noise sampling to adjust acoustic filtering parameters in real-time. In some instances, the system may characterize ambient noise patterns to automatically select improved noise cancellation algorithms. In some instances, the method may comprise maintaining noise reduction effectiveness across varying environmental conditions through adaptive parameter tuning.Integrating Complementary Health Monitoring Data

[0101] In some embodiments, the method may comprise integrating complementary health monitoring data. In some cases, the method may comprise incorporating heart rate variability, physical activity levels, and blood pressure trends from wearable devices to provide context for stenosis measurements. In some instances, the integrated data may reveal correlations between lifestyle factors and stenosis progression. In some instances, the method may compriseWSGR Docket No. 69145-701.601generating comprehensive health insights that combine vascular measurements with broader cardiovascular indicators.Expanding Fingerprint Database Representation

[0102] In some embodiments, the method may comprise expanding fingerprint database representation across diverse populations. In some cases, the method may comprise incorporating data from pediatric patients, individuals with anatomical variations, and historically underrepresented demographic groups. In some instances, the method may comprise developing specialized algorithms for analyzing atypical stenosis patterns. In some instances, the method may comprise implementing risk-adjusted analysis protocols based on population-specific factors.Modular Software Architecture

[0103] In some embodiments, the method may comprise implementing modular software architecture supporting continuous enhancement. In some cases, the method may comprise allowing seamless integration with emerging Al models, including transformer-based architectures for enhanced pattern recognition. In some instances, the method may comprise maintaining compatibility with evolving healthcare data standards such as FHIR and HL7. In some instances, the method may comprise supporting automated software updates to incorporate new diagnostic capabilities and security features.Comprehensive Accessibility Features

[0104] In some embodiments, the method may comprise providing comprehensive accessibility features. In some cases, the method may comprise offering voice-guided instructions in multiple languages for device positioning and measurement procedures. In some instances, the method may comprise implementing configurable user interfaces adaptable to various visual, auditory, and cognitive needs. In some instances, the method may comprise maintaining compliance with accessibility standards across all user interaction points.Robust Validation and Risk Management

[0105] In some embodiments, the method may comprise implementing robust validation and risk management protocols. In some cases, the method may comprise conducting large-scale clinical trials to validate system accuracy across diverse populations and use cases. In some instances, the method may comprise employing multiple verification algorithms to minimize false positives and negatives in stenosis detection. In some instances, the method may comprise implementing graduated alert systems that distinguish between routine observations and clinically significant findings requiring immediate attention.Advanced Predictive AnalyticsWSGR Docket No. 69145-701.601

[0106] In some embodiments, the method may comprise implementing advanced predictive analytics capabilities. In some cases, the method may comprise utilizing machine learning models to forecast stenosis progression based on historical measurement patterns and risk factors. In some instances, the method may comprise generating personalized risk projections that account for individual patient characteristics and lifestyle factors. In some instances, the method may comprise providing early warning indicators when progression patterns suggest increased probability of adverse events.Diverse Clinical Integration Scenarios

[0107] In some embodiments, the method may comprise supporting diverse clinical integration scenarios. In some cases, the method may comprise allowing emergency department triage through rapid stenosis assessment integrated with vital sign monitoring. In some instances, the method may comprise supporting primary care monitoring by combining regular stenosis measurements with annual health assessments. In some instances, the method may comprise facilitating remote patient monitoring through secure data sharing between home measurements and clinical systems.Global Accessibility and Deployment Flexibility

[0108] In some embodiments, the method may comprise ensuring global accessibility and deployment flexibility. In some cases, the method may comprise adapting system requirements to function effectively in low-resource settings with limited healthcare infrastructure. In some instances, the method may comprise providing offline processing capabilities when continuous network connectivity is unavailable. In some instances, the method may comprise implementing power management features to support extended operation in areas with unreliable power supply.Modular Diagnostic Expansion

[0109] In some embodiments, the method may comprise supporting modular diagnostic expansion. In some cases, the method may comprise incorporating additional vascular health biomarkers as new sensing technologies become available. In some instances, the method may comprise integrating complementary diagnostic modalities such as blood flow velocity measurements or tissue elastography. In some instances, the method may comprise maintaining extensible data structures to accommodate future diagnostic parameters without system redesign.Regulatory Compliance

[0110] In some embodiments, the method may comprise ensuring regulatory compliance across jurisdictions. In some cases, the method may comprise maintaining documentation and validation protocols aligned with FDA, CE marking, and other relevant regulatory requirements. In some instances, the method may comprise implementing quality management systems that supportWSGR Docket No. 69145-701.601medical device approval processes. In some instances, the method may comprise providing audit trails and performance monitoring to demonstrate ongoing compliance with safety and effectiveness standards.[OHl] The present disclosure provides computer systems that are programmed to implement methods of the disclosure. FIG. 5 shows a computer system 501 that is programmed or otherwise configured to perform data acquisition, frequency or time-domain analysis, and machinelearning-based assessment of arterial health. The computer system 501 may regulate various aspects of the multi-stage workflow of the present disclosure, such as, for example, receiving vascular acoustic signals, applying noise-reduction and windowing techniques, converting the processed data to a frequency or time-based representation, and generating stenosis predictions using a trained neural network. The computer system 501 may be an electronic device of a user or a computer system that is remotely located with respect to the electronic device. The electronic device may be a mobile electronic device.

[0112] The computer system 501 includes a central processing unit (CPU, also “processor” and “computer processor” herein) 502, which may be a single core or multi core processor, or a plurality of processors for parallel processing. The computer system 501 also includes memory or memory location 503 (e.g., random-access memory, read-only memory, flash memory), electronic storage unit 504 (e.g., hard disk), communication interface 505 (e.g., network adapter) for communicating with one or more other systems, and peripheral devices 506, such as cache, other memory, data storage and / or electronic display adapters. The memory 503, storage unit 504, interface 505 and peripheral devices 506 are in communication with the CPU 502 through a communication bus (solid lines), such as a motherboard. The storage unit 504 may be a data storage unit (or data repository) for storing data. The computer system 501 may be operatively coupled to a computer network (“network”) 100 with the aid of the communication interface 505.The network 507 may be the Internet, an internet and / or extranet, or an intranet and / or extranet that is in communication with the Internet. The network 507 in some cases is a telecommunication and / or data network. The network 507 may include one or more computer servers, which may configure distributed computing, such as cloud computing. The network 507, in some cases with the aid of the computer system 501, may implement a peer-to-peer network, which may configure devices coupled to the computer system 501 to behave as a client or a server.

[0113] The CPU 502 may execute a sequence of machine-readable instructions, which may be embodied in a program or software. The instructions may be stored in a memory location, such as the memory 503. The instructions may be directed to the CPU 502, which may subsequentlyWSGR Docket No. 69145-701.601program or otherwise configure the CPU 502 to implement methods of the present disclosure. Examples of operations performed by the CPU 502 may include fetch, decode, execute, and writeback.

[0114] The CPU 502 may be part of a circuit, such as an integrated circuit. One or more other components of the system 501 may be included in the circuit. In some cases, the circuit is an application specific integrated circuit (ASIC).

[0115] The storage unit 504 may store files, such as drivers, libraries and saved programs. The storage unit 504 may store user data, e.g., user preferences and user programs. The computer system 501 in some cases may include one or more additional data storage units that are external to the computer system 501, such as located on a remote server that is in communication with the computer system 501 through an intranet or the Internet.

[0116] The computer system 501 may communicate with one or more remote computer systems through the network 507. For instance, the computer system 501 may communicate with a remote computer system of a user (e.g., a patient, a healthcare professional, or another authorized stakeholder). Examples of remote computer systems include personal computers (e.g., portable PC), slate or tablet PC’s (e.g., Apple® iPad, Samsung® Galaxy Tab), telephones, Smart phones (e.g., Apple® iPhone, Android-configured device, Blackberry®), or personal digital assistants. The user may access the computer system 501 via the network 507.

[0117] Methods as described herein may be implemented by way of machine (e.g., computer processor) executable code stored on an electronic storage location of the computer system 501, such as, for example, on the memory 510 or electronic storage unit 504. The machine executable or machine readable code may be provided in the form of software. During use, the code may be executed by the processor 502. In some cases, the code may be retrieved from the storage unit 504 and stored on the memory 503 for ready access by the processor 502. In some situations, the electronic storage unit 504 may be precluded, and machine-executable instructions are stored on memory 503.

[0118] The code may be pre-compiled and configured for use with a machine having a processer adapted to execute the code, or may be compiled during runtime. The code may be supplied in a programming language that may be selected to configure the code to execute in a pre-compiled or as-compiled fashion.

[0119] Aspects of the systems and methods provided herein, such as the computer system 501, may be embodied in programming. Various aspects of the technology may be thought of as “products” or “articles of manufacture” typically in the form of machine (or processor) executable code and / or associated data that is carried on or embodied in a type of machineWSGR Docket No. 69145-701.601readable medium. Machine-executable code may be stored on an electronic storage unit, such as memory (e.g., read-only memory, random-access memory, flash memory) or a hard disk.“Storage” type media may include any or all of the tangible memory of the computers, processors or the like, or associated modules thereof, such as various semiconductor memories, tape drives, disk drives and the like, which may provide non-transitory storage at any time for the software programming. All or portions of the software may at times be communicated through the Internet or various other telecommunication networks. Such communications, for example, may configure loading of the software from one computer or processor into another, for example, from a management server or host computer into the computer platform of an application server. Thus, another type of media that may bear the software elements includes optical, electrical and electromagnetic waves, such as used across physical interfaces between local devices, through wired and optical landline networks and over various air-links. The physical elements that carry such waves, such as wired or wireless links, optical links or the like, also may be considered as media bearing the software. As used herein, unless restricted to non-transitory, tangible “storage” media, terms such as computer or machine “readable medium” refer to any medium that participates in providing instructions to a processor for execution.

[0120] Hence, a machine readable medium, such as computer-executable code, may take many forms, including but not limited to, a tangible storage medium, a carrier wave medium or physical transmission medium. Non-volatile storage media include, for example, optical or magnetic disks, such as any of the storage devices in any computer(s) or the like, such as may be used to implement the databases, etc. shown in the drawings. Volatile storage media include dynamic memory, such as main memory of such a computer platform. Tangible transmission media include coaxial cables; copper wire and fiber optics, including the wires that comprise a bus within a computer system. Carrier-wave transmission media may take the form of electric or electromagnetic signals, or acoustic or light waves such as those generated during radio frequency (RF) and infrared (IR) data communications. Common forms of computer-readable media therefore include for example: a floppy disk, a flexible disk, hard disk, magnetic tape, any other magnetic medium, a CD-ROM, DVD or DVD-ROM, any other optical medium, punch cards paper tape, any other physical storage medium with patterns of holes, a RAM, a ROM, a PROM and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave transporting data or instructions, cables or links transporting such a carrier wave, or any other medium from which a computer may read programming code and / or data. Many of these forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.WSGR Docket No. 69145-701.601

[0121] The computer system 501 may include or be in communication with an electronic display 105 that comprises a user interface (UI) 509 for providing, for example, instructions for device placement, real-time feedback regarding arterial sound recordings, or interactive controls for data visualization and machine-learning analysis results. Examples of UIs include, without limitation, a graphical user interface (GUI) and a web-based user interface.

[0122] Methods and systems of the present disclosure may be implemented by way of one or more algorithms. An algorithm may be implemented by way of software upon execution by the central processing unit 502. The algorithm can, for example, analyze incoming acoustic signals, apply noise-reduction techniques, identify systolic cycles, and generate an output indicative of an estimated arterial stenosis level.Computer systems

[0123] The present disclosure provides computer systems that are programmed to implement methods of the disclosure. FIG. 5 shows a computer system 501 that is programmed or otherwise configured to regulate and monitor the heat transfer in a heat transfer system as per the given specifications. The computer system 501 may regulate various aspects of the heat transfer and control operations of the present disclosure, such as, for example, the direction and control of heat flow between the first and second streams via the intermediate heat transfer fluid. The computer system 501 may be an electronic device of a user or a computer system that is remotely located with respect to the electronic device. The electronic device may be a mobile electronic device.

[0124] The computer system 501 includes a central processing unit (CPU, also “processor” and “computer processor” herein) 502, which may be a single core or multi core processor, or a plurality of processors for parallel processing. The computer system 501 also includes memory or memory location 503 (e.g., random-access memory, read-only memory, flash memory), electronic storage unit 504 (e.g., hard disk), communication interface 505 (e.g., network adapter) for communicating with one or more other systems, and peripheral devices 506, such as cache, other memory, data storage and / or electronic display adapters. The memory 503, storage unit 504, interface 505 and peripheral devices 506 are in communication with the CPU 502 through a communication bus (solid lines), such as a motherboard. The storage unit 504 may be a data storage unit (or data repository) for storing data. The computer system 501 may be operatively coupled to a computer network (“network”) 507 with the aid of the communication interface 505.The network 507 may be the Internet, an internet and / or extranet, or an intranet and / or extranet that is in communication with the Internet. The network 507 in some cases is a telecommunication and / or data network. The network 507 may include one or more computerWSGR Docket No. 69145-701.601servers, which may configure distributed computing, such as cloud computing. The network 507, in some cases with the aid of the computer system 501, may implement a peer-to-peer network, which may configure devices coupled to the computer system 501 to behave as a client or a server.

[0125] The CPU 502 may execute a sequence of machine-readable instructions, which may be embodied in a program or software. The instructions may be stored in a memory location, such as the memory 503. The instructions may be directed to the CPU 502, which may subsequently program or otherwise configure the CPU 502 to implement methods of the present disclosure. Examples of operations performed by the CPU 502 may include fetch, decode, execute, and writeback.

[0126] The CPU 502 may be part of a circuit, such as an integrated circuit. One or more other components of the system 501 may be included in the circuit. In some cases, the circuit is an application specific integrated circuit (ASIC).

[0127] The storage unit 504 may store files, such as drivers, libraries and saved programs. The storage unit 504 may store user data, e.g., user preferences and user programs. The computer system 501 in some cases may include one or more additional data storage units that are external to the computer system 501, such as located on a remote server that is in communication with the computer system 501 through an intranet or the Internet.

[0128] The computer system 501 may communicate with one or more remote computer systems through the network 507. For instance, the computer system 501 may communicate with a remote computer system of a user (e.g., an engineer overseeing the operation of the heat transfer system). Examples of remote computer systems include personal computers (e.g., portable PC), slate or tablet PC’s (e.g., Apple® iPad, Samsung® Galaxy Tab), telephones, Smart phones (e.g., Apple® iPhone, Android-configured device, Blackberry®), or personal digital assistants. The user may access the computer system 501 via the network 507.

[0129] Methods as described herein may be implemented by way of machine (e.g., computer processor) executable code stored on an electronic storage location of the computer system 501, such as, for example, on the memory 503 or electronic storage unit 504. The machine executable or machine readable code may be provided in the form of software. During use, the code may be executed by the processor 502. In some cases, the code may be retrieved from the storage unit 504 and stored on the memory 503 for ready access by the processor 502. In some situations, the electronic storage unit 504 may be precluded, and machine-executable instructions are stored on memory 503.WSGR Docket No. 69145-701.601

[0130] The code may be pre-compiled and configured for use with a machine having a processer adapted to execute the code, or may be compiled during runtime. The code may be supplied in a programming language that may be selected to configure the code to execute in a pre-compiled or as-compiled fashion.

[0131] Aspects of the systems and methods provided herein, such as the computer system 501, may be embodied in programming. Various aspects of the technology may be thought of as “products” or “articles of manufacture” typically in the form of machine (or processor) executable code and / or associated data that is carried on or embodied in a type of machine readable medium. Machine-executable code may be stored on an electronic storage unit, such as memory (e.g., read-only memory, random-access memory, flash memory) or a hard disk.“Storage” type media may include any or all of the tangible memory of the computers, processors or the like, or associated modules thereof, such as various semiconductor memories, tape drives, disk drives and the like, which may provide non-transitory storage at any time for the software programming. All or portions of the software may at times be communicated through the Internet or various other telecommunication networks. Such communications, for example, may configure loading of the software from one computer or processor into another, for example, from a management server or host computer into the computer platform of an application server. Thus, another type of media that may bear the software elements includes optical, electrical and electromagnetic waves, such as used across physical interfaces between local devices, through wired and optical landline networks and over various air-links. The physical elements that carry such waves, such as wired or wireless links, optical links or the like, also may be considered as media bearing the software. As used herein, unless restricted to non-transitory, tangible “storage” media, terms such as computer or machine “readable medium” refer to any medium that participates in providing instructions to a processor for execution.

[0132] Hence, a machine readable medium, such as computer-executable code, may take many forms, including but not limited to, a tangible storage medium, a carrier wave medium or physical transmission medium. Non-volatile storage media include, for example, optical or magnetic disks, such as any of the storage devices in any computer(s) or the like, such as may be used to implement the databases, etc. shown in the drawings. Volatile storage media include dynamic memory, such as main memory of such a computer platform. Tangible transmission media include coaxial cables; copper wire and fiber optics, including the wires that comprise a bus within a computer system. Carrier-wave transmission media may take the form of electric or electromagnetic signals, or acoustic or light waves such as those generated during radio frequency (RF) and infrared (IR) data communications. Common forms of computer-readableWSGR Docket No. 69145-701.601media therefore include for example: a floppy disk, a flexible disk, hard disk, magnetic tape, any other magnetic medium, a CD-ROM, DVD or DVD-ROM, any other optical medium, punch cards paper tape, any other physical storage medium with patterns of holes, a RAM, a ROM, a PROM and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave transporting data or instructions, cables or links transporting such a carrier wave, or any other medium from which a computer may read programming code and / or data. Many of these forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.

[0133] The computer system 501 may include or be in communication with an electronic display 508 that comprises a user interface (UI) 509 for providing, for example, real-time temperature readings, flow directions, and other critical data related to the operation of the heat transfer system. The UI 509 may also present visual alerts if temperature or flow parameters deviate from predefined thresholds, as well as interactive controls to adjust system settings. Examples of UIs include, without limitation, a graphical user interface (GUI) and a web-based user interface.

[0134] Methods and systems of the present disclosure may be implemented by way of one or more algorithms. An algorithm may be implemented by way of software upon execution by the central processing unit 502. The algorithm can, for example, calculate and adjust the improved heat flow between the first and second streams by regulating the intermediate heat transfer fluid. It may monitor the temperature variations and make necessary adjustments to ensure efficient heat transfer, ensuring that the system operates within the desired parameters.EXAMPLES

[0135] The following illustrative examples are representative of embodiments of the software applications, systems, and methods described herein and are not meant to be limiting in any way.Example 1: Development of an Acoustic-Based Arterial Stenosis Detection System

[0136] In this case, a system is provided that incorporates an integrated acoustic sensing and signal processing setup. The system quantifies arterial blood flow characteristics and stenosis levels through acoustic analysis, achieving improvements in accessibility, temporal resolution, and cost-efficiency compared to traditional clinical methods. The system serves as a scalable diagnostic tool for at-home arterial health monitoring.

[0137] The system incorporates a specialized acoustic transducer array that achieves high-fidelity detection of blood flow sounds, with particular emphasis on the infrasonic range (sub-20 Hz). The system utilizes an array of complementary transducers that enhance signal captureWSGR Docket No. 69145-701.601across different frequency ranges. The acoustic sensing unit includes an optimized membrane for skin contact, specialized signal conditioning circuitry, and precise analog-to-digital conversion circuitry configured to capture low-amplitude acoustic signals with high accuracy.

[0138] The testing system includes CT-scan based carotid artery models manufactured using compliant resin (e.g., Elastic 50A resin or any other bio-compatible material) through additive manufacturing processes. The models simulate various stenosis levels (e.g., 0, 30, 40, 50, 60, 70, 74, 78, 88% occlusion or any number between 0-100% occlusion), with calculations based on cross-sectional area reduction. The system includes a custom-configured pump configured to replicate physiological blood flow characteristics and a blood-mimicking fluid comprising a specific mixture of glycerin, water, and xanthan gum for simulating blood flow dynamics.

[0139] The system implements a comprehensive signal processing pipeline that extracts and analyzes frequency components across a 0.001-800 Hz range. The processing system identifies individual carotid pulse cycles and processes a number of samples sufficient for statistical significance (e.g. n=30) for each stenosis level. Each sample undergoes normalization followed by Fast Fourier Transform (FFT) analysis. The system performs statistical analysis to evaluate the differences between frequency bins of variable sizes across stenosis levels. The system generates correlation matrices for each stenosis level, creating unique "fingerprints" for different stenosis levels through visualization of positive and negative correlations.

[0140] The system includes a deep learning network utilizing an optimization algorithm (e.g., Adam optimization algorithm), achieving an improved accuracy in stenosis level prediction. The network processes a training dataset of a sufficient size to generate, with each sample comprising 0.001-800Hz samples for each stenosis level. The system validates performance using a separate testing database of previously unseen recordings.Example 2: System Applications and Implementations

[0141] The system includes capabilities for monitoring large vessel occlusions beyond the carotid arteries. The system incorporates functionality for assessing bruits in different locations across the body, for such as bruits of carotid arteries, cervical veins or arteriovenous (AV) connections, aorta, renal arteries, iliac arteries, hepatic artery, and splenic artery, and femoral artery blockages associated with diabetic complications. Through specific hardware and software modifications, the system includes coronary artery assessment capabilities for myocardial infarction risk monitoring and ability to assess bruits across different vessels in the body.

[0142] The system includes a positioning system for accurate identification of internal and common carotid arteries, incorporating personalized features for individual patient anatomies. The system implements user interface elements based on iterative usability studies, ensuring easeWSGR Docket No. 69145-701.601of use across different user populations. The system incorporates compatibility with clinical validation protocols, allowing the collection and analysis of human subject data. The processing pipeline includes advanced algorithms for plaque morphology assessment, analyzing different forms of plaque development within the arteries and their impact on stroke risk profiles.

[0143] The system integrates with existing healthcare workflows, providing continuous monitoring capabilities during extended assessment periods. The system configures healthcare providers to track and analyze arterial health changes over time, supporting data-driven intervention decisions when warranted by the acoustic analysis results.Example 3: Signal Analysis and Stenosis Classification

[0144] The system incorporates acoustic signal processing capabilities for stenosis classification across multiple blockage levels. The processing system analyzes acoustic signatures in the range of 0.001-800 Hz, with particular focus on infrasonic frequencies below 20 Hz. The system implements Fast Fourier Transform analysis with 1 Hz resolution bins, creating detailed frequency or time domain representations of blood flow characteristics. The analysis system processes multiple samples per measurement (e.g. n=15), allowing robust statistical validation of detected patterns.

[0145] For each stenosis level, the system generates correlation matrices using Spearman correlation analysis (CI=95%) across all frequency bins. The system creates unique "fingerprints" for different stenosis levels (e.g., 0-100% occlusion) by visualizing positive and negative correlations between frequency components. These fingerprints serve as reference patterns for subsequent stenosis classification through pattern matching algorithms.

[0146] The system includes a deep learning network utilizing an optimization algorithm for stenosis classification. For example, an Adam optimization algorithm may be used. The network processes frequency or time domain data from the acoustic analysis pipeline, achieving an improved classification accuracy. The system validates performance using independent test datasets not used during training, ensuring robust classification across different measurement conditions.Example 4: Device Positioning and User Interface

[0147] The system incorporates positioning assistance mechanisms for accurate placement over carotid arteries. The positioning system includes sensors for detecting proper contact with the skin surface and alignment with target vessels. The system provides feedback to users during device placement, ensuring improved signal acquisition conditions for subsequent analysis.

[0148] For measurement validation, the system implements quality control protocols for recorded signals. The analysis system evaluates signal characteristics including amplitude,WSGR Docket No. 69145-701.601frequency content, and noise levels to ensure reliable data acquisition. When signal quality falls outside acceptable parameters, the system indicates required adjustments to device positioning or measurement conditions.Example 5: Clinical Validation Implementation

[0149] The system includes testing protocols utilizing carotid phantom models for performance validation. The testing system implements blood flow simulation using a blood-mimicking fluid composed of glycerin, water, and xanthan gum. The flow control system replicates physiological blood flow patterns through artificial vessel models with precisely controlled stenosis levels.

[0150] The system incorporates plaque morphology assessment capabilities through acoustic pattern analysis. The analysis system evaluates frequency or time components associated with different plaque characteristics, allowing assessment of plaque stability and development. This analysis contributes to overall stroke risk assessment based on both stenosis level and plaque characteristics.

[0151] For clinical validation, the system includes protocols for comparison with established diagnostic methods including carotid ultrasound and angiogram results. The validation system configures correlation between acoustic signatures and clinically verified stenosis levels, supporting development of robust classification algorithms.

[0152] While preferred embodiments of the present invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in practicing the invention. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

WSGR Docket No. 69145-701.601CLAIMSWhat is claimed is:

1. A method for monitoring arterial health, comprising:(a) receiving physiological sound data from a mechanical transducer device positioned proximate to an artery;(b) processing the physiological sound data to generate acoustic frequency or time domain data;(c) analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a measure of stenosis in the artery; and(d) outputting an indication of the determined measure of stenosis in the artery.

2. The method of claim 1, wherein the physiological sound data is infrasound or audible range acoustic vibration data.

3. The method of any one of claims 1-2, further comprising:(a) analyzing a plurality of changes in the acoustic frequency or time domain data over a period of time;(b) predicting future stenosis progression based on the plurality of changes over the period of time; and(c) generating one or more risk assessments based on the predicted future stenosis progression.

4. The method of any one of claims 1-3, further comprising:(a) detecting systolic cycles within the acoustic frequency or time domain data; and (b) windowing the acoustic frequency or time domain data prior to generating the acoustic frequency or time domain data.

5. The method of any one of claims 1-4, further comprising:(a) receiving environmental noise data from at least one secondary mechanical transducer device; and(b) applying noise reduction to the acoustic frequency or time domain data based on the environmental noise data.

6. The method of any one of claims 1-5, further comprising:(a) receiving positioning data indicating a position of the mechanical transducer device relative to the artery; and(b) providing feedback for adjusting the position of the mechanical transducer device based on the positioning data.WSGR Docket No. 69145-701.6017. The method of any one of claims 1-6, further comprising:(a) analyzing amplitude, frequency content, and noise levels of the physiological sound data to determine a signal quality; and(b) when the signal quality falls outside a specified threshold, transmitting pressure fluctuations to the mechanical transducer device using a pressure sensor.

8. The method of any one of claims 1-7, wherein the method is performed at home or outside of a healthcare environment.

9. A system for monitoring arterial health, the system comprising:a housing, andat least one mechanical transducer device configured to capture physiological sound signals,wherein the system is configured to be placed proximate to an artery of a subject, wherein the system further comprises a computer-readable medium storing instructions that, when executed by a computer, cause it to perform:(a) processing the physiological sound signals to generate acoustic frequency or time domain data;(b) analyzing the acoustic frequency or time domain data and one or more supplementary patient metrics with at least one machine learning or deep learning model to determine a degree of stenosis in the artery; and(c) outputting an indication of a determined degree of arterial stenosis.

10. The system of claim 9, wherein the at least one mechanical transducer device collects infrasound or audible range acoustic vibration data.

11. The system of claim 9 or claim 10, wherein the housing comprises an acoustic membrane configured to contact the subject’s skin.

12. The system of claim 11, wherein the acoustic membrane is configured to capture infrasonic frequency vibrations below 20 Hz.

13. The system of claim 11 or claim 12, wherein the acoustic membrane is configured to capture audible range acoustic vibrations.

14. The system of any one of claims 9-13, further comprising a supplementary sensor assembly configured to detect applied contact pressure.

15. The system of any one of claims 9-14, further comprising a digital signal processing (DSP) unit configured to extract time or frequency information indicative of vascular flow characteristics from the physiological sound signals.WSGR Docket No. 69145-701.60116. The system of any one of claims 9-15, further comprising an interface module configured to display real-time visualizations of the physiological sound signals.

17. The system of any one of claims 9-16, further comprising a data integration platform.

18. The system of claim 17, wherein the data integration platform is configured to synchronize the analyzed acoustic frequency or time domain data with electronic health records (EHRs).

19. The system of any one of claims 9-18, further comprising a server architecture.

20. The system of claim 19, wherein the server architecture comprises pre-trained machine learning or deep learning models configured to perform stenosis analysis.

21. The system of any one of claims 9-20, further comprising a layered hardware architecture configured to allow upgrades or replacements of modules without affecting the entire system.

22. The system of any one of claims 9-21, further comprising a data validation system configured to flag poor-quality measurements.

23. The system of any one of claims 9-22, further comprising a data security assembly configured to perform encryption of subject health information.

24. The system of any one of claims 9-23, further comprising a telemedicine interface configured to allow real-time consultation with a healthcare provider based on the determined measure of arterial stenosis.

25. The system of any one of claims 9-24, wherein the system is handheld.

26. The system of any one of claims 9-25, wherein the system is portable.

Images