System and method for physiological monitoring using biophotonics and machine learning analysis
The system addresses challenges in non-invasive wearable monitoring by capturing high-fidelity physiological parameters through coherent light and optical sensors, achieving clinical-grade accuracy and continuous health monitoring with motion description modeling and closed-loop feedback.
Patent Information
- Application Number
- PCT/EP2026/063625
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-09
- Filing Date
- 2026-05-11
- Publication Date
- 2026-11-12
AI Technical Summary
Existing non-invasive wearable physiological monitoring technologies face challenges in accurately measuring high-fidelity physiological parameters like detailed heart sounds and blood flow indicators, are susceptible to motion artifacts, and require complex user calibration for cuffless blood pressure measurement, while conventional optical methods struggle with signal-to-noise ratio degradation due to spatial mismatches and lack sensitivity to sub-dermal dynamics.
A system utilizing coherent light sources and optical sensors captures high-frequency image frames of dynamic speckle patterns, subdivides the field of view into sub-regions, applies closed-loop feedback for SNR optimization, and integrates machine learning to derive clinical parameters like cuffless blood pressure and heart sound components through motion description modeling.
The system achieves clinical-grade accuracy in deriving high-fidelity physiological parameters without user calibration, overcoming spatial mismatches and noise issues, and enables continuous, non-invasive health monitoring across diverse form factors.
Smart Images

Figure EP2026063625_12112026_PF_FP_ABST
Abstract
Description
SYSTEM AND METHOD FOR PHYSIOLOGICAL MONITORING USING BIOPHOTONICS AND MACHINE LEARNING ANALYSISBACKGROUNDTechnical Field
[0001] The disclosure relates generally to physiological monitoring systems and, more particularly, to a system and a method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment.Description of the Related Art
[0002] The present disclosure generally relates to physiological monitoring systems, particularly those utilizing non-invasive sensors integrated into wearable devices or other proximate devices. Continuous monitoring of biovitals like heart sounds, respiratory sounds, heart rate (HR), blood pressure (BP), respiration rate (RR), and oxygen saturation (SpCh) are crucial for health management, disease detection, and optimizing treatments. There is a significant and ongoing trend towards developing non-invasive and convenient methods for physiological monitoring, especially through wearable devices that allow for continuous data collection during daily activities, away from traditional clinical settings.
[0003] Various technologies are currently employed for physiological assessment. Wearable devices, such as smartwatches, fitness bands, patches, and sensor-integrated clothing, commonly incorporate optical sensing techniques. These techniques typically use light sources, like Light Emitting Diodes (LEDs), to illuminate the skin and optical sensors, like photodiodes, to detect changes in light reflected or scattered back from tissue, often using principles like photoplethysmography (PPG) to estimate parameters such as heart rate and blood oxygen saturation. The integration of multiple sensor types into single devices or platforms is also explored.
[0004] Furthermore, the application of computational algorithms, including machine learning (ML), deep learning (DL), and artificial intelligence (Al), to analyze data gathered from wearable sensors is an established practice. These algorithms are utilized for diverse functions, such as extracting relevant features from complex time-series sensor data, monitoring physiological signs to detect potential illness or adverse conditions,predicting future health states or outcomes, and sometimes adapting recommendations or alerts based on the collected data. Concepts like analyzing physiological trends over extended periods (longitudinal analysis) and incorporating user-specific information (like demographics) into the analysis are also known areas of investigation. Efforts are continually made to improve the quality and reliability of signals obtained from wearable sensors.
[0005] Despite these ongoing developments, significant challenges and limitations remain in the field of non-invasive, wearable physiological monitoring. Accurately measuring certain key vital signs continuously, conveniently, and without requiring user calibration remains difficult. For example, obtaining reliable, cuffless blood pressure measurements that meet clinical accuracy standards is a well-known challenge for many current wearable technologies. Additionally, extracting high-fidelity information beyond basic parameters like heart rate, such as detailed acoustic signatures corresponding to heart sounds, respiratory sounds, or indicators of blood flow turbulence, using non-invasive optical methods presents considerable technical hurdles. Many existing optical techniques may primarily analyze bulk changes in light absorption or scattering, potentially overlooking subtle but information-rich dynamic patterns in the light interacting with the tissue surface caused by physiological processes.
[0006] Moreover, ensuring high signal quality and robustness against motion artifacts is a persistent issue for sensors worn during daily activities. Integrating advanced, multi-parameter sensing capabilities into comfortable, unobtrusive, and diverse form factors suitable for long-term use also poses design and technical difficulties.
[0007] However, existing optical monitoring techniques, such as PPG, typically rely on measuring bulk changes in light absorption or scattering associated with blood volume pulses. Such conventional methods are fundamentally limited by their reliance on gross volumetric changes, which lack the requisite sensitivity and sampling frequency to capture high-fidelity cutaneous vibrations and sub-dermal physiological dynamics, such as detailed heart sound components SI through S4, respiratory sounds, or markers of turbulent blood flow. Furthermore, a persistent technical hurdle in biophotonic sensing arises from the spatial mismatch between the coherent light beam diameter, often ranging from 2 mm to 5 mm, and the smaller dimensions of target vascular structures. This mismatch leads to significant signal-to-noise ratio (SNR) degradation when the entire field of view (FOV) of a sensor is processed as a single data point, as physiological signals are often diluted by reflected light from non-dynamic surrounding tissue.
[0008] Additionally, achieving clinical-grade accuracy in cuffless blood pressure monitoring currently necessitates complex, user-specific calibration procedures, as existing systems lack the technical capability to dynamically optimize hardware parameters in realtime or map quantified tissue surface motion dynamics directly to acoustic and hemodynamic markers across diverse populations. There remains a need for a technical approach that utilizes motion description modeling and temporal decorrelation analysis to decode high-fidelity physiological signatures from dynamic speckle patterns, while overcoming hardware-level SNR limitations through closed-loop feedback and spatial subdivision of the sensing field.SUMMARY
[0009] In view of the foregoing, there is provided a system for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment. The system comprises a coherent light source, an optical sensor and a processing unit. The coherent light source is configured to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue. The optical sensor is configured to capture high-frequency image frames of light reflected from the biological tissue at a sampling frequency of at least 200 Hz, wherein the high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity. The processing unit is communicatively coupled to the optical sensor and the coherent light source. The processing unit is configured to execute instructions to: (i) transform the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer at least one of velocity dynamics or pressure dynamics of the biological tissue from temporal decorrelation of the dynamic speckle patterns; (ii) optimize a quality of the time-series signal through a closed-loop feedback mechanism by processing each sub-region to determine a signal-to-noise ratio (SNR), computing a Quality Assurance (QA) score for the time-series signal of each sub-region based on the SNR and a measure of signal consistency comprising one or more of temporal stability, waveform repeatability, or amplitude variance across consecutive segments, selecting a sub-region yielding the highest SNR among the plurality of sub-regions for feature extraction, and dynamically adjusting at least one hardware parameter of the coherent light source or the optical sensor in real-time based on the QA score to maintain the SNR of the selected sub-region above apredetermined threshold; (iii) apply at least one of a statistical model or a machine learning (ML) model to a multidimensional feature set extracted from the optimized time-series signal, wherein the multidimensional feature set comprises at least two of non-linear entropy features, frequency-domain spectral densities, and time-domain statistical features, wherein the statistical model or ML model is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers; and (iv) derive the plurality of clinical physiological parameters from the acoustic and hemodynamic markers output by the statistical model or ML model. The parameters comprise at least one of cuffless blood pressure, heart sound components SI through S4, indicators of turbulent blood flow, respiratory sounds, respiratory rate, heart rate, or blood oxygen saturation. The system is configured to generate an output comprising the clinical physiological parameters for use in physiological monitoring.
[0010] In some embodiments, the processing unit is configured to dynamically adjust the hardware parameter by modulating the optical power of the coherent light source within a range of 0.01 mW to 5 mW to compensate for variations in tissue absorption or ambient light interference.
[0011] In some embodiments, the the system applies the machine learning (ML) model, and the ML model is optimized via transfer learning, comprising a baseline model trained on a first dataset of biovital waveform signals including one or more of phonocardiogram (PCG) signals, photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, or digital stethoscope acoustic recordings, subsequently fine-tuned using a second dataset of biophotonic signals captured by the optical sensor, and wherein the machine learning model includes a synthetic data generation layer configured to address class imbalance by generating synthetic signal segments using a synthetic minority oversampling technique (SMOTE) or a generative model.
[0012] In some embodiments, the processing unit is further configured to perform normative modeling by comparing the derived clinical physiological parameters against a population-based statistical benchmark adjusted for user-specific covariates comprising at least two of age, sex, height, weight, body mass index (BMI), ethnicity, Fitzpatrick skin phototype, or smoking status.
[0013] In some embodiments, the processing unit is further configured to perform longitudinal analysis by tracking deviations in the clinical physiological parameters over a plurality of time points to generate a prognostic score indicating one or more of a healthtrajectory, progression or regression of a detected condition, a response to a therapeutic intervention, or a predicted effectiveness of a therapeutic strategy.
[0014] In some embodiments, the system further comprises at least one motion sensor comprising an accelerometer or a gyroscope configured to detect motion of a subject, operator, system, or host device, and the processing unit is configured to utilize data from the motion sensor to perform artifact correction by filtering out signal components corresponding to said detected motion from the quantified tissue surface motion dynamics.
[0015] In some embodiments, the processing unit is configured to perform automated labeling of the time-series signal by applying a machine learning segmentation model trained to identify heart sound components SI through S4 and detect and classify cardiac murmurs.
[0016] In some embodiments, the processing unit is further configured to perform biometric identification by extracting a unique biovital profile of the subject comprising a heart sound signature, cardiac rhythm signature, vascular pulse waveform signature, or respiratory pattern signature and matching said biovital profile against a registered biometric database.
[0017] In some embodiments, the system is integrated into, mounted on, or embedded within one or more of a wearable device, a wearable garment or accessory, a portable electronic device, a stationary object or fixture, a vehicle or vehicle component, a piece of fitness equipment, or a medical wellness device.
[0018] In some embodiments, the processing unit is configured to generate an automated diagnostic report comprising derived clinical physiological parameters, a severity stratification for detected cardiovascular or respiratory conditions, and personalized therapeutic suggestions.
[0019] In some embodiments, the optical sensor comprises one or more of a CCD sensor, a CMOS sensor, an sCMOS sensor, an optical flow sensor, a near-infrared (NIR) sensor, a short-wave infrared (SWIR) sensor, a microbolometer array, a thermal imaging sensor, or an event-based vision sensor. In some embodiments, the system further comprises an optical assembly including one or more of a refractive lens, a diffractive optical element (DOE), a metalens, a metasurface, a holographic optical element (HOE), a Fresnel lens, a GRIN lens, a microlens array, an optical filter, a polarizer, or a beam splitter.
[0020] In one aspect, there is provided a method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment. The method comprises: (i) emitting,via a coherent light source, light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue; (ii) capturing, via an optical sensor, high-frequency image frames of light reflected from the biological tissue at a sampling frequency of at least 200 Hz to encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity; (iii) transforming, via a processing unit, the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of sub-regions and independently determining spatial-temporal energy distribution changes for each subregion to infer at least one of velocity dynamics or pressure dynamics of the biological tissue from temporal decorrelation of the dynamic speckle patterns; (iv) optimizing a quality of the time-series signal through a closed-loop feedback mechanism by determining a signal-to-noise ratio (SNR), computing a Quality Assurance (QA) score for each subregion, selecting a sub-region yielding the highest SNR, and dynamically adjusting at least one hardware parameter of the light source or sensor in real-time based on the QA score; (v) applying a statistical model or machine learning (ML) model to a multidimensional feature set extracted from the optimized time-series signal to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers; and (vi) deriving the plurality of clinical physiological parameters, including cuffless blood pressure, heart sound components SI through S4, or indicators of turbulent blood flow, and generating an output comprising the clinical physiological parameters for use in physiological monitoring.
[0021] In some embodiments, the dynamic adjustment of the hardware parameter comprises modulating the optical power of the coherent light source within a range of 0.01 mW to 5 mW.
[0022] In some embodiments, the ML model is optimized via transfer learning from a baseline model trained on a first dataset of biovital waveform signals and subsequently fine-tuned using a second dataset of biophotonic signals, and wherein the machine learning model includes a synthetic data generation layer to address class imbalance via SMOTE or a generative model.
[0023] In some embodiments, the method further comprises performing normative modeling by comparing the derived clinical physiological parameters against a populationbased statistical benchmark adjusted for user-specific covariates.
[0024] In some embodiments, the method includes performing longitudinal analysis by tracking deviations in the clinical physiological parameters over time to generate a prognostic score.
[0025] In some embodiments, the method further comprises utilizing data from at least one motion sensor to perform artifact correction by filtering out signal components corresponding to detected motion of a subject, operator, system, or host device. In some embodiments, the method includes performing automated labeling of the time-series signal by applying a machine learning segmentation model to identify heart sound components SI through S4 and classify cardiac murmurs.
[0026] In some embodiments, the method includes performing biometric identification by extracting a unique biovital profile and matching it against a registered biometric database. In some embodiments, the method is performed using a system integrated into wearable garments, portable electronic devices, stationary fixtures, vehicles, or fitness equipment.
[0027] In another aspect, there is provided a non-transitory computer-readable medium storing instructions for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment that, when executed by a processing unit, cause the processing unit to perform operations comprising: (i) controlling a coherent light source to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue; (ii) commanding an optical sensor to capture high-frequency image frames encoding dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity; (iii) transforming the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of discrete sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer velocity or pressure dynamics from temporal decorrelation of the dynamic speckle patterns; (iv) optimizing a quality of the time-series signal through a closed-loop feedback mechanism by computing Quality Assurance (QA) scores and dynamically adjusting at least one hardware parameter of the coherent light source or the optical sensor in real-time based on the QA scores; (v) applying at least one of a statistical model or a machine learning (ML) model to a multidimensional feature set to map quantified tissue surface motion dynamics to acoustic and hemodynamic markers; and (vi) deriving the plurality of clinicalphysiological parameters from the acoustic and hemodynamic markers to generate an output for physiological monitoring.
[0028] The method, system, and computer-readable medium provide several significant technical benefits over conventional physiological monitoring through their innovative hardware-software integration. By analyzing the temporal decorrelation of dynamic speckle patterns within reflected biophotonic signals at high sampling frequencies (at least 200 Hz), the system captures high-fidelity cutaneous and sub-dermal vibrations that are invisible to standard photoplethysmography (PPG). Unlike PPG, which relies on bulk changes in light absorption, the system utilizes motion description modelling, such as, akin to high-performance optical tracking, to quantify tissue surface dynamics (velocity and pressure) directly. This technical approach establishes a high-fidelity pathway to derive complex biovitals, including cuffless blood pressure estimates reaching Grade A accuracy, detailed heart sound components SI through S4, and specific indicators of turbulent blood flow, which have historically presented significant technical hurdles for non-invasive optical sensing.
[0029] A primary technical advantage of the system is the resolution of the spatial mismatch between the laser beam diameter (typically 2 mm to 5 mm) and target vascular structures. By sub-dividing the sensor’s field of view (FOV) into a plurality of discrete subregions and independently processing each for signal-to-noise ratio (SNR) optimization, the system identifies and isolates information-rich regions from surrounding non-dynamic tissue. This is further enhanced by a closed-loop feedback mechanism that dynamically adjusts hardware parameters, such as laser power or sensor gain, in real-time based on a Quality Assurance (QA) score, ensuring robust signal acquisition across diverse skin phototypes and demographics without the need for manual user calibration.
[0030] Furthermore, the integration of transfer learning architecture provides a unique technical bridge between distinct sensing modalities, leveraging established acoustic datasets (e.g., phonocardiograms) to optimize biophotonic signal interpretation. This allows for high-fidelity diagnostic modeling even with minimal initial device-specific data, facilitating normative modeling and longitudinal analysis that can track subtle health trajectories and responses to therapeutic interventions over time. The system’s versatility enables it to capture these clinical-grade signals through thin layers of clothing and across diverse form factors, ranging from wearables like rings and watches to integration into stationary objects like vehicle components or furniture, offering a comprehensive sensingsolution that is well-suited for continuous, non-invasive health management and personalized medicine.
[0031] These and other aspects of the embodiments herein will be better appreciated and understood when considered in conjunction with the following description and the accompanying drawings. It should be understood, however, that the following descriptions, while indicating preferred embodiments and numerous specific details thereof, are given by way of illustration and not of limitation. Many changes and modifications may be made within the scope of the embodiments herein without departing from the spirit thereof, and the embodiments herein include all such modifications.BRIEF DESCRIPTION OF DRAWINGS
[0032] Implementations of the disclosure will now be described, by way of example only, with reference to the accompanying drawings, in which:
[0033] FIG. 1 is a block diagram that illustrates a system for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment according to some embodiments herein;
[0034] FIG. 2 illustrates an exploded view of the system of FIG. 1 according to some embodiments herein;
[0035] FIGS. 3A & 3B are data flow diagrams illustrating a method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment according to some embodiments herein; and
[0036] FIG. 4 is a representative hardware environment for practicing the embodiments herein with respect to FIG. 1 through 3B in accordance with the embodiments herein.DETAILED DESCRIPTION OF THE DRAWINGS
[0037] The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein.Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
[0038] As mentioned, there remains a need for a system and a method deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment. Various embodiments disclosed herein provide a system and a method deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment. Referring now to the drawings, and more particularly to FIGS. 1 through 4, where similar reference characters denote corresponding features consistently throughout the figures, preferred embodiments are shown.
[0039] FIG. 1 is a block diagram that illustrates a system for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment according to some embodiments herein. The system 100 includes a coherent light source 102, an optical sensor 104, and a processing unit 108 communicatively connected to the coherent light source and the optical sensor 104. The coherent light source 102 is configured to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue 106. In some embodiments, the coherent light source 102 is a laser module that emits light at an optical power ranging from 0.01 mW to 5 mW, enabling the capture of physiological data directly from the skin or through thin layers of clothing.
[0040] The optical sensor 104 is configured to capture high-frequency image frames of light reflected from the biological tissue 106 at a sampling frequency of at least 200 Hz. The high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity.
[0041] In some embodiments, the optical sensor 104 includes one or more of: (a) a charge-coupled device (CCD) image sensor, (b) a complementary metal-oxide-semiconductor (CMOS) image sensor, (c) a scientific complementary metal-oxide-semiconductor (sCMOS) image sensor, (d) an optical flow sensor of the type used in motion-tracking peripherals, (e) a near-infrared (NIR) image sensor responsive to wavelengths between approximately 700 nm and 1100 nm, (f) a short-wave infrared (SWIR) image sensor comprising an indium gallium arsenide (InGaAs) detector array, responsive to wavelengths between approximately 1100 nm and 2500 nm, (g) a mid-wave or long-wave infrared (MWIR / LWIR) image sensor comprising a microbolometer orpyroelectric detector array, (h) a thermal imaging sensor, (i) an event-based or neuromorphic vision sensor, or (j) a combination thereof.
[0042] In some embodiments, the optical flow sensor typically utilized in high-performance motion-tracking peripherals. Unlike conventional CMOS image sensors that prioritize static image fidelity, the optical sensor 104 of the system 100 is technically superior for dynamic speckle pattern capture due to its integrated motion description modeling hardware logic. The optical sensor 104 is designed for high-speed, sub-pixel displacement tracking, which provides the requisite sensitivity to detect the minute temporal decorrelation of speckle patterns caused by cutaneous vibrations. This specialized hardware architecture enables the capture of high-frequency reflected light at sampling rates exceeding 200 Hz with minimal computational overhead, as the optical sensor 104 itself is optimized to identify and quantify spatial shifts in texture, the interference patterns of the coherent light, rather than merely recording pixel intensities. This technical configuration is critical for isolating high-fidelity physiological signatures from the optical noise inherent in sub-dermal sensing.
[0043] The processing unit 108 is configured to execute specialized instructions to transform the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics. The processing unit 108 sub-divides a field of view (FOV) of the optical sensor 104 into a plurality of discrete sub-regions and independently determines spatial-temporal energy distribution changes for each sub-region to infer velocity dynamics or pressure dynamics of the biological tissue 106 from temporal decorrelation of the dynamic speckle patterns. The subdivision of the FOV allows the system 100 to overcome technical hurdles where the diameter of the laser beam (e.g., 2 to 5 mm) exceeds the diameter of target vascular structures, thereby identifying regions with optimal signal-to-noise ratio (SNR).
[0044] In some embodiments, the transformation of high-frequency image frames into a time-series signal is characterized by the execution of a motion description logic across consecutive image frames. The processing unit 108 is configured to perform an optical flow analysis or a related motion estimation technique to quantify sub-pixel displacements between consecutive frames within each sub-region. By applying at least one of a machine learning model or a statistical heuristic to these consecutive frames, the system 100 determines the spatial -temporal energy distribution changes that represent the mechanical vibrations of the tissue surface. This multi-frame analysis enables the system 100 to differentiate between the high-frequency temporal decorrelation caused byphysiological activity and gross movement, thereby providing a high-fidelity input for deriving clinical physiological parameters.
[0045] The processing unit 108 further configured to optimize a quality of the timeseries signal through a closed-loop feedback mechanism. The closed-loop feedback mechanism operates by: (i) processing each sub-region to determine an SNR; (ii) computing a Quality Assurance (QA) score based on the SNR and a measure of signal consistency comprising temporal stability, waveform repeatability, or amplitude variance; (iii) selecting a sub-region yielding the highest SNR; and (iv) dynamically adjusting at least one hardware parameter of the coherent light source 102 or the optical sensor 104 in realtime based on the QA score to maintain the SNR above a predetermined threshold. In some embodiments, the processing unit 108 dynamically adjusts hardware parameters by modulating the optical power of the coherent light source 102 within a range of 0.01 mW to 5 mW to compensate for variations in tissue absorption.
[0046] In some embodiments, the processing unit 108 is further configured to perform automated labeling of the time-series signal by applying a machine learning segmentation model trained to (i) identify and label one or more heart sound components selected from SI, S2, S3, and S4, and (ii) detect and classify cardiac murmurs. The cardiac murmurs include at least one of murmurs associated with turbulent blood flow across one or more heart valves, murmurs associated with valvular stenosis or regurgitation, murmurs associated with blood flow interaction with atherosclerotic plaques in arterial vasculature, murmurs associated with arterial stenosis, or continuous murmurs associated with abnormal vascular shunting.
[0047] The processing unit 108 is further configured to extract a feature set comprising non-linear entropy features, frequency-domain spectral densities, and timedomain statistical features from the optimized signal. The processing unit 108 then applies a statistical model or a machine learning (ML) model to the multidimensional feature set to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers. The system 100 derives clinical physiological parameters comprising at least one of cuffless blood pressure, heart sound components SI through S4, indicators of turbulent blood flow, respiratory sounds, or blood oxygen saturation.
[0048] In some embodiments, the ML model is optimized via transfer learning, comprising a baseline model trained on a first dataset of biovital waveform signals subsequently fine-tuned using a second dataset of biphotonic signals captured by the optical sensor 104. The first dataset includes one or more of phonocardiogram (PCG) signals,photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, respiratory sound signals, capnography signals, ballistocardiogram (BCG) signals, seismocardiogram (SCG) signals, impedance cardiography (ICG) signals, arterial blood pressure waveforms, pulse wave velocity or pulse transit time waveforms, laser Doppler or Doppler ultrasound velocity waveforms, near-infrared spectroscopy (NIRS) waveforms, mechanocardiogram or vibrocardiogram signals, accelerometer-derived cardiac or respiratory waveforms, and digital stethoscope acoustic recordings. The machine learning model includes a synthetic data generation layer configured to address class imbalance by generating synthetic signal segments using a synthetic minority oversampling technique (SMOTE) or a generative model.
[0049] In some embodiments, the system 100 further comprises at least one motion sensor, such as an accelerometer or a gyroscope, configured to detect motion of the subject or the system 100 itself. The processing unit 108 utilizes data from the motion sensor to perform artifact correction by filtering out motion-related signal components from the quantified tissue surface motion dynamics. In some embodiments, the processing unit 108 is configured to perform normative modeling by comparing derived parameters against population-based statistical benchmarks adjusted for user-specific covariates such as age, BMI, or Fitzpatrick skin phototype. In some embodiments, the processing unit 108 performs longitudinal analysis by tracking deviations over time to generate a prognostic score indicating a health trajectory or a response to therapeutic intervention.
[0050] In some embodiments, the system 100 is integrated into a wearable device such as a watch, a ring, an ear-worn device, or a skin-adhering patch. In some embodiments, the system 100 is integrated into a vehicle component, such as a steering wheel or a seat, or a stationary object such as a smart mirror or a chair. The processing unit 108 is further configured to execute a diagnostic output generator 216 to produce an automated diagnostic report including (i) the derived clinical physiological parameters, (ii) a severity stratification for one or more detected conditions selected from cardiovascular conditions, respiratory conditions, peripheral arterial disease, hypertension, and heart valve conditions, and (iii) personalized therapeutic suggestions.
[0051] The processing unit 108 is further configured to perform biometric identification by extracting a unique biovital profile of the subject from the quantified tissue surface motion dynamics. The biovital profile includes one or more of a heart sound signature, a cardiac rhythm signature, a vascular pulse waveform signature, a respiratorypattern signature, or a combination thereof, and matching said biovital profile against a registered biometric database.
[0052] In some embodiments, the system 100 is integrated into, mounted on, or embedded within one or more of (a) a wearable device worn directly on a body of the subject, comprising a watch, a wristband, a bracelet, a ring, an armband, an ankle band, an ear-worn device, an earbud, a hearing aid, a headphone, a headband, a strap, a helmet, a choker, a necklace, eyeglasses, or a wearable medical device, (b) a wearable garment or accessory, comprising a shirt, an undergarment, a brassiere, a glove, a sock, a shoe, a belt, or a skin-adhering patch, (c) a portable electronic device, comprising a mobile phone, a tablet, a laptop computer, a smartwatch, or a computer peripheral including a computer mouse or keyboard, (d) a stationary object or fixture, comprising a chair, a bed, a mattress, a pillow, a toilet, a toilet seat, a desk, a table, a door, a mirror, or a smart mirror, (e) a vehicle or vehicle component, comprising a steering wheel, a seat, a seatbelt, a dashboard, or an interior surface of an automobile, (f) a piece of fitness or exercise equipment, comprising a treadmill, a stationary bicycle, a rowing machine, or a weight-training apparatus, or (g) a medical or wellness device, comprising a stethoscope, a blood pressure monitor, an examination table, or a hospital bed.
[0053] In some embodiments, the processing unit 108 is configured to generate an automated diagnostic report comprising (i) the derived clinical physiological parameters, (ii) a severity stratification for one or more detected conditions selected from cardiovascular conditions, respiratory conditions, peripheral arterial disease, hypertension, and heart valve conditions, and (iii) personalized therapeutic suggestions.
[0054] FIG. 2 illustrates an exploded view of the system 100 of FIG. 1 according to some embodiments herein. The system 100 includes a coherent light source 102, an optical sensor 104, an optical assembly 202, a motion sensor 204, and a processing unit 108, which further comprises a spatio-temporal transformation module 206, a closed-loop feedback controller 208, a multidimensional feature extractor 210, a statistical model or a machine learning (ML) model 212, a clinical parameter derivator 214, and a diagnostic output generator 216. The coherent light source 102 is operably coupled to the processing unit 108 to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue 106. The optical sensor 104 is configured to capture high-frequency image frames of light reflected from the biological tissue 106 at a sampling frequency of at least 200 Hz to encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity. In some embodiments, the optical sensor 104 comprises an opticalflow sensor of the type used in motion-tracking peripherals, a CMOS sensor, or an eventbased vision sensor. In some embodiments, the optical assembly 202 is disposed in the optical path between the biological tissue 106 and the optical sensor 104 to focus or shape reflected light using a diffractive optical element (DOE), a metalens, or a polarizer.
[0055] In some embodiments, the optical assembly 202 includes one or more of (a) a refractive lens or lens stack, (b) a diffractive optical element (DOE), (c) a metalens comprising a sub -wavelength nanostructured array configured to focus, collimate, or shape the reflected light, (d) a metasurface configured to perform polarization filtering, wavefront shaping, or spectral filtering of the reflected light, (e) a holographic optical element (HOE), (f) a Fresnel lens, (g) a gradient-index (GRIN) lens, (h) a microlens array, (i) a bandpass, longpass, shortpass, or notch optical filter configured to selectively transmit one or more wavelength bands, (j) a polarizer or polarization-sensitive element, (k) a beam splitter or beam combiner; or (1) a combination thereof.
[0056] In some embodiments, the motion sensor 204, comprising an accelerometer or a gyroscope, provides data to the processing unit 108 to perform artifact correction by filtering out signal components corresponding to detected motion of the subject or the system.
[0057] The spatio-temporal transformation module 206 is configured to transform the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics. This is achieved by sub-dividing a field of view (FOV) of the optical sensor 104 into a plurality of discrete sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer velocity or pressure dynamics from temporal decorrelation of the speckle patterns. The spatial subdivision (e.g., segmenting a 100x100 resolution into 25x25 FOV frames) enables the system 100 to identify regions with an optimal signal-to-noise ratio (SNR) despite the laser beam diameter (2 to 5 mm) exceeding the size of target vascular structures.
[0058] The resulting time-series signal, generated by the spatio-temporal transformation module 206, constitutes a biophotonic-based biosignal representing the quantified tissue surface motion dynamics. This biosignal differs from conventional PPG in that it is a physical representation of tissue velocity and pressure derived from the temporal decorrelation of optical speckle, rather than a measure of bulk light absorption. This biosignal effectively bridges the gap between raw optical data and clinical hemodynamics, serving as a high-fidelity digital proxy for the mechanical activities of the cardiovascular system, such as heart valve closure and turbulent blood flow. By mappingthis quantified motion directly to hemodynamic markers, the system 100 achieves a pathway to derive clinical parameters, specifically cuffless blood pressure, with clinical-grade (e.g., Grade A) accuracy without requiring the user-specific calibration typically necessitated by existing non-invasive technologies. s
[0059] The closed-loop feedback controller 208 is configured to optimize signal quality through a closed-loop feedback mechanism by calculating a Quality Assurance (QA) score for each sub-region based on the SNR and signal consistency, selecting the subregion yielding the highest SNR, and dynamically adjusting hardware parameters of the coherent light source 102 or optical sensor 104 in real-time. In some embodiments, the hardware adjustment includes modulating the optical power of the coherent light source 102 within a range of 0.01 mW to 5 mW to compensate for tissue absorption variations.
[0060] The multidimensional feature extractor 210 is configured to extract a feature set comprising at least two of non-linear entropy features, frequency-domain spectral densities, and time-domain statistical features from the optimized signal. The statistical model or ML model 212 is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers. In some embodiments, the ML model 212 is optimized via transfer learning, utilizing a baseline model trained on biovital waveform signals (e.g., PCG or digital stethoscope recordings) and fine-tuned using biphotonic signals, and may include a synthetic data generation layer (e.g., SMOTE) to address class imbalance. The clinical parameter derivator 214 derives the plurality of clinical physiological parameters, including cuffless blood pressure, heart sound components SI through S4, and indicators of turbulent blood flow. In some embodiments, the processing unit 108 performs automated labeling to identify heart sound components and classify murmurs associated with valvular stenosis or atherosclerotic plaques. In some embodiments, the processing unit 108 performs normative modeling against population benchmarks adjusted for user-specific covariates such as age, BMI, and Fitzpatrick skin phototype, or performs longitudinal analysis to generate a prognostic score. In some embodiments, the system 100 performs biometric identification by matching a unique biovital profile against a registered database. The diagnostic output generator 216 produces an automated diagnostic report comprising the derived parameters, a severity stratification for detected conditions, and personalized therapeutic suggestions.
[0061] FIGS. 3 A & 3B are data flow diagrams illustrating a method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment accordingto some embodiments herein. At step 302, light is emitted within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue 106 via a coherent light source 102. At step 304, high-frequency image frames of light reflected from the biological tissue 106 are captured at a sampling frequency of at least 200 Hz via an optical sensor 104. The high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity.
[0062] At step 306, the high-frequency image frames are transformed into a timeseries signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor 104 into a plurality of sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer at least one of velocity dynamics or pressure dynamics of the biological tissue 106 from temporal decorrelation of the dynamic speckle patterns via a processing unit 108. The subdivision (e.g., segmenting a 100x100 resolution into 25x25 frames) allows the system 100 to identify information-rich regions with an optimal signal -to-noise ratio (SNR) despite a laser beam diameter (2 to 5 mm) that may exceed the size of target vascular structures.
[0063] At step 308, a quality of the time-series signal is optimized through a closed-loop feedback mechanism via a processing unit 108 by: (i) processing each sub-region to determine a signal-to-noise ratio (SNR); (ii) computing a Quality Assurance (QA) score for the time-series signal of each sub-region based on the SNR and a measure of signal consistency comprising temporal stability, waveform repeatability, or amplitude variance; (iii) selecting a sub-region yielding the highest SNR among the plurality of sub-regions for feature extraction; and (iv) dynamically adjusting at least one hardware parameter of the coherent light source 102 or the optical sensor 104 in real-time based on the QA score to maintain the SNR of the selected sub-region above a predetermined threshold.
[0064] At step 310, at least one of a statistical model or a machine learning (ML) model 212 is applied to a multidimensional feature set extracted from the optimized timeseries signal via a processing unit 108. The multidimensional feature set includes at least two of non-linear entropy features, frequency-domain spectral densities, and time-domain statistical features, and the statistical model or ML model 212 is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers. In some embodiments, the ML model 212 is optimized via transfer learning, including a baseline model trained on a first dataset of biovital waveform signals (e.g., PCG signals or digital stethoscope recordings) subsequently fine-tuned using a second dataset of biphotonicsignals, and includes a synthetic data generation layer configured to address class imbalance using SMOTE or a generative model.
[0065] At step 312, the plurality of clinical physiological parameters are derived from the acoustic and hemodynamic markers output by the statistical model or ML model 212 via a processing unit 108. The parameters include at least one of cuffless blood pressure, heart sound components SI through S4, indicators of turbulent blood flow, respiratory sounds, respiratory rate, heart rate, or blood oxygen saturation. In some embodiments, the method further comprises performing normative modeling by comparing derived clinical physiological parameters against a population-based statistical benchmark adjusted for user-specific covariates such as age, sex, BMI, and Fitzpatrick skin phototype. In some embodiments, the method includes performing longitudinal analysis by tracking deviations in clinical physiological parameters over a plurality of time points to generate a prognostic score.
[0066] In some embodiments, the method includes performing automated labeling of the time-series signal to identify SI through S4 heart sounds and classify cardiac murmurs. In some embodiments, the method includes performing biometric identification by matching a unique biovital profile against a registered biometric database. In some embodiments, the method utilizes data from at least one motion sensor 204 to perform artifact correction by filtering out signal components corresponding to detected motion of the subject, operator, or system.
[0067] At step 314, an output including the clinical physiological parameters for use in physiological monitoring is generated via the processing unit 108. In some embodiments, the output comprises an automated diagnostic report including a severity stratification for detected cardiovascular or respiratory conditions and personalized therapeutic suggestions. In some embodiments, the method is performed using the system 100 integrated into a wearable device, a garment, a portable electronic device, a stationary fixture, or a vehicle component.
[0068] A representative hardware environment for practicing the embodiments herein is depicted in FIG. 4, with reference to FIGS. 1 through 3B. This schematic drawing illustrates a hardware configuration of a server / computer system / computing device in accordance with the embodiments herein. The system includes at least one processing device CPU 10 that may be interconnected via system bus 14 to various devices such as a random access memory (RAM) 12, read-only memory (ROM) 16, and an input / output (I / O) adapter 18. The I / O adapter 18 can connect to peripheral devices, such as disk units 38 andprogram storage devices 40 that are readable by the system. The system can read the inventive instructions on the program storage devices 40 and follow these instructions to execute the methodology of the embodiments herein. The system further includes a user interface adapter 22 that connects a keyboard 28, mouse 30, speaker 32, microphone 34, and / or other user interface devices such as a touch screen device (not shown) to the bus 14 to gather user input. Additionally, a communication adapter 20 connects the bus 14 to a data processing network 42, and a display adapter 24 connects the bus 14 to a display device 26, which provides a graphical user interface (GUI) 36 of the output data in accordance with the embodiments herein, or which may be embodied as an output device such as a monitor, printer, or transmitter, for example.
[0069] The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and / or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the scope of appended claims.
Claims
CLAIMSWhat is claimed is:
1. A system for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment, the system comprising:a coherent light source configured to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue;an optical sensor configured to capture high-frequency image frames of light reflected from the biological tissue at a sampling frequency of at least 200 Hz, wherein the high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity; anda processing unit communicatively coupled to the optical sensor and the coherent light source, wherein the processing unit is configured to execute instructions to:(a) transform the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer at least one of velocity dynamics or pressure dynamics of the biological tissue from temporal decorrelation of the dynamic speckle patterns;(b) optimize a quality of the time-series signal through a closed-loop feedback mechanism by (i) processing each sub-region to determine a signal-to- noise ratio (SNR), (ii) computing a Quality Assurance (QA) score for the timeseries signal of each sub-region based on the signal-to-noise ratio (SNR) and a measure of signal consistency comprising one or more of: temporal stability, waveform repeatability, or amplitude variance across consecutive segments of the time-series signal; (iii) selecting a sub-region yielding the highest SNR among the plurality of sub-regions for feature extraction, and (iv) dynamically adjusting at least one hardware parameter of the coherent light source or the optical sensor in real-time based on the QA score to maintain the SNR of the selected sub-region above a predetermined threshold;(c) apply at least one of a statistical model or a machine learning (ML) model to a multidimensional feature set extracted from the optimized time-series signal, wherein the multidimensional feature set comprises at least two of non-linearentropy features, frequency-domain spectral densities, and time-domain statistical features, wherein the statistical model or ML model is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers;(d) derive the plurality of clinical physiological parameters from the acoustic and hemodynamic markers output by the statistical model or ML model, wherein the parameters comprise at least one of cuffless blood pressure, heart sound components SI through S4, indicators of turbulent blood flow, respiratory sounds, respiratory rate, heart rate, or blood oxygen saturation,wherein the system is configured to generate an output comprising the clinical physiological parameters for use in physiological monitoring.
2. The system of claim 1, wherein the processing unit is configured to dynamically adjust the hardware parameter by modulating the optical power of the coherent light source within a range of 0.01 mW to 5 mW to compensate for variations in tissue absorption or ambient light interference.
3. The system of claim 1, wherein the system applies the machine learning (ML) model, and the ML model is optimized via transfer learning, comprising a baseline model trained on a first dataset of biovital waveform signals subsequently fine-tuned using a second dataset of biophotonic signals captured by the optical sensor, wherein the first dataset comprises one or more of: phonocardiogram (PCG) signals, photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, respiratory sound signals, capnography signals, ballistocardiogram (BCG) signals, seismocardiogram (SCG) signals, impedance cardiography (ICG) signals, arterial blood pressure waveforms, pulse wave velocity or pulse transit time waveforms, laser Doppler or Doppler ultrasound velocity waveforms, near-infrared spectroscopy (NIRS) waveforms, mechanocardiogram or vibrocardiogram signals, accelerometer-derived cardiac or respiratory waveforms, and digital stethoscope acoustic recordings; and wherein the machine learning model includes a synthetic data generation layer configured to address class imbalance by generating synthetic signal segments using a synthetic minority oversampling technique (SMOTE) or a generative model.
4. The system of claim 1, wherein the processing unit is further configured to perform normative modeling by comparing the derived clinical physiological parameters against apopulation-based statistical benchmark adjusted for user-specific covariates comprising at least two of: age, sex, height, weight, body mass index (BMI), body surface area (BSA), ethnicity, Fitzpatrick skin phototype, resting heart rate, resting blood pressure, and smoking status.
5. The system of claim 1, wherein the processing unit is further configured to perform longitudinal analysis by tracking deviations in the clinical physiological parameters over a plurality of time points to generate a prognostic score indicating one or more of: a health trajectory of the subject, a progression or regression of a detected condition, a response to a therapeutic intervention, or a predicted effectiveness of a therapeutic strategy.
6. The system of claim 1, further comprising at least one motion sensor comprising an accelerometer or a gyroscope, wherein the at least one motion sensor is configured to detect motion of (i) a subject from whom the biological tissue measurements are being acquired, (ii) an operator handling the system during measurement acquisition, (iii) the system itself, or (iv) a host device or platform to which the system is physically mounted, integrated, or communicatively coupled, and wherein the processing unit is configured to utilize data from the at least one motion sensor to perform artifact correction by filtering out signal components, corresponding to said detected motion from any one or more of (i) through (iv), from the quantified tissue surface motion dynamics.
7. The system of claim 1, wherein the processing unit is further configured to perform automated labeling of the time-series signal by applying a machine learning segmentation model trained to (i) identify and label one or more heart sound components selected from SI, S2, S3, and S4, and (ii) detect and classify cardiac murmurs, wherein the cardiac murmurs comprise at least one of murmurs associated with turbulent blood flow across one or more heart valves, murmurs associated with valvular stenosis or regurgitation, murmurs associated with blood flow interaction with atherosclerotic plaques in arterial vasculature, murmurs associated with arterial stenosis, or continuous murmurs associated with abnormal vascular shunting.
8. The system of claim 1, wherein the processing unit is further configured to perform biometric identification by extracting a unique biovital profile of the subject from the quantified tissue surface motion dynamics, wherein the biovital profile comprises one ormore of: a heart sound signature, a cardiac rhythm signature, a vascular pulse waveform signature, a respiratory pattern signature, or a combination thereof, and matching said biovital profile against a registered biometric database.
9. The system of claim 1, wherein the system is integrated into, mounted on, or embedded within one or more of:(a) a wearable device worn directly on a body of the subject, comprising a watch, a wristband, a bracelet, a ring, an armband, an ankle band, an ear-worn device, an earbud, a hearing aid, a headphone, a headband, a strap, a helmet, a choker, a necklace, eyeglasses, or a wearable medical device;(b) a wearable garment or accessory, comprising a shirt, an undergarment, a brassiere, a glove, a sock, a shoe, a belt, or a skin-adhering patch;(c) a portable electronic device, comprising a mobile phone, a tablet, a laptop computer, a smartwatch, or a computer peripheral including a computer mouse or keyboard;(d) a stationary object or fixture, comprising a chair, a bed, a mattress, a pillow, a toilet, a toilet seat, a desk, a table, a door, a mirror, or a smart mirror;(e) a vehicle or vehicle component, comprising a steering wheel, a seat, a seatbelt, a dashboard, or an interior surface of an automobile;(f) a piece of fitness or exercise equipment, comprising a treadmill, a stationary bicycle, a rowing machine, or a weight-training apparatus; or(g) a medical or wellness device, comprising a stethoscope, a blood pressure monitor, an examination table, or a hospital bed.
10. The system of claim 1, wherein the processing unit is configured to generate an automated diagnostic report comprising (i) the derived clinical physiological parameters, (ii) a severity stratification for one or more detected conditions selected from cardiovascular conditions, respiratory conditions, peripheral arterial disease, hypertension, and heart valve conditions, and (iii) personalized therapeutic suggestions.
11. The system of claim 1, wherein the optical sensor comprises one or more of:(a) a charge-coupled device (CCD) image sensor;(b) a complementary metal-oxide-semiconductor (CMOS) image sensor;(c) a scientific complementary metal-oxide-semiconductor (sCMOS) image sensor;(d) an optical flow sensor of the type used in motion-tracking peripherals;(e) a near-infrared (NIR) image sensor responsive to wavelengths between approximately 700 nm and 1100 nm;(f) a short-wave infrared (SWIR) image sensor comprising an indium gallium arsenide (InGaAs) detector array, responsive to wavelengths between approximately 1100 nm and 2500 nm;(g) a mid-wave or long-wave infrared (MWIR / LWIR) image sensor comprising a microbolometer or pyroelectric detector array;(h) a thermal imaging sensor;(i) an event-based or neuromorphic vision sensor; or(j) a combination thereof.
12. The system of claim 1, further comprising an optical assembly disposed in an optical path between the biological tissue and the optical sensor, wherein the optical assembly comprises one or more of:(a) a refractive lens or lens stack;(b) a diffractive optical element (DOE);(c) a metalens comprising a sub -wavelength nanostructured array configured to focus, collimate, or shape the reflected light;(d) a metasurface configured to perform polarization filtering, wavefront shaping, or spectral filtering of the reflected light;(e) a holographic optical element (HOE);(f) a Fresnel lens;(g) a gradient-index (GRIN) lens;(h) a microlens array;(i) a bandpass, longpass, shortpass, or notch optical filter configured to selectively transmit one or more wavelength bands;(j) a polarizer or polarization-sensitive element;(k) a beam splitter or beam combiner; or(l) a combination thereof.
13. A method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment, the method comprising:emitting, via a coherent light source, light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue;capturing, via an optical sensor, high-frequency image frames of light reflected from the biological tissue at a sampling frequency of at least 200 Hz, wherein the high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity;transforming, via a processing unit, the high-frequency image frames into a timeseries signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer at least one of velocity dynamics or pressure dynamics of the biological tissue from temporal decorrelation of the dynamic speckle patterns;optimizing a quality of the time-series signal through a closed-loop feedback mechanism by: (i) processing each sub-region to determine a signal-to-noise ratio (SNR); (ii) computing a Quality Assurance (QA) score for the time-series signal of each sub-region based on the signal-to-noise ratio (SNR) and a measure of signal consistency comprising one or more of: temporal stability, waveform repeatability, or amplitude variance across consecutive segments of the time-series signal; (iii) selecting a sub-region yielding the highest SNR among the plurality of sub-regions for feature extraction; and (iv) dynamically adjusting at least one hardware parameter of the coherent light source or the optical sensor in real-time based on the QA score to maintain the SNR of the selected sub-region above a predetermined threshold;applying at least one of a statistical model or a machine learning (ML) model to a multidimensional feature set extracted from the optimized time-series signal, wherein the multidimensional feature set comprises at least two of non-linear entropy features, frequency-domain spectral densities, and time-domain statistical features, and wherein the statistical model or ML model is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers;deriving the plurality of clinical physiological parameters from the acoustic and hemodynamic markers output by the statistical model or ML model, wherein the parameters comprise at least one of cuffless blood pressure, heart sound components SI through S4,indicators of turbulent blood flow, respiratory sounds, respiratory rate, heart rate, or blood oxygen saturation; andgenerating an output comprising the clinical physiological parameters for use in physiological monitoring.
14. The method of claim 13, wherein the dynamic adjustment of the hardware parameter comprises modulating the optical power of the coherent light source within a range of 0.01 mW to 5 mW to compensate for variations in tissue absorption or ambient light interference.
15. The method of claim 13, wherein the method applies the machine learning (ML) model, and the ML model is optimized via transfer learning, comprising a baseline model trained on a first dataset of biovital waveform signals subsequently fine-tuned using a second dataset of biophotonic signals captured by the optical sensor, wherein the first dataset comprises one or more of: phonocardiogram (PCG) signals, photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, respiratory sound signals, capnography signals, ballistocardiogram (BCG) signals, seismocardiogram (SCG) signals, impedance cardiography (ICG) signals, arterial blood pressure waveforms, pulse wave velocity or pulse transit time waveforms, laser Doppler or Doppler ultrasound velocity waveforms, near-infrared spectroscopy (NIRS) waveforms, mechanocardiogram or vibrocardiogram signals, accelerometer-derived cardiac or respiratory waveforms, and digital stethoscope acoustic recordings; and wherein the machine learning model includes a synthetic data generation layer configured to address class imbalance by generating synthetic signal segments using a synthetic minority oversampling technique (SMOTE) or a generative model.
16. The method of claim 13, further comprising performing normative modeling by comparing the derived clinical physiological parameters against a population-based statistical benchmark adjusted for user-specific covariates comprising at least two of: age, sex, height, weight, body mass index (BMI), body surface area (BSA), ethnicity, Fitzpatrick skin phototype, resting heart rate, resting blood pressure, and smoking status.
17. The method of claim 13, further comprising performing longitudinal analysis by tracking deviations in the clinical physiological parameters over a plurality of time pointsto generate a prognostic score indicating one or more of: a health trajectory of the subject, a progression or regression of a detected condition, a response to a therapeutic intervention, or a predicted effectiveness of a therapeutic strategy.
18. The method of claim 13, further comprising: detecting, via at least one motion sensor comprising an accelerometer or a gyroscope, motion of (i) a subject from whom the biological tissue measurements are being acquired, (ii) an operator handling a system performing the method during measurement acquisition, (iii) the system itself, or (iv) a host device or platform to which the system is physically mounted, integrated, or communicatively coupled; and performing artifact correction by filtering out signal components corresponding to said detected motion from any one or more of (i) through (iv) from the quantified tissue surface motion dynamics.
19. The method of claim 13, further comprising performing automated labeling of the timeseries signal by applying a machine learning segmentation model trained to (i) identify and label one or more heart sound components selected from SI, S2, S3, and S4, and (ii) detect and classify cardiac murmurs, wherein the cardiac murmurs comprise at least one of murmurs associated with turbulent blood flow across one or more heart valves, murmurs associated with valvular stenosis or regurgitation, murmurs associated with blood flow interaction with atherosclerotic plaques in arterial vasculature, murmurs associated with arterial stenosis, or continuous murmurs associated with abnormal vascular shunting.
20. The method of claim 13, further comprising performing biometric identification by extracting a unique biovital profile of the subject from the quantified tissue surface motion dynamics, wherein the biovital profile comprises one or more of: a heart sound signature, a cardiac rhythm signature, a vascular pulse waveform signature, a respiratory pattern signature, or a combination thereof, and matching said biovital profile against a registered biometric database.
21. The method of claim 13, wherein the method is performed using a system integrated into, mounted on, or embedded within one or more of:(a) a wearable device worn directly on a body of the subject, comprising a watch, a wristband, a bracelet, a ring, an armband, an ankle band, an ear-worn device, an earbud, ahearing aid, a headphone, a headband, a strap, a helmet, a choker, a necklace, eyeglasses, or a wearable medical device;(b) a wearable garment or accessory, comprising a shirt, an undergarment, a brassiere, a glove, a sock, a shoe, a belt, or a skin-adhering patch;(c) a portable electronic device, comprising a mobile phone, a tablet, a laptop computer, a smartwatch, or a computer peripheral including a computer mouse or keyboard;(d) a stationary object or fixture, comprising a chair, a bed, a mattress, a pillow, a toilet, a toilet seat, a desk, a table, a door, a mirror, or a smart mirror;(e) a vehicle or vehicle component, comprising a steering wheel, a seat, a seatbelt, a dashboard, or an interior surface of an automobile;(f) a piece of fitness or exercise equipment, comprising a treadmill, a stationary bicycle, a rowing machine, or a weight-training apparatus; or(g) a medical or wellness device, comprising a stethoscope, a blood pressure monitor, an examination table, or a hospital bed.
22. A non-transitory computer-readable medium storing instructions to perform a method for deriving a plurality of clinical physiological parameters via tissue surface motion description modeling for non-invasive physiological monitoring and diagnostic assessment that, when executed by a processing unit, cause the processing unit to perform the steps comprising:(a) controlling a coherent light source to emit light within a wavelength range of 400 nm to 2500 nm onto a surface of a biological tissue;(b) commanding an optical sensor to capture high-frequency image frames of light reflected from the biological tissue at a sampling frequency of at least 200 Hz, wherein the high-frequency image frames encode dynamic speckle patterns indicative of cutaneous, sub-dermal, or vascular physiological activity;(c) transforming the high-frequency image frames into a time-series signal representing quantified tissue surface motion dynamics by sub-dividing a field of view (FOV) of the optical sensor into a plurality of discrete sub-regions and independently determining spatial-temporal energy distribution changes for each sub-region to infer at least one of velocity dynamics or pressure dynamics of the biological tissue from temporal decorrelation of the dynamic speckle patterns;(d) optimizing a quality of the time-series signal through a closed-loop feedback mechanism by: (i) processing each sub-region to determine a signal-to-noise ratio (SNR); (ii) computing a Quality Assurance (QA) score for the time-series signal of each sub-region based on the signal-to-noise ratio (SNR) and a measure of signal consistency comprising one or more of: temporal stability, waveform repeatability, or amplitude variance across consecutive segments of the time-series signal; (iii) selecting a sub-region yielding the highest SNR among the plurality of sub-regions for feature extraction; and (iv) dynamically adjusting at least one hardware parameter of the coherent light source or the optical sensor in real-time based on the QA score to maintain the SNR of the selected sub-region above a predetermined threshold;(e) applying at least one of a statistical model or a machine learning (ML) model to a multidimensional feature set extracted from the optimized time-series signal, wherein the multidimensional feature set comprises at least two of non-linear entropy features, frequency-domain spectral densities, and time-domain statistical features, and wherein the statistical model or ML model is configured to map the quantified tissue surface motion dynamics to acoustic and hemodynamic markers;(f) deriving the plurality of clinical physiological parameters from the acoustic and hemodynamic markers output by the statistical model or ML model, wherein the parameters comprise at least one of cuffless blood pressure, heart sound components SI through S4, indicators of turbulent blood flow, respiratory sounds, respiratory rate, heart rate, or blood oxygen saturation; and(g) generating an output comprising the clinical physiological parameters for use in physiological monitoring.
