[go: up one dir, main page]

WO2025231209A1 - Prédiction non invasive de résistance à l'insuline à l'aide de données de capteur - Google Patents

Prédiction non invasive de résistance à l'insuline à l'aide de données de capteur

Info

Publication number
WO2025231209A1
WO2025231209A1 PCT/US2025/027242 US2025027242W WO2025231209A1 WO 2025231209 A1 WO2025231209 A1 WO 2025231209A1 US 2025027242 W US2025027242 W US 2025027242W WO 2025231209 A1 WO2025231209 A1 WO 2025231209A1
Authority
WO
WIPO (PCT)
Prior art keywords
level
wearer
computing device
wearable computing
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/US2025/027242
Other languages
English (en)
Inventor
Ahmed Mohamed Ibrahim Abdelhadi METWALLY
Javier L. Prieto
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Google LLC
Original Assignee
Google LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Google LLC filed Critical Google LLC
Publication of WO2025231209A1 publication Critical patent/WO2025231209A1/fr
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Measuring devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor or mobility of a limb
    • A61B5/1118—Determining activity level
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration or pH-value ; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid or cerebral tissue
    • A61B5/14532—Measuring characteristics of blood in vivo, e.g. gas concentration or pH-value ; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid or cerebral tissue for measuring glucose, e.g. by tissue impedance measurement
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/48—Other medical applications
    • A61B5/4806—Sleep evaluation
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
    • A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
    • A61B5/6802—Sensor mounted on worn items
    • A61B5/681—Wristwatch-type devices
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
    • G16H40/63—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for local operation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the present disclosure relates generally to the non-invasive prediction of a health risk using sensor data obtained from one or more sensors on a wearable device. More particularly, the present disclosure relates to predicting insulin resistance using sensor data obtained from sensors on the wearable device.
  • Insulin resistance is a common precursor to type 2 diabetes and most diagnosed individuals express IR. Insulin resistance represents a fundamental breakdown in the body's ability to regulate blood sugar levels. It occurs when cells in muscles, liver (hepatic), and fat (adipose) tissue become less responsive to the hormone insulin. Insulin's primary function is to shuttle glucose from the bloodstream into cells for energy utilization. With insulin resistance, this process becomes impaired, resulting in elevated blood sugar. Chronic insulin resistance puts the person at significant risk for prediabetes, type 2 diabetes, and heart disease. Contributing factors include excess body weight, particularly visceral fat, physical inactivity, and genetic predisposition. Insulin resistance may manifest subtly with high blood pressure or abnormal cholesterol profiles.
  • IR insulin resistance
  • the gold standard is the hyperinsulinemic euglycemic clamp, which is performed in research facilities but is expensive and time-consuming.
  • the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) and minimal model-based glucose and insulin measures are cheaper and faster alternatives that can be performed in clinical labs.
  • these methods each require a clinical lab visit, are still expensive, and cannot be performed regularly.
  • glucotyping via continuous glucose monitoring (CGM) and/or daily finger pricks are invasive tests and are typically not available as options for individuals who have not yet been diagnosed with diabetes but would like to monitor IR to look for trends and take preventative action before IR turns into type 2 diabetes.
  • a computing system to predict insulin resistance in a wearer of a wearable computing device includes a machine-learned model trained to predict insulin resistance in the wearer of the wearable computing device based at least in part on non-invasive biometric data associated with the wearer; one or more processors; and one or more non-transitory computer-readable media that collectively store instructions that, when executed by the one or more processors, cause the computing system to perform operations, the operations comprising: (i) receiving, via the one or more processors, the non-invasive biometric data from one or more sensors associated with the wearable computing device; and (ii) implementing, via the one or more processors, the machine-learned model to determine if the non-invasive biometric data associated with the wearer is indicative of insulin resistance.
  • the machine-learned model can be trained based at least in part on demographic training data, biometric training data, preexisting health training data, or a combination thereof.
  • the demographic training data can include body mass index, age, height, weight, gender, waist circumference, body composition, or a combination thereof.
  • the biometric training data can include data associated with resting heart rate, heart rate, heart rate variability, interbeat interval, sleep score, bedtime, wake time, sleep stage, sleep duration, step count, electrodermal activity (EDA), temperature, calories burned, active zone minutes, stress score, VO2 max level, respiratory rate, SpCh level, PPG waveform data, or a combination thereof.
  • EDA electrodermal activity
  • the preexisting health training data comprises data associated with a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alkaline phosphatase (ALP) level, an aspartate aminotransferase (AST) level, an alanine aminotransferase (ALT) level, or a combination
  • ALP
  • the preexisting health training data can include continuous glucose monitoring training data
  • the biometric training data can include raw photoplethysmogram training data
  • the machine-learned model can be further trained to predict insulin resistance in the wearer of the wearable computing device based on demographic data, preexisting health data, or both associated with the wearer.
  • the demographic data can include body mass index, age, height, weight, gender, waist circumference, body composition, or a combination thereof
  • the preexisting health data comprises data associated with a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alkaline phosphatase (ALP) level, an aspartate
  • ALP al
  • the non-invasive biometric data associated with the wearer can include data associated with resting heart rate, heart rate, heart rate variability, interbeat interval, sleep score, bedtime, wake time, sleep stage, sleep duration, step count, electrodermal activity (EDA), temperature, calories burned, active zone minutes, stress score, VO2 max level, respiratory rate, SpCh level, PPG waveform data, or a combination thereof.
  • EDA electrodermal activity
  • the operations can further include calculating, via the one or more processors, a predicted Homeostatic Model Assessment for Insulin Resistance (HOMA- IR) score for the wearer based at least in part on the biometric data associated with the wearer.
  • HOMA-IR Homeostatic Model Assessment for Insulin Resistance
  • a predicted HOMA-IR score at or below a predetermined first level can indicate that the wearer is insulin sensitive
  • a predicted HOMA-IR score at or above a predetermined second level can indicate that the wearer is insulin resistant
  • a predicted HOMA-IR score between the predetermined first level and the predetermined second level can indicate that the wearer is impaired insulin resistant.
  • the operations can further include displaying, via the one or more processors, a notification on a display of the computing system alerting the wearer of the predicted HOMA-IR score.
  • a computer-implemented method for prediction of insulin resistance of a wearer of a wearable computing device is provided that is capable of conducting the functionality described above with respect to the computing system.
  • a wearable computing device is provided that is capable of conducting the functionality described above with respect to the computing system.
  • FIG. 1 provides a front perspective view of a wearable computing device on a wrist of a user according to one embodiment of the present disclosure
  • FIG. 2 provides a rear perspective view of the wearable computing device of FIG. 1;
  • FIG. 3 illustrates various components of an example system that can be utilized according to one embodiment of the present disclosure
  • FIG. 4 provides a schematic diagram of an example set of devices that are able to communicate according to one embodiment of the present disclosure
  • FIG. 5 depicts a block diagram of an example computing system that can perform non-invasive predictions of insulin resistance-related health metrics for a patient according to example embodiments of the present disclosure
  • FIG. 6 depicts a block diagram of an example computing device that can perform non-invasive predictions of insulin resistance-related health metrics for a patient according to example embodiments of the present disclosure
  • FIG. 7 depicts a block diagram of an example computing device that can perform non-invasive predictions of insulin resistance-related health metrics for a patient according to example embodiments of the present disclosure
  • FIG. 8 depicts a flow chart diagram of components of an example machine-learned model for predicting insulin resistance in a wearer of a wearable computing device according to example embodiments of the present disclosure
  • FIG. 9A is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on non-invasive biometric data and demographic data associated with the wearer;
  • FIG. 9B is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on selected non-invasive biometric data and demographic data associated with the wearer;
  • FIG. 9C is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on non-invasive biometric data, demographic data, and fasting glucose (predetermined) associated with the wearer;
  • FIG. 9D is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on non-invasive biometric data, demographic data, and a lipid and complete metabolic panel (predetermined) associated with the wearer;
  • FIG. 9E is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on non-invasive biometric data associated with the wearer;
  • FIG. 9F is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA- IR scores were determined based on demographic data associated with the wearer;
  • FIG. 10 depicts chart diagram of an example, non-limiting computer-implemented method for predicting insulin resistance in a wearer of a wearable computing device according to example embodiments of the present disclosure.
  • HOMA-IR is a mathematical model that is utilized to assign a value to an individual related to that individual’s glucotype.
  • the individual can be labeled as insulin sensitive, impaired insulin resistant (early signs of insulin resistance), or insulin resistant based on the individual’s HOMA-IR score or value, where the HOMA-IR score or value is calculated according to the following formula (I):
  • HOMA-IR (Fasting Insulin (microliters/liter) x Fasting Glucose (mmol/Liter))/22.5 (I)
  • a HOMA-IR score or value of 1.1 or less is suggestive of insulin sensitivity
  • a HOMA-IR score or value between 1.1 and 2.9 is suggestive of impaired insulin resistance
  • a HOMA-IR score or value of 2.9 or greater is suggestive of insulin resistance.
  • the present disclosure contemplates utilizing a machine-learned model to predict a HOMA-IR score or value, and insulin sensitivity, impaired insulin resistance, or insulin resistance in individual in an efficient and non-invasive manner utilizing at least biometric data obtained from a wearable computing device. For instance, the present disclosure contemplates utilizing biometric data obtained from one sensors associated with or located on the wearable computing device to predict a HOMA-IR score or value for a wearer of the wearable computing device.
  • the one or more sensors can include a photoplethy smogram (PPG) sensor, a motion sensor, and additional sensor electrodes.
  • the non-invasive biometric data that can be obtained from such sensors can include data associated with resting heart rate, heart rate, heart rate variability, interbeat interval, sleep score, bedtime, wake time, sleep stage, sleep duration, step count, electrodermal activity (EDA), temperature, calories burned, active zone minutes, stress score, VO2 max level, respiratory rate, SpCh level, PPG waveform data, or a combination thereof, although it is also to be understood that other data can be utilized alternatively and/or in addition to the sensor data features listed.
  • EDA electrodermal activity
  • the resolution of the biometric data can be optimized to obtain the most accurate prediction of a HOMA-IR score or value.
  • the biometric data can be collected and averaged over a time period of at least about 1 day, such as at least about 1 week, such as at least about 1 month in order for the machine-learned model to output the best prediction based on the non-invasive biometric data input into the model.
  • the biometric data may first be input into a large sensor model (LSM) or a PPG foundational model, which are example methods for inputting the biometric data into the machine-learned model.
  • LSM large sensor model
  • PPG foundational model are example methods for inputting the biometric data into the machine-learned model.
  • a wearer of a wearable computing device contemplated by the present disclosure is deemed to be insulin sensitive, impaired insulin resistant, or insulin resistant under HOMA-IR standards can be predicted based at least in part on non-invasive biometric data collected by sensors contained on or within a wearable computing device, where such data is analyzed via a machine-learned model to predict a HOMA-IR score or value.
  • the HOMA-IR score or value is predicted, it is useful to notify the wearer of this information, such as via a notification on a display of the wearable device or another computing device (e.g., table, mobile phone, laptop, personal computer, etc.) so the wearer can be educated and/or make lifestyle, diet, and/or other changes, particularly when the wearer is predicted to be impaired insulin resistant or insulin resistant.
  • a notification on a display of the wearable device or another computing device e.g., table, mobile phone, laptop, personal computer, etc.
  • demographic data for the wearer can be used alone or in addition to the biometric data to predict whether the wearer of a wearable computing device contemplated by the present disclosure is deemed to be insulin sensitive, impaired insulin resistant, or insulin resistant under HOMA-IR standards.
  • the present inventors have found that use of the demographic data alone does not provide as accurate a prediction as the biometric data alone, the biometric data combined with the demographic data alone, or specific features of the biometric data and the demographic data combined together.
  • the demographic data associated with the wearer can include body mass index, age, height, weight, gender, waist circumference, body composition, or a combination thereof, although it is also to be understood that other demographic data can be utilized alternatively and/or in addition to the demographic data features listed.
  • preexisting health data for the wearer can be used alone or in addition to the biometric data and/or the demographic data for the wearer to predict whether the wearer of a wearable computing device contemplated by the present disclosure is deemed to be insulin sensitive, impaired insulin resistant, or insulin resistant under HOMA-IR standards.
  • the preexisting health data can include data associated with a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alkaline phosphatase (ALP) level, an aspartate aminotransferase (AST) level, an alanine aminotransferase (ALT) level, or a combination thereof, although it
  • the present disclosure contemplates utilizing specific biometric data and demographic data for the wearer to obtain an accurate prediction as to whether the wearer of a wearable computing device contemplated by the present disclosure is deemed to be insulin sensitive, impaired insulin resistant, or insulin resistant.
  • the biometric data can include a first biometric parameter and a second biometric parameter
  • the demographic data can include a first demographic data parameter to strike a balance between the amount of sensor and demographic data needed without sacrificing the accuracy of the prediction of insulin resistance.
  • the first biometric parameter can include resting heart rate (e.g., during sleep) and the second biometric parameter can include a sleep- related parameter (e.g., sleep score or amount of sleep).
  • the demographic data can include body mass index.
  • a third biometric parameter can include a step count, a number of active minutes, or any other feature related to physical activity.
  • a second demographic parameter can include age, gender, weight, waist circumference, height, or body composition.
  • the present disclosure is related to a wearable computing device, particularly a wearable computing device having a photoplethysmogram (PPG) sensor, a motion sensor, and additional sensor electrodes.
  • PPG photoplethysmogram
  • the data collected for the individual wearer or user is introduced into one or more machine-learned models via a computer- implemented method for accurately predicting insulin resistance in the wearer.
  • the present inventors have found that the particular models described herein can accurately predict insulin sensitivity, impaired insulin resistance, and insulin resistance in a non-invasive manner that does not require the user to undergo invasive testing (e.g., blood draws, finger pricks, etc.) or invasive monitoring (e.g., continuous glucose monitoring).
  • invasive testing e.g., blood draws, finger pricks, etc.
  • invasive monitoring e.g., continuous glucose monitoring
  • the PPG sensor(s) of the present disclosure use one or more emitters, such as light-emitting diodes (LEDs) to emit controlled pulses of light and use one or more detectors, such as photodiodes, to capture the returned light.
  • LEDs light-emitting diodes
  • detectors such as photodiodes
  • the emitted photons reflect off the skin, tissue, bones, blood, etc. of a wearer, and a processer controls the emitter(s) and converts the analog current received by the detector(s) into a digital PPG signal.
  • Signal changes associated with peripheral perfusion, due to contract of the heart enable the wearable computing device to measure the wearer’s pulsatility and heart rate, resting heart rate, interbeat interval, heart rate variability, etc. in the PPG signal.
  • the PPG sensors can be configured for use at various light wavelengths such as green (centered at 528 nanometers (nm)), red (centered at 660 nm), and infrared (centered at 940 nm), where the amplitude of the reflected light for the green, red, and infrared wavelengths changes with every heartbeat.
  • green centered at 528 nanometers (nm)
  • red centered at 660 nm
  • infrared centered at 940 nm
  • the peak to peak amplitude for green PPG signals is greater than the peak to peak amplitude for red PPG signals and infrared PPG signals when there is a clear pulsatile signal such as that associated with a heartbeat.
  • the motion sensor(s) of the present disclosure are configured for sensing and outputting movement data indicative of the motion of the wearer of the wearable monitoring device to calculate, inter alia, step count and other movement-related metrics.
  • step count it is to be understood that data collected by the motion sensor(s) can also be used to identify periods of relative stillness for determining sleep-related metrics.
  • the motion sensors can include one or more accelerometers for sensing movement data.
  • the wearable monitoring device can include one or more accelerometers for sensing acceleration or other movement data in each of, for example, three directions (x, y, and z), which may be orthogonal.
  • the accelerometer can be a triaxial accelerometer.
  • the motion sensor(s) additionally can include one or more gyroscopes for sensing rotation data.
  • the wearable monitoring device can include one or more gyroscopes for sensing rotation about each of, for example, three axes, which may be orthogonal.
  • the motion sensor(s) additionally can include one or more altimeters, such as a pressure or barometric altimeter.
  • the devices, methods, and systems of the present disclosure analyze signals collected from the motion sensor(s), PPG sensor(s), and optionally other sensor electrodes, and optionally in conjunction with demographics data associated with the wearer and/or preexisting health data to predict insulin resistance with high confidence. This can be accomplished via a one or more machine-learned models embedded within the wearable computing device or computing system associated with the wearable computing device.
  • the machine-learned models can be trained at least in part on demographic training data, biometric training data, and preexisting health training data from a wide population of users.
  • the biometric training data can include raw photoplethysmogram training data
  • the preexisting health training data can include biomarker training data, continuous glucose monitoring training data, or a combination thereof.
  • at least non-invasive biometric data associated with the wearer of the wearable computing device can be input into the model to predict whether or not the wearer is insulin resistant (or insulin sensitive, impaired insulin resistant, etc.) based on a calculated HOMA-IR score or value.
  • the disclosed devices, systems, and methods allow for prediction of a wearer’s status regarding insulin sensitivity, impaired insulin resistance, or insulin resistance to be accurately determined in a non-invasive manner, which can result in the wearer making healthy lifestyle, diet, and other changes to prevent the onset of prediabetes or type 2 diabetes.
  • FIGS. 1-2 illustrate perspective views of a wearable computing device 102 according to the present disclosure.
  • the wearable computing device 102 may be worn on a user or wearer’s forearm 101 like a wristwatch.
  • the wearable computing device 102 may include a wristband 103 for securing the wearable computing device 102 to the user or wearer’s forearm 101.
  • the wearable computing device 102 may be worn at any other suitable location by a user, such as, for example, on an ankle.
  • the wearable computing device 102 can include a ring, band, earring, necklace, or any other wearable device known by one of skill in the art.
  • the wearable computing device 102 has an outer covering 105 and a housing 104 that contains the electronics associated with the wearable computing device 102.
  • the outer covering 105 may be constructed of glass, polycarbonate, acrylic, or similar.
  • the wearable computing device 102 includes an electronic display screen 106 arranged within the housing 104 and viewable through the outer covering 105.
  • the wearable computing device 102 may also include one or more buttons 108 that may be implemented to provide a mechanism to activate various sensors of the wearable computing device 102 to collect certain health data of the user.
  • the electronic display 106 may cover an electronics package (not shown), which may also be housed within the housing 104.
  • one or more motion sensors 124 as described in detail above may be contained within the housing 104 of the wearable computing device 102, which can be used, inter alia, to calculate step count, etc.
  • the housing 104 of the wearable computing device 102 further includes a dorsal wrist-side face 110 configured to sit against a dorsal wrist of a user when being worn by the user, and one or more PPG sensors 126 as described above can be disposed on the dorsal wrist-side face 110 of the housing 104 of the wearable computing device 102 so as to maintain skin contact with the user when being worn on the wrist by the user.
  • a plurality of sensor electrodes 125 can be positioned on the dorsal wrist-side face 110 of the housing 104 so as to maintain skin contact with the user when being worn on the wrist by the user.
  • each of the sensor electrodes 125 may be configurable to measure, at least, electrical impedance of the user at a location of the skin contact on the dorsal wrist, which can be associated with electrodermal activity data.
  • the wearable computing device 102 may also include at least one additional biometric sensor electrode in addition to the PPG sensors 126 and the impedance sensor electrodes.
  • the additional biometric sensor electrode 125 may include one or more temperature sensors (such as an ambient temperature sensor or a skin temperature sensor), a humidity sensor, a light sensor, a pressure sensor, a microphone, or an optical sensor. In some instances, the sensor electrodes 125 may generally be arranged around the PPG sensors 126. [52] Referring now to FIG. 3, components of an example computing system 100 of the wearable computing device 102 that can be utilized in accordance with various embodiments are illustrated. In particular, as shown, the computing system 100 may also include at least one processor 112 communicatively coupled to the motion sensor(s) 124, the PPG sensor(s) 126, and any other sensors present, such as the sensor electrodes 125 (e.g., electrodermal activity electrodes). Moreover, in an embodiment, the processor(s) 112 may be a central processing unit (CPU) or graphics processing unit (GPU) for executing instructions that can be stored in a memory 114, such as flash memory or DRAM, among other such options.
  • CPU central processing unit
  • the memory 114 may include RAM, ROM, FLASH memory, or other non-transitory digital data storage, and may include a control program comprising sequences of instructions 118 which, when loaded from the memory 114 and executed using the processor(s) 112, cause the processor(s) 112 to perform the functions that are described herein.
  • the computing system 100 can include many types of memory, data storage, or computer-readable media, such as data storage for program instructions for execution by any suitable processor. The same or separate storage can be used for images or data, a removable memory can be available for sharing information with other devices, and any number of communication approaches can be available for sharing with other devices.
  • the computing system 100 includes any suitable display 106, such as a touch screen, organic light emitting diode (OLED), or liquid crystal display (LCD), although devices might convey information via other means, such as through audio speakers, projectors, or casting the display or streaming data to another device, such as a mobile phone, wherein an application on the mobile phone displays the data.
  • a touch screen such as a touch screen, organic light emitting diode (OLED), or liquid crystal display (LCD)
  • OLED organic light emitting diode
  • LCD liquid crystal display
  • the computing system 100 may also include one or more wireless components 212 operable to communicate with one or more electronic devices within a communication range of the particular wireless channel.
  • the wireless channel can be any appropriate channel used to enable devices to communicate wirelessly, such as Bluetooth, cellular, NFC, Ultra- Wideband (UWB), or Wi-Fi channels. It should be understood that the computing system 100 can have one or more conventional wired communications connections as known in the art.
  • the computing system 100 also includes one or more power components 208, such as a battery operable to be recharged through conventional plug-in approaches, or through other approaches such as capacitive charging through proximity with a power mat or other such device.
  • the computing system 100 can also include at least one additional input-output (I/O) component 122 able to receive conventional input from a user.
  • This conventional input can include, for example, a push button, touch pad, touch screen, wheel joystick, keyboard, mouse, keypad, or any other such device or element whereby a user can input a command to the computing system 100.
  • the I/O component(s) 122 may be connected by a wireless infrared or Bluetooth or other link as well in some embodiments.
  • the computing system 100 may also include a microphone or other audio capture element that accepts voice or other audio commands.
  • the computing system 100 may not include any buttons at all, but might be controlled only through a combination of visual and audio commands, such that a user can control the wearable computing device 102 without having to be in contact therewith.
  • the I/O components 122 may also include one or more of the sensor electrodes 125 described herein, optical sensors, barometric sensors (e.g., altimeter, etc.), and the like.
  • the computing system 100 can include a motion sensor 124, as well as a driver 214 and a photoplethy smogram (PPG) sensor 126 that includes at least some combination of one or more emitters 127 and one or more detectors 128 for measuring data for one or more metrics of a human body via optical signals, such as for a person wearing the wearable computing device 102.
  • the PPG sensor 126 may be arranged within the housing 104 and at least partially exposed through the dorsal wrist-side face 110 of the housing 104.
  • the sensor electrodes 125 may be positioned around the PPG sensor 126 on the wristside face 110 of the housing 104.
  • the various components of the PPG sensor 126 may be positioned around the sensor electrodes 125 and/or in another other suitable configuration such as adjacent to, interspersed with, surrounded by, or on top of the PPG sensor 126.
  • the emitters 127 and detectors 128 of the PPG sensor 126 as shown in FIG. 3 are capable of being used, in one example, for obtaining optical PPG measurements as part of the PPG sensor 126.
  • Some PPG technologies rely on detecting light at a single spatial location, adding signals taken from two or more spatial locations, or an algorithmic combination thereof. Both of these approaches result in a single spatial measurement from which the heart rate (HR) estimate (or other physiological metrics) can be determined.
  • HR heart rate
  • a PPG device employs a single light source coupled to a single detector (i.e., a single light path).
  • a PPG device may employ multiple light sources coupled to a single detector or multiple detectors (i.e., two or more light paths).
  • a PPG device employs multiple detectors coupled to a single light source or multiple light sources (i.e., two or more light paths).
  • the light source(s) may be configured to emit one or more of green, red, infrared (IR) light, as well as any other suitable wavelengths in the spectrum.
  • a PPG device may employ a single light source and two or more light detectors each configured to detect a specific wavelength or wavelength range.
  • each detector is configured to detect a different wavelength or wavelength range from one another.
  • two or more detectors are configured to detect the same wavelength or wavelength range.
  • one or more detectors configured to detect a specific wavelength or wavelength range different from one or more other detectors).
  • the PPG device may determine an average of the signals resulting from the multiple light paths before determining an HR estimate or other physiological metrics.
  • the PPG sensor 126 can be used to collect non-invasive biometric data related to resting heart rate, heart rate variability, interbeat interval, etc.
  • the emitters 127 and detectors 128 may be coupled to the processor 112 directly or indirectly using driver circuitry by which the processor 112 may drive the emitters 127 and obtain signals from the detectors 128.
  • a server computing system 130 can communicate with the wireless networking components 212 via the one or more networks 180, which may include one or more local area networks, wide area networks, UWB, and/or internetworks using any of terrestrial or satellite links.
  • the server computing system 130 executes control programs and/or application programs that are configured to perform some of the functions described herein.
  • FIG. 4 a schematic diagram of an environment 300 in which aspects of various embodiments can be implemented is illustrated.
  • a user might have a number of different devices that are able to communicate using at least one wireless communication protocol.
  • the user might have a smartwatch 302 or fitness tracker (such as wearable computing device 102), which the user would like to be able to communicate with a smartphone 304 and a tablet computer 306.
  • the ability to communicate with multiple devices can enable a user to obtain information from the smartwatch 302, e.g., data captured using a sensor on the smartwatch 302, using an application installed on either the smartphone 304 or the tablet computer 306.
  • the user may also want the smartwatch 302 to be able to communicate with a service provider 308, or other such entity, which is able to obtain and process data from the smartwatch and provide functionality that may not otherwise be available on the smartwatch or the applications installed on the individual devices.
  • the smartwatch 302 may be able to communicate with the service provider 308 through at least one network 310, such as the Internet or a cellular network, or may communicate over a wireless connection such as Bluetooth® to one of the individual devices, which can then communicate over the at least one network.
  • a network 310 such as the Internet or a cellular network
  • Bluetooth® wireless connection
  • a user or wearer may want to allow the system 100 to access health data 172 from an external source 170 not associated with the wearable computing device 102, such as an external health records system (see FIG. 5).
  • the health data 172 can include preexisting health data of the user or wearer in the form of electronic health records that can include biomarker data.
  • biomarker data could have been previously collected in an invasive or non-invasive manner and can include biomarkers from blood testing, and this data can be related to a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alkaline phosphatase (ALP) level, an aspartate aminotransferase (AST) level,
  • a user or wearer may also want the devices to be able to communicate in a number of ways or with certain aspects. For example, the user or wearer may want communications between the devices to be secure, particularly where the data may include personal health data or other such communications. The device or application providers may also be required to secure this information in at least some situations. The user may want the devices to be able to communicate with each other concurrently, rather than sequentially. This may be particularly true where pairing may be required, as the user may prefer that each device be paired at most once, such that no manual pairing is required.
  • the user may also desire the communications to be as standards-based as possible, not only so that little manual intervention is required on the part of the user but also so that the devices can communicate with as many other types of devices as possible, which is often not the case for various proprietary formats.
  • a user may thus desire to be able to walk in a room with one device and have such device automatically communicate with another target device with little to no effort on the part of the user.
  • a device will utilize a communication technology such as Wi-Fi to communicate with other devices using wireless local area networking (WLAN).
  • WLAN wireless local area networking
  • Smaller or lower capacity devices, such as many Internet of Things (loT) devices instead utilize a communication technology such as Bluetooth®, and in particular Bluetooth Low Energy (BLE) which has very low power consumption.
  • the environment 300 illustrated in FIG. 4 enables data to be captured, processed, and displayed in a number of different ways.
  • data may be captured using sensors on the smartwatch 302, but due to limited resources on the smartwatch 302, the data may be transferred to the smartphone 304 or the service provider 308 (or a cloud resource) for processing, and results of that processing may then be presented back to that user on the smartwatch 302, smartphone 304, and/or another such device associated with that user, such as the tablet computer 306.
  • a user may also be able to provide input such as health data 172 from an external data source 170 (see FIG. 5) using an interface on any of these devices, which can then be considered when making that determination.
  • the data collected from the motion sensor(s) 124, the sensor electrodes 125, and/or the PPG sensor(s) 126 can be utilized in one or more digital signal processing and/or machine-learned algorithms in order to predict insulin resistance in a wearer of the wearable computing device 102.
  • FIG. 5 depicts a block diagram of an example computing system 500 that can predict insulin sensitivity (HOMA-IR score or value of 1.1 or less), impaired insulin resistance (HOMA-IR score or value between 1.1 and 2.9), and/or insulin resistance (HOMA- IR score or value of 2.9 or greater) in a wearer or user according to example embodiments of the present disclosure.
  • the system 500 includes the wearable computing device 102 described above, a server computing system 130, and a training computing system 150 that are communicatively coupled over a network 180.
  • the system 500 can also include an external data source 170, which can house health data 172, such as electronic health records obtained from invasive and/or non-invasive testing associated with the wearer or user.
  • health data 172 such as electronic health records obtained from invasive and/or non-invasive testing associated with the wearer or user.
  • the wearable computing device 102 includes one or more processors 112 and a memory 114.
  • the one or more processors 112 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected.
  • the memory 114 can include one or more non -transitory computer-readable storage media, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof.
  • the memory 114 can store data 116 and instructions 118 which are executed by the processor 112 to cause the wearable computing device 102 to perform operations.
  • the wearable computing device 102 can store or include one or more machine-learned models 120 for predicting insulin resistance in a wearer.
  • the one or more machine-learned models 120 can be or can otherwise include various machine- learned models such as neural networks (e.g., deep neural networks), large language models (LLMs), or other types of machine-learned models, including non-linear models and/or linear models.
  • Neural networks can include feed-forward neural networks, recurrent neural networks (e.g., long short-term memory recurrent neural networks), convolutional neural networks or other forms of neural networks.
  • Some example machine-learned models can leverage an attention mechanism such as self-attention.
  • some example machine-learned models can include multi-headed self-attention models (e.g., transformer models). Further, the one or more machine-learned models 120 can be trained as a regression model and can be evaluated using a leave-one-out cross-validation (LOOCV) process.
  • LOOCV leave-one-out cross-validation
  • the one or more models 120 can be received from the server computing system 130 over network 180, stored in the wearable computing device memory 114, and then used or otherwise implemented by the one or more processors 112.
  • the wearable computing device 102 can implement multiple parallel instances of a machine-learned model 120 to predict insulin resistance.
  • one or more machine-learned models 140 to predict insulin resistance can be included in or otherwise stored and implemented by the server computing system 130 that communicates with the wearable computing device 102 according to a client-server relationship.
  • the machine-learned models 140 can be implemented by the server computing system 140 as a portion of a web service.
  • one or more models 120 can be stored and implemented at the wearable computing device 102 and/or one or more models 140 can be stored and implemented at the server computing system 130.
  • the wearable computing device 102 can also include one or more user input components 122 that receives user input.
  • the user input component 122 can be a touch-sensitive component (e.g., a touch-sensitive display screen or a touch pad) that is sensitive to the touch of a user input object (e.g., a finger or a stylus).
  • the touch-sensitive component can serve to implement a virtual keyboard.
  • Other example user input components include a microphone, a traditional keyboard, or other means by which a user can provide user input.
  • the server computing system 130 includes one or more processors 132 and a memory 134.
  • the one or more processors 132 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected.
  • the memory 134 can include one or more non-transitory computer-readable storage media, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof.
  • the memory 134 can store data 136 and instructions 138 which are executed by the processor 132 to cause the server computing system 130 to perform operations.
  • the server computing system 130 includes or is otherwise implemented by one or more server computing devices. In instances in which the server computing system 130 includes plural server computing devices, such server computing devices can operate according to sequential computing architectures, parallel computing architectures, or some combination thereof.
  • the server computing system 130 can store or otherwise include one or more machine-learned models 140 for predicting insulin resistance.
  • the models 140 can be or can otherwise include various machine-learned models.
  • Example machine-learned models include neural networks or other multi-layer non-linear models.
  • Example neural networks include feed forward neural networks, deep neural networks, recurrent neural networks, and convolutional neural networks.
  • Some example machine- learned models can leverage an attention mechanism such as self-attention.
  • some example machine-learned models can include multi-headed self-attention models (e.g., transformer models).
  • the one or more machine-learned models 140 can be evaluated using a leave-one-out cross-validation (LOOCV) process.
  • LOOCV leave-one-out cross-validation
  • the wearable computing device 102 and/or the server computing system 130 can train the models 120 and/or 140 via interaction with the training computing system 150 that is communicatively coupled over the network 180.
  • the training computing system 150 can be separate from the server computing system 130 or can be a portion of the server computing system 130.
  • the training computing system 150 can include one or more processors 152 and a memory 154.
  • the one or more processors 152 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected.
  • the memory 154 can include one or more non-transitory computer-readable storage media, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof.
  • the memory 154 can store data 156 and instructions 158 which are executed by the processor 152 to cause the training computing system 150 to perform operations.
  • the training computing system 150 includes or is otherwise implemented by one or more server computing devices.
  • the training computing system 150 can include a model trainer 160 that trains the machine-learned models 120 and/or 140 stored at the wearable computing device 102 and/or the server computing system 130 using various training or learning techniques, such as, for example, backwards propagation of errors.
  • a loss function can be backpropagated through the model(s) to update one or more parameters of the model(s) (e.g., based on a gradient of the loss function).
  • Various loss functions can be used such as mean squared error, likelihood loss, cross entropy loss, hinge loss, and/or various other loss functions.
  • Gradient descent techniques can be used to iteratively update the parameters over a number of training iterations.
  • performing backwards propagation of errors can include performing truncated backpropagation through time.
  • the model trainer 160 can perform a number of generalization techniques (e.g., weight decays, dropouts, etc.) to improve the generalization capability of the models being trained.
  • the model trainer 160 can train the machine-learned models 120 and/or 140 based on a set of training data 162.
  • the training data 162 can include, for example, demographic training data, biometric training data, biomarker training data (data associated with a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alka
  • the training examples can be provided by the wearable computing device 102.
  • the model 120 provided to the wearable computing device 102 can be trained by the training computing system 150 on, in part, user-specific data received from the wearable computing device 102. In some instances, this process can be referred to as personalizing the model.
  • the model trainer 160 includes computer logic utilized to provide desired functionality.
  • the model trainer 160 can be implemented in hardware, firmware, and/or software controlling a general purpose processor.
  • the model trainer 160 includes program files stored on a storage device, loaded into a memory, and executed by one or more processors.
  • the model trainer 160 includes one or more sets of computer-executable instructions that are stored in a tangible computer-readable storage medium such as RAM, hard disk, or optical or magnetic media.
  • the network 180 can be any type of communications network, such as a local area network (e.g., intranet), wide area network (e.g., Internet), or some combination thereof and can include any number of wired or wireless links.
  • communication over the network 180 can be carried via any type of wired and/or wireless connection, using a wide variety of communication protocols (e.g., TCP/IP, HTTP, SMTP, FTP), encodings or formats (e.g., HTML, XML), and/or protection schemes (e.g., VPN, secure HTTP, SSL).
  • the machine-learned models described in this specification may be used in a variety of tasks, applications, and/or use cases.
  • the input to the machine- learned model(s) of the present disclosure can be demographic data, non-invasive biometric data from sensors on the wearable device, health data (e.g., biomarker data) from an external data source, or a combination thereof.
  • the machine-learned model(s) can then process the aforementioned data to generate an output.
  • the machine-learned model(s) can process the data to generate a recognition output.
  • the machine-learned model(s) can process the data to generate a prediction output.
  • the machine-learned model(s) can process the data to generate a classification output.
  • the machine-learned model(s) can process the data to generate a segmentation output.
  • the machine-learned model(s) can process the data to generate a visualization output.
  • the machine-learned model(s) can process the data to generate a diagnostic output.
  • the machine-learned model(s) can process the data to generate a detection output.
  • the input to the machine-learned model(s) of the present disclosure can be statistical data.
  • Statistical data can be, represent, or otherwise include data computed and/or calculated from some other data source.
  • the machine-learned model(s) can process the statistical data to generate an output.
  • the machine-learned model(s) can process the statistical data to generate a recognition output.
  • the machine-learned model(s) can process the statistical data to generate a prediction output.
  • the machine-learned model(s) can process the statistical data to generate a classification output.
  • the machine-learned model(s) can process the statistical data to generate a segmentation output.
  • the machine-learned model(s) can process the statistical data to generate a visualization output.
  • the machine-learned model(s) can process the statistical data to generate a diagnostic output.
  • FIG. 5 illustrates one example computing system 500 that can be used to implement the machine learning aspects of the present disclosure.
  • the wearable computing device 102 can include the model trainer 160 and the training dataset 162.
  • the models 120 can be both trained and used locally at the wearable computing device 102.
  • the wearable computing device 102 can implement the model trainer 160 to personalize the models 120 based on user-specific data.
  • FIG. 6 depicts a block diagram of an example computing device 600 that performs according to example embodiments of the present disclosure.
  • the computing device 600 can be a wearable computing device or a server computing device.
  • the computing device 600 includes a number of applications (e.g., applications 1 through N). Each application contains its own machine learning library and machine-learned model(s). For example, each application can include a machine-learned model.
  • Example applications include a biometric sensor application, a demographic application, an electronic health record application, etc.
  • each application of the computing device 600 can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, and/or additional components.
  • each application can communicate with each device component using an API (e.g., a public API).
  • the API used by each application is specific to that application.
  • FIG. 7 depicts a block diagram of an example computing device 700 that performs according to example embodiments of the present disclosure.
  • the computing device 700 can be a wearable computing device or a server computing device.
  • the computing device 700 includes a number of applications (e.g., applications 1 through N). Each application is in communication with a central intelligence layer. Example applications include a biometric sensor application, a demographic application, an electronic health record application, etc. In some implementations, each application can communicate with the central intelligence layer (and model(s) stored therein) using an API (e.g., a common API across all applications).
  • applications e.g., applications 1 through N.
  • Each application is in communication with a central intelligence layer.
  • Example applications include a biometric sensor application, a demographic application, an electronic health record application, etc.
  • each application can communicate with the central intelligence layer (and model(s) stored therein) using an API (e.g., a common API across all applications).
  • the central intelligence layer includes a number of machine-learned models. For example, as illustrated in FIG. 7, a respective machine-learned model can be provided for each application and managed by the central intelligence layer. In other implementations, two or more applications can share a single machine-learned model. For example, in some implementations, the central intelligence layer can provide a single model for all of the applications. In some implementations, the central intelligence layer is included within or otherwise implemented by an operating system of the computing device 700.
  • the central intelligence layer can communicate with a central device data layer.
  • the central device data layer can be a centralized repository of data for the computing device 700. As illustrated in FIG. 7, the central device data layer can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, and/or additional components. In some implementations, the central device data layer can communicate with each device component using an API (e.g., a private API).
  • an API e.g., a private API
  • data from an external data source 170 can be input into the model or models in addition to demographic data and non-invasive biometric data received from the wearable computing device 102.
  • health data 172 can be preexisting and can be obtained from electronic health record data associated with the wearer.
  • the electronic health record data can include non-invasive test, invasive test data, or a combination thereof.
  • a user or wearer may want to allow the system 100 to access the health data 172 from the external source 170 not associated with the wearable computing device 102, such as an external electronic health records system.
  • the health data 172 can thus include any preexisting health data of the user or wearer in the form of electronic health records, which can include demographic data, biometric data, biomarker data, etc.
  • the biomarker data can include data associated with a complete blood count, a comprehensive metabolic panel, an insulin level, a blood glucose level, a total cholesterol level, an HDL cholesterol level, an LDL cholesterol level, a triglyceride level, an HbAlc level, a high-sensitivity C-reactive protein level, a gamma-glutamyl transferase level, a testosterone level, a blood urea nitrogen level, a creatinine level, an Estimated Glomerular Filtration Rate (eGFR), a sodium level, a potassium level, a chloride level, a carbon dioxide level, a calcium level, a total protein level, an albumin level, a globulin level, an albumin/globulin ratio, a total bilirubin level, an alkaline phosphatas
  • the workflow 800 includes a step 802 of compiling training data, which is then subjected to a step 804 of dimensionality reduction 804, followed by a step 806 of data splitting, followed by a step 808 of applying a regression model, after which, in step 810, a prediction is made as to whether a person exhibits insulin sensitivity, impaired insulin resistance, or insulin resistance. Then, in step 812, statistical analysis can be performed to compare the model predictions to actual HOMA-IR targets, scores, or values 810 for insulin sensitivity, impaired insulin resistance, or insulin resistance to determine the accuracy of the predicted HOMA-IR scores or values.
  • the method 1000 can include a step 1002 of training a machine-learned model at least in part on demographic training data, biometric training data, continuous glucose monitoring training data, raw photoplethysmogram training data, or a combination thereof.
  • the method 1000 can include a step 1004 of receiving, via one or more processors, non- invasive biometric data from one or more sensors associated with the wearable computing device.
  • the method 1000 can include a step 1006 of implementing, via the one or more processors, a machine-learned model to determine if the non-invasive biometric data associated with the wearer is indicative of insulin resistance, wherein the machine-learned model is configured to predict insulin resistance in the wearer based at least in part on the non-invasive biometric data associated with the wearer.
  • the method 1000 can include a step 1008 of calculating, via the one or more processors, a predicted Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) score for the wearer based at least in part on the biometric data associated with the wearer.
  • HOMA-IR Homeostatic Model Assessment for Insulin Resistance
  • the method 100 can include a step 1010 of displaying, via the one or more processors, a notification on a display of the wearable computing device alerting the wearer of the predicted HOMA-IR score.
  • the method 1000 may also include utilizing the machine- learned model (e.g., a large language models (LLMs)) to make recommendations to the user as to how to reverse insulin resistance.
  • the machine- learned model e.g., a large language models (LLMs)
  • FIGS. 9A-9F four baseline experiments were conducted to assess the performance of non-invasively predicting insulin resistance using the machine-learned model, systems, and methods of the present disclosure.
  • the machine-learned model was trained as a regression model and a leave-one-out cross-validation (LOOCV) process was utilized to evaluate the model and assess the model’s performance in predicting insulin sensitivity based on HOMA-IR scores within ranges associated with insulin sensitivity, impaired insulin resistance, or insulin resistance in a non-invasive manner.
  • LOCV leave-one-out cross-validation
  • FIG. 9A is a graph comparing actual HOMA-IR scores to predicted HOMA-IR score according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on non-invasive biometric data and demographic data associated with the wearer;
  • FIG. 9B is a graph comparing actual HOMA- IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on selected non- invasive biometric data and demographic data associated with the wearer;
  • FIG. 9A is a graph comparing actual HOMA-IR scores to predicted HOMA-IR score according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on selected non- invasive biometric data and demographic data associated with the wearer;
  • FIG. 9C is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on non-invasive biometric data, demographic data, and fasting glucose (predetermined) associated with the wearer;
  • FIG. 9D is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on non-invasive biometric data, demographic data, and a lipid and complete metabolic panel (predetermined) associated with the wearer;
  • FIG. 9C is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on non-invasive biometric data, demographic data, and a lipid and complete metabolic panel (predetermined) associated with the wearer;
  • FIG. 9D is a graph comparing actual HOMA-IR scores to predicted
  • FIG. 9E is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on non-invasive biometric data associated with the wearer; and FIG. 9F is a graph comparing actual HOMA-IR scores to predicted HOMA-IR scores according to an example embodiment of the present disclosure, where the predicted HOMA-IR scores were determined based on demographic data associated with the wearer.
  • the annotated dashed lines on the x-axis and y-axis are the cutoff points for insulin sensitivity (HOMA-IR scores or values of 1.1 or less), impaired insulin sensitivity (HOMA-IR scores or values between 1.1 and 2.9), and insulin resistance (HOMA-IR scores or values of 2.9 or greater) for the true HOMA-IR scores or values and the predicted HOMA-IR scores or values, respectively.
  • the demographic data analyzed included body mass index (BMI) and age, while the biometric features analyzed included resting heart rate (RHR), heart rate (HR), daily step count, sleep duration, and heart rate variability (HRV) as summarized in terms of mean, median, standard deviation, maximum value, and minimum value. It should be understood that when listed as selected biometric features in Table 1 below, such data only includes mean stats only for resting heart rate (RHR), heart rate, daily step count, and sleep duration.
  • insulin resistance can be predicted from non-invasive biometric data (e.g., resting heart rate (RHR), daily step count, sleep duration, and heart rate variability (HRV)), non-invasive biometric data (e.g., resting heart rate (RHR), daily step count, sleep duration, and heart rate variability (HRV)) and demographic data (e.g., BMI and age), and/or non-invasive biometric data, demographic data (e.g., BMI and age) and previously collected invasive biometric data (e.g., blood testing data such as fasting glucose, lipid/complete metabolic panel, HbAlc) as summarized below in Table 1 : Table 1: Summary of Various Machine-Learned Model Inputs for Predicting IR
  • a user or wearer may be provided with controls allowing the user or wearer to make an election as to both if and when systems, programs, or features described herein may enable collection of user information (e.g., information about a user’s health data, activities, social network, social actions, profession, a user’s preferences, or a user’s current location, etc.), and if the user is sent content or communications from a server.
  • user information e.g., information about a user’s health data, activities, social network, social actions, profession, a user’s preferences, or a user’s current location, etc.
  • certain data may be treated in one or more ways before it is stored or used, so that personally identifiable information is removed.
  • a user’s identity may be treated so that no personally identifiable information can be determined for the user, or a user’s geographic location may be generalized where location information is obtained (such as to a city, ZIP code, or state level), so that a particular location of a user cannot be determined.
  • location information such as to a city, ZIP code, or state level
  • the wearer or user may have control over what information is collected about the user, how that information is used, and what information is provided to the user.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Physics & Mathematics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Molecular Biology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biophysics (AREA)
  • Veterinary Medicine (AREA)
  • Artificial Intelligence (AREA)
  • Primary Health Care (AREA)
  • Physiology (AREA)
  • Data Mining & Analysis (AREA)
  • Epidemiology (AREA)
  • Psychiatry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Signal Processing (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Computation (AREA)
  • Business, Economics & Management (AREA)
  • General Business, Economics & Management (AREA)
  • Emergency Medicine (AREA)
  • Optics & Photonics (AREA)
  • Dentistry (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

L'invention concerne un système informatique, un procédé mis en oeuvre par ordinateur et un dispositif informatique pouvant être porté destinés à prédire la résistance à l'insuline chez un porteur du dispositif informatique pouvant être porté sans nécessiter de tests invasifs supplémentaires en dehors des données de test qui peuvent déjà être disponibles, bien que cela ne soit pas requis. Par exemple, un modèle d'apprentissage automatique est entraîné à prédire la résistance à l'insuline chez le porteur du dispositif informatique pouvant être porté sur la base, au moins en partie, de données biométriques non invasives associées au porteur. Ensuite, un ou plusieurs supports lisibles par ordinateur non transitoires amènent le système informatique à effectuer des opérations par l'intermédiaire d'un ou de plusieurs processeurs. Les opérations consistent à recevoir les données biométriques non invasives provenant d'un ou de plusieurs capteurs associés au dispositif informatique pouvant être porté ; et à mettre en oeuvre le modèle d'apprentissage automatique afin de déterminer si les données biométriques non invasives associées au porteur indiquent une résistance à l'insuline.
PCT/US2025/027242 2024-05-03 2025-05-01 Prédiction non invasive de résistance à l'insuline à l'aide de données de capteur Pending WO2025231209A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US202463642369P 2024-05-03 2024-05-03
US63/642,369 2024-05-03

Publications (1)

Publication Number Publication Date
WO2025231209A1 true WO2025231209A1 (fr) 2025-11-06

Family

ID=95895585

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2025/027242 Pending WO2025231209A1 (fr) 2024-05-03 2025-05-01 Prédiction non invasive de résistance à l'insuline à l'aide de données de capteur

Country Status (1)

Country Link
WO (1) WO2025231209A1 (fr)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210050089A1 (en) * 2019-08-13 2021-02-18 Twin Health, Inc. Metabolic health using a precision treatment platform enabled by whole body digital twin technology

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210050089A1 (en) * 2019-08-13 2021-02-18 Twin Health, Inc. Metabolic health using a precision treatment platform enabled by whole body digital twin technology

Similar Documents

Publication Publication Date Title
US11872041B1 (en) Determining mental health and cognitive state through physiological and other non-invasively obtained data
JP7127086B2 (ja) ヘルストラッキングデバイス
EP2457500B1 (fr) Capteur à anneau alimenté par induction
JP7113833B2 (ja) 体液中のグルコースレベルを示すグルコースモニタリングデータを分析するための、コンピュータにより実施される方法および携帯型装置、ならびにコンピュータプログラム製品
Penders et al. Wearable sensors for healthier pregnancies
EP2479692A2 (fr) Capteur d'humeur
EP2457505A1 (fr) Diagnostic et surveillance de la dyspnée
EP2458544A1 (fr) Enregistrement et analyse de données sur un avatar 3D
US20120130202A1 (en) Diagnosis and Monitoring of Musculoskeletal Pathologies
US11478186B2 (en) Cluster-based sleep analysis
AU2022201287A1 (en) Cluster-based sleep analysis
CN119731744A (zh) 健身疲劳分数确定和管理技术
WO2025231209A1 (fr) Prédiction non invasive de résistance à l'insuline à l'aide de données de capteur
US20160051189A1 (en) Method and apparatus for processing food information
EP4663119A1 (fr) Surveillance de la santé cardiovasculaire à l'aide de données de capteur
WO2025128655A1 (fr) Système et procédé de détection de disparition du pouls utilisant un dispositif informatique pouvant être porté sur soi
KR20250176121A (ko) 센서 데이터를 사용한 심혈관 건강 모니터링
JP2025185721A (ja) センサデータを使用した心血管の健康の監視
Gunay Modelling of sleep behaviors of patients with mood disorders
Akhtar et al. Implementing wearable sensor technology for the determination of a biomarker profile for cancer-related fatigue
HK1248006B (zh) 健康跟踪设备