Method for performing blood pressure estimation

Abstract

This application discloses blood pressure estimation using photoplethysmographic pulse wave measurements. A device can obtain a heartbeat profile for a current subject, the heartbeat profile based on photoplethysmographic pulse wave (PPG) data associated with a set of wavelength channels. The device can determine a global model comprising a plurality of classes, each class associated with a subject identifier and a blood pressure level (BPL) identifier. The device can identify one class as a closest class for the current subject based on the heartbeat profile for the current subject. The device can identify a local subject associated with the closest class based on the subject identifier associated with the closest class. The device can generate a BPL estimation model for the current subject based on heartbeat profile data for the local subject. The device can determine a BPL estimate for the current subject based on the heartbeat profile for the current subject and using the BPL estimation model.

Description

TECHNICAL FIELD

[0001] The present application relates to blood pressure estimation using photoplethysmographic pulse wave measurements. BACKGROUND

[0002] Photoplethysmography (PPG) is an optical technique that can be used to detect volume changes in blood in the peripheral circulation (blood volume changes due to the pumping action of the heart). PPG is a non-invasive method of measurement at the skin surface (e.g., at the fingertip, wrist, earlobe, etc.). PPG devices can take the form of, for example, a multi-spectral sensor device (e.g., a binary multi-spectral (BMS) sensor device) that provides heartbeat time series data associated with a plurality of wavelength channels (e.g., 64 wavelength channels). The multi-spectral sensor device can include a plurality of sensor elements (e.g., optical, spectral, and / or image sensors) that each receive one of the plurality of wavelength channels (via a respective region of a multi-spectral filter) in order to capture the heartbeat time series data.

[0003] SUMMARY

[0004] According to some embodiments, a method can include obtaining, by a device, a heartbeat profile for a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels; determining, by the device, a global model that includes a plurality of classes, each class of the plurality of classes being associated with a subject identifier and a blood pressure level (BPL) identifier; identifying, by the device, one class of the plurality of classes included in the global model as a closest class for the current subject, the closest class being identified based on the heartbeat profile for the current subject; identifying, by the device, a local subject associated with the closest class, the local subject being identified based on a subject identifier associated with the closest class; selecting, by the device, at least two classes associated with the local subject for use in generating a BPL estimation model for the current subject, wherein the at least two classes include the closest class and at least one other class associated with the local subject, the at least one other class being associated with a BPL identifier that is different from a BPL identifier of the closest class; generating, by the device, the BPL estimation model based on heartbeat profile data for the at least two classes associated with the local subject; and determining, by the device, a BPL estimate for the current subject based on the heartbeat profile for the current subject and using the BPL estimation model.

[0005] According to some embodiments, a method may include: obtaining, by a device, a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels; determining, by the device, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier; identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on the heartbeat profile of the current subject; identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category; and determining, by the device and based on identifying the local subject, a local subject transfer set associated with the local subject. a local subject transfer set including one or more heartbeat profiles of the local subject collected at a reference BPL; obtaining, by the device, a current subject transfer set associated with the current subject, the current subject transfer set including one or more heartbeat profiles of the current subject collected at a reference BPL; creating, by the device, a transferred current subject set based on the current subject transfer set and the local subject transfer set, wherein the transferred current subject set includes a plurality of transferred heartbeat profiles associated with the current subject; generating, by the device, a BPL estimation model based on the local subject set associated with the identified local subject; and determining, by the device, a BPL estimation for the current subject based on the transferred current subject set and using the BPL estimation model.

[0006] According to some embodiments, a method may include: obtaining, by a device, a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels; determining, by the device, a global model comprising a plurality of categories, each of the plurality of categories being associated with a subject identifier and a BPL identifier; identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on the heartbeat profile of the current subject; identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category; generating, by the device, a BPL estimation model for the current subject based on the heartbeat profile data of the local subject; and determining, by the device, a BPL estimation for the current subject based on the heartbeat profile of the current subject and using the BPL estimation model.

[0007] 1. A method comprising:

[0008] obtaining, by the device, a heartbeat profile of a current subject, the heartbeat profile being based on photoplethysmography (PPG) data associated with a set of wavelength channels;

[0009] determining, by the apparatus, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a blood pressure level (BPL) identifier;

[0010] identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on a heartbeat profile of the current subject;

[0011] identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category;

[0012] selecting, by the device, at least two categories associated with the local subject for use in generating a BPL estimation model for the current subject,

[0013] wherein the at least two categories include the closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier different from the BPL identifier of the closest category;

[0014] generating, by the device, the BPL estimation model based on the heartbeat profile data regarding the at least two categories associated with the local subject; and

[0015] A BPL estimate for the current subject is determined, by the apparatus, based on the heartbeat profile of the current subject and using the BPL estimation model.

[0016] 2. The method according to 1, wherein determining the global model comprises:

[0017] obtaining a plurality of heartbeat profiles associated with a plurality of local subjects,

[0018] wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and

[0019] The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects.

[0020] 3. The method of 1, wherein the global model is generated using a nonlinear classifier.

[0021] 4. The method of 1, wherein selecting the at least two classes for generating a BPL estimation model for the current subject comprises:

[0022] identifying a linear region of a set of classes associated with the local subject, the set of classes including the closest class and the at least one other class; and

[0023] selecting the at least two classes based on the linear region of the set of classes associated with the local subject.

[0024] 5. The method of 1, wherein the BPL estimation model is a quantitative model generated using a partial least squares regression method.

[0025] 6. The method of 1, wherein the BPL estimation model is a qualitative model generated using a classifier.

[0026] 7. The method of 1, wherein the BPL estimation comprises an estimated BPL of the current subject.

[0027] 8. The method of 1, further comprising:

[0028] providing information associated with the BPL estimation of the current subject.

[0029] 9. A method comprising:

[0030] obtaining, by a device, a heartbeat profile of a current subject, the heartbeat profile based on photoplethysmographic (PPG) data associated with a set of wavelength channels;

[0031] determining, by the device, a global model comprising a plurality of classes, each class of the plurality of classes associated with a subject identifier and a blood pressure level (BPL) identifier;

[0032] identifying, by the device, one class of the plurality of classes included in the global model as a closest class for the current subject, the closest class identified based on the heartbeat profile of the current subject;

[0033] identifying, by the device, a local subject associated with the closest class, the local subject identified based on the subject identifier associated with the closest class;

[0034] determining, by the device, and based on identifying the local subject, a local subject transfer set associated with the local subject, the local subject transfer set comprising one or more heartbeat profiles of the local subject collected at or near a reference BPL;

[0035] obtaining, by the device, a current subject delivery set associated with the current subject, the current subject delivery set comprising one or more heartbeat profiles of the current subject collected at the reference BPL;

[0036] creating, by the device, a delivered current subject set based on the current subject delivered set and the local subject delivered set,

[0037] wherein the transferred current subject set comprises a plurality of transferred heartbeat profiles associated with the current subject;

[0038] generating, by the apparatus, a BPL estimation model based on a set of local subjects associated with the identified local subjects; and

[0039] A BPL estimate for the current subject is determined, by the device, based on the communicated set of current subjects and using the BPL estimation model.

[0040] 10. The method according to claim 9, wherein determining the global model comprises:

[0041] obtaining a plurality of heartbeat profiles associated with the plurality of local subjects,

[0042] wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and

[0043] The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects.

[0044] 11. The method of claim 9, wherein the global model is generated using a nonlinear classifier.

[0045] 12. The method of claim 9, wherein the transferred current subject set is created using one of:

[0046] Mean difference correction,

[0047] Segment-wise direct standardization,

[0048] Generalized least squares method,

[0049] Quadrature signal correction method, or

[0050] One or more additional model transfer methods.

[0051] 13. The method according to 9, wherein the BPL estimation model is a quantitative model generated using a partial least squares regression method.

[0052] 14. The method of 9, wherein the BPL estimate comprises an estimated BPL of the current subject.

[0053] 15. The method according to 9, wherein the BPL estimation model is a qualitative model generated using a classifier.

[0054] 16. The method according to item 9, further comprising:

[0055] Information associated with the BPL estimate for the current subject is provided.

[0056] 17. A method comprising:

[0057] obtaining, by the device, a heartbeat profile of a current subject, the heartbeat profile being based on photoplethysmography (PPG) data associated with a set of wavelength channels;

[0058] determining, by the apparatus, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a blood pressure level (BPL) identifier;

[0059] identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on a heartbeat profile of the current subject;

[0060] identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category;

[0061] generating, by the device, a BPL estimation model for the current subject based on the heartbeat profile data of the local subject; and

[0062] A BPL estimate for the current subject is determined, by the apparatus, based on the heartbeat profile of the current subject and using the BPL estimation model.

[0063] 18. The method according to 17, wherein generating the BPL estimation model comprises:

[0064] selecting at least two categories associated with the local subject for use in generating the BPL estimation model for the current subject,

[0065] wherein the at least two categories include the closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier different from the BPL identifier of the closest category; and

[0066] The BPL estimation model is generated based on heartbeat profile data about the at least two categories associated with the local subject.

[0067] 19. The method according to 17, wherein generating the BPL estimation model comprises:

[0068] determining a local subject transfer set associated with the local subject, the local subject transfer set comprising one or more heartbeat profiles of the local subject collected at a reference BPL;

[0069] obtaining a current subject delivery set associated with the current subject, the current subject delivery set comprising one or more heartbeat profiles of the current subject collected at the reference BPL;

[0070] creating a delivered current subject set based on the current subject delivery set and the local subject delivery set,

[0071] wherein the transferred current subject set comprises a plurality of transferred heartbeat profiles associated with the current subject; and

[0072] generating the BPL estimation model based on a set of local subjects associated with the identified local subjects,

[0073] wherein the BPL estimate for the current subject is determined based on the transferred current subject set and using the BPL estimation model.

[0074] 20. The method according to 17, wherein determining the global model comprises:

[0075] obtaining a plurality of heartbeat profiles associated with a plurality of local subjects,

[0076] wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and

[0077] The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects, the global model being generated using a nonlinear classifier. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1A and Figure 1B is an illustration of an example implementation described herein.

[0079] Figure 2 is an illustration of an example associated with a global model associated with multiple categories as described herein.

[0080] Figure 3 and Figure 4 are illustrations showing examples of identifying the closest class associated with a heartbeat profile data based on heartbeat profile data associated with a current subject as described herein.

[0081] Figure 5 is an illustration of an example environment in which the systems and / or methods described herein can be implemented.

[0082] Figure 6 is an illustration of example components of one or more devices of Figure 5

[0083] Figure 7-9 is a flowchart of an example process for determining blood pressure level estimates based on PPG data as described herein. DETAILED DESCRIPTION

[0084] The detailed description of the following example embodiments refers to the accompanying drawings. The same reference numbers in different drawings can identify the same or similar elements or features. Also, while the following description in one example can use a multi-spectral sensor device, the principles, processes, operations, techniques, and methods described herein can be used with any other type of sensing device, such as a spectrometer, optical sensor, spectral sensor, etc.

[0085] As described above, a multi-spectral sensor device can be capable of measuring, obtaining, collecting, or otherwise determining heartbeat time series data associated with a plurality (e.g., 16, 32, 64, etc.) of wavelength channels. Such data is referred to herein as PPG data. In some cases, a multi-spectral sensor device can be used to collect PPG data associated with a subject, and the PPG data can be used to generate a heartbeat profile based on which a blood pressure level (BPL) of the subject can be estimated. However, variations between subjects based on PPG-based heartbeat profiles are typical, and even for the same subject, a heartbeat profile can show different trends of variations at different regions of BPL. Therefore, conventional techniques for estimating BPL based on a heartbeat profile generated from PPG data typically show large bias and poor prediction performance, and thus are unreliable.

[0086] ​Some implementations described herein provide a device that provides improved PPG-based BPL estimation. In some implementations, the device generates a BPL estimation model for a current subject (i.e., the subject for which a BPL estimation is being determined) based on heartbeat profile data about the local subject (i.e., the particular subject identified from a global model associated with a plurality of subjects) and determines a BPL estimation for the current subject based on the current subject’s heartbeat profile and using the BPL estimation model. In some implementations, the manner in which the BPL estimation model is generated can depend on the availability of heartbeat profile data about the current subject and / or heartbeat profile data about global subjects associated with the global model. A variety of techniques for generating a BPL estimation model are provided below.

[0087] Figure 1A and Figure 1B is an illustration of an example implementation 100 described herein.

[0088] As shown in Figure 1A , the multispectral sensor device can be positioned relative to a skin surface of a current subject (i.e., the subject for which a BPL estimation is being determined). For example, as shown in Figure 1A , the multispectral sensor device can be a device worn on the wrist of the current subject. In some implementations, the multispectral sensor device can be positioned at another location on the body (such as a fingertip, an arm, a leg, an earlobe, etc.) relative to the skin surface. In some implementations, the multispectral sensor device includes a BMS sensing device that operates in, for example, the visible (VIS) spectrum, the near-infrared (NIR) spectrum, etc.

[0089] As shown by reference numeral 105, the multispectral sensor device can determine (e.g., measure, collect, acquire, etc.) PPG data (e.g., raw heartbeat data) associated with N (N > 1) wavelength channels. For each of the N wavelength channels, the PPG data includes photometric response data indicative of blood volume under the skin surface at the location of the multispectral sensor device at a given point in time.

[0090] As shown by reference numeral 110, the estimation device can obtain the PPG data from the multispectral sensor device. As described herein, the estimation device is a device capable of generating a BPL estimation model associated with the current subject and / or determining a BPL estimation for the current subject using the BPL estimation model. In some implementations, the estimation device can be integrated with (e.g., in the same package, the same housing, the same chip, etc.) the multispectral sensor device. Alternatively, the estimation device can be separate from (e.g., located remotely relative to) the multispectral sensor device.

[0091] In some embodiments, the estimation device can obtain the PPG data in real time or near real time (e.g., when the multispectral sensor device is configured to provide the PPG data when the multispectral sensor device obtains the PPG data). Additionally or alternatively, the estimation device can obtain the PPG data based on the multispectral sensor device (e.g., automatically) periodically providing the PPG data (e.g., every second, every five seconds, etc.). Additionally or alternatively, the estimation device can obtain the PPG data from the multispectral sensor device based on requesting the PPG data from the multispectral sensor device.

[0092] As shown in mark 115, the estimation device can determine a global model including multiple categories. In some embodiments, one or more categories of the global model can be used as the basis for generating a BPL estimation model, as described below. In some embodiments, each category of the global model includes heartbeat profile data associated with a subject identifier and a BPL identifier. That is, each category can be associated with a subject identifier corresponding to a local subject in a group of local subjects (e.g., a subject for whom a set of heartbeat profiles have been collected at corresponding known BPLs) and a BPL identifier indicating the BPL when the associated heartbeat profile data is collected. In some embodiments, as described below, the estimation device identifies the closest category for the current subject based on the global model, and generates a BPL estimation model for the current subject based on the heartbeat profile data of the local subject associated with the identified closest category.

[0093] In some embodiments, the estimation device can generate a global model. For example, the estimation device can obtain the heartbeat profile associated with a group of local subjects, wherein each heartbeat profile is the heartbeat profile of a local subject in the local subjects and is associated with the corresponding known BPL. As a specific example, the estimation device can obtain a first group of heartbeat profiles associated with a first local subject and a second group of heartbeat profiles associated with a second local subject. Here, the first heartbeat profile in the first group of heartbeat profiles can be associated with the first known BPL of the first local subject (for example, having been collected at the first known BPL), the second heartbeat profile in the first group of heartbeat profiles can be associated with the second known BPL of the first local subject, and so on. Therefore, the first group of heartbeat profiles can include a group of heartbeat profiles of the first local subject, and each heartbeat profile is associated with the corresponding known BPL of the first local subject. Similarly, the first heartbeat profile in the second group of heartbeat profiles can be associated with the first known BPL of the second local subject, and the second heartbeat profile in the second group of heartbeat profiles can be associated with the second known BPL of the second local subject, and so on. Thus, the second set of heartbeat profiles may include a set of heartbeat profiles for a second local subject, each heartbeat profile being associated with a corresponding known BPL for the second local subject. Each set of heartbeat profiles associated with other local subjects may include similar information. In some embodiments, the estimation device obtains the heartbeat profiles associated with a set of local subjects from another device (e.g., a device configured with a database that stores heartbeat profiles for subjects collected at known BPLs).

[0094] In some embodiments, a set of heartbeat profiles for a given local subject can include one or more heartbeat profiles collected at a reference BPL (e.g., measured by a reference blood pressure monitor (such as a clinically approved blood pressure monitor, a home blood pressure monitor, etc.)). In some embodiments, a set of heartbeat profiles for a given local subject can include at least one additional heartbeat profile collected at a BPL that differs from the reference BPL by at least a specific amount (e.g., 20 millimeters of mercury (mmHg)). In some embodiments, a set of heartbeat profiles for a given local subject includes heartbeat profiles associated with at least two BPLs: one or more heartbeat profiles associated with the reference BPL and one or more heartbeat profiles associated with a BPL that differs from the reference BPL (e.g., by a specific amount).

[0095] In some embodiments, the estimation device generates a global model based on the heartbeat profile associated with a group of local subjects. For example, the estimation device can generate a global model including multiple categories. Here, the first category can correspond to the heartbeat profile associated with the first local subject and the first known BPL, the second category can correspond to the heartbeat profile associated with the first local subject and the second known BPL, the third category can correspond to the heartbeat profile associated with the first local subject and the third known BPL, the fourth category can correspond to the heartbeat profile associated with the second local subject and the fourth known BPL, the fifth category can correspond to the heartbeat profile associated with the second local subject and the fifth known BPL, the sixth category can correspond to the heartbeat profile associated with the third local subject and the sixth known BPL, the seventh category can correspond to the heartbeat profile associated with the third local subject and the seventh known BPL, and so on for each heartbeat profile in the group of heartbeat profiles. Generally speaking, in the global model, each heartbeat profile of each local subject at each BPL is regarded as a category (for example, so that multi-label full combination classification is performed to generate a global model). In some embodiments, the estimation device may generate the global model using a nonlinear classifier, such as a support vector machine (SVM) classifier or a hierarchical SVM classifier (ILM).

[0096] In some embodiments, the estimation device may store the global model (e.g., so that the global model can be used to generate a BPL estimation model at a later time). Additionally or alternatively, (e.g., when the estimation device is not configured to generate a global model) the estimation device may determine the global model based on receiving the global model from another device.

[0097] Figure 2 is a diagram showing an example of a global model, as described above, where each BPL for each local subject is considered as a different category in the global model. Figure 2 In the example shown, there are 13 categories associated with five local subjects (e.g. Figure 2 Here, Figure 2 Each region indicated in is associated with one of 13 categories, where each category corresponds to a different combination of local subject identifier and BPL. It is worth noting that the identification of points within each region is crucial for understanding Figure 2 The concepts shown are not required; therefore, Figure 2 There is no clear delineation of each point in the diagram. Figure 2 are provided as examples only. Other examples may be Figure 2 The examples described are different.

[0098] Back to Figure 1AAs shown in mark 120, the estimation device may identify one of the multiple categories included in the global model as the closest category for the current subject. In some embodiments, the estimation device may identify the closest category based on the heartbeat profile of the current subject. For example, the estimation device may classify the heartbeat profile of the current subject into one of the categories included in the global model. Here, the category into which the heartbeat profile of the current subject is classified in the global model may be identified as the closest category. In some embodiments, the estimation device may use pattern matching to identify the closest category.

[0099] As shown in reference 125, the estimation device can identify the local subject associated with the closest category. In some embodiments, the local subject is identified based on the subject identifier associated with the closest category. For example, as described above, the estimation device can identify the closest category for the current subject. Here, since each category of the global model is associated with a subject identifier, the estimation device can identify the local subject associated with the closest category based on the subject identifier of the closest category.

[0100] like Figure 1B As shown by mark 130 in , the estimation device may generate a BPL estimation model of the current subject based on the heartbeat profile data of the local subject associated with the closest category (referred to herein as the identified local subject).

[0101] In some embodiments, when generating a BPL estimation model, the estimation device may select at least two categories associated with the identified local subject, and generate the BPL estimation model based on the heartbeat profile data of the at least two categories associated with the identified local subject. In some embodiments, the at least two categories may include a closest category and at least one other category associated with the identified local subject (e.g., at least one other category associated with a BPL identifier that is different from the BPL identifier of the closest category). In some embodiments, when selecting at least two categories, the estimation device may identify a linear region of a set of categories associated with the identified local subject, and select at least two categories based on the linear region of the set of categories associated with the identified local subject. Therefore, in some embodiments, the linear region of the identified local subject can be used in association with an in-situ generated BPL estimation model (e.g., a local regression model) that can be used to estimate the BPL of the current subject.

[0102] In practice, different current subject's heartbeat profiles may lead to the selection of different class combinations as the basis for generating the BPL estimation model, depending on the location and classifier used. Figure 3An example is shown associated with the validation set of the example current subject's data shown within the PCA-SVM plot of the identified local subject. Here, it can be seen that the unknown spectrum of the current subject's heartbeat profile data (in Figure 3 BP68 CS The star symbol symbol) is located in the first category associated with the first BPL of the local subject (in Figure 3 Zhongyou area BP57 LS identification) and a second category associated with a second BPL of the local subject (in Figure 3 Zhongyou area BP88 LS identification) or a third category associated with a third BPL of a local subject (in Figure 3 Zhongyou Area BP117 LS In this example, 57 of the 70 samples associated with the current subject select the first category and the second category as the two categories based on which the BPL estimation model is generated, and 13 of the 70 samples select the first category and the third category as the two categories based on which the BPL estimation model is generated. It is worth noting that in some scenarios, these two categories may not be the two categories associated with the BPL that is closest to the BPL of the current subject. Figure 4 An example of this is shown in . Figure 4 In the unknown spectrum of the current subject's heartbeat profile data (given by Figure 4 Marked as BP119 CS The star symbol symbol) is located in the first category associated with the first BPL of the local subject (in Figure 4 Zhongyou area BP72 LS identification) and a second category associated with a second BPL of the local subject (in Figure 4 Zhongyou Area BP152 LS identification) or a third category associated with a third BPL of a local subject (in Figure 4 Zhongyou Area BP174 LS In this example, some samples associated with the current subject select the first and third categories as the two categories (even though the second category has a BPL that is closer to the actual BPL of the current subject than the first or third categories). In some embodiments, this selection is a result of one category exhibiting nonlinearity relative to the other categories in the x matrix space. As indicated above, Figure 3 and Figure 4 It is provided as an example only. Other examples may be related to Figure 3 and Figure 4 The examples described are different.

[0103] In some embodiments, as described above, the estimation device can generate a BPL estimation model based on at least two categories of heartbeat profile data associated with a local subject. For example, the estimation device can obtain heartbeat profile data associated with at least two categories and can generate a local regression model based on the heartbeat profile data. In some embodiments, the BPL estimation model generated based on the heartbeat profile data associated with at least two categories can be a quantitative model generated using partial least squares regression. In some embodiments, the BPL estimation provided by this BPL estimation model can include an estimated BPL of the current subject (e.g., a quantitative estimate of the BPL of the current subject). In some embodiments, the BPL estimate can be a qualitative BPL estimate (e.g., an indication of whether the BPL of the current subject increases, decreases, or remains unchanged, or an indication of whether the current subject is in hypotension, normal state, prehypertension, hypertension stage 1, hypertension stage 2, or hypertensive crisis).

[0104] In some embodiments, a signal point calibration (SPC) usage model can be used in conjunction with generating a BPL estimation model. Here, to generate the BPL estimation model, the estimation device can determine a local subject transfer set associated with the identified local subject. The local subject transfer set can include one or more heartbeat profiles of the identified local subject collected at a reference BPL. Next, the estimation device can obtain a current subject transfer set associated with the current subject. The current subject transfer set can include one or more heartbeat profiles of the current subject collected at the reference BPL. According to the SPC usage model, PPG signals for the current subject should be (previously) collected at one or more known BPLs to provide a set of heartbeat profiles associated with the current subject. This set of heartbeat profiles can then be used to perform model transfer, which can reduce the bias associated with cross-subject model predictions. In some embodiments, the set of heartbeat profiles for the current subject can include one or more heartbeat profiles collected at a reference BPL (e.g., measured by a reference blood pressure monitor (such as a clinically approved blood pressure monitor, a home blood pressure monitor, etc.)). In some embodiments, a set of heartbeat profiles for the current subject includes heartbeat profiles associated with two or more BPLs: one or more heartbeat profiles associated with a reference BPL and one or more heartbeat profiles associated with a BPL that differs from the reference BPL (e.g., differs by a specific amount).

[0105] In some embodiments, the estimation device can create a current subject set of transmission based on the current subject transmission set and the local subject transmission set. The current subject set of transmission can include a plurality of heartbeat profiles of transmission associated with the current subject. As an example, the estimation device can use mean difference correction (MDC) to create the current subject set of transmission. According to MDC, the centroid of the current subject transmission set can be mapped to the centroid of the local subject transmission set. Here, when the centroid is mapped to the same position, the deviation can be eliminated. The result of the mapping is the current subject transmission set of transmission. In some embodiments, the estimation device can use, for example, MDC, segmented direct standardization, generalized least squares method, orthogonal signal correction method or other technology to create the current subject of transmission. It is worth noting that in some cases, the BPL estimation model can be used to provide a classification output (rather than an estimated BPL value).

[0106] Next, the estimation device may generate a BPL estimation model based on a set of local subjects associated with the identified local subjects (e.g., a linear region of a set of heartbeat profiles for the local subjects). For example, the estimation device may generate the BPL estimation model based on the set of local subjects using linear regression. In some embodiments, the BPL estimation model generated based on the set of local subjects associated with the identified local subjects may be a quantitative model generated using partial least squares regression. In some embodiments, the BPL estimate provided by this BPL estimation model may include an estimated BPL of the current subject (e.g., a quantitative estimate of the BPL of the current subject). In some embodiments, the BPL estimation model generated based on the set of local subjects associated with the identified local subjects may be a qualitative model generated using a classifier. In some embodiments, the BPL estimate provided by this BPL estimation model may include a classification output associated with the current subject, such as an indication of whether the BPL of the current subject has increased, decreased, or remained unchanged, or an indication of whether the current subject is in a specific state (e.g., hypotension, normal, prehypertension, hypertension stage 1, hypertension stage 2, hypertensive crisis, etc.).

[0107] like Figure 1BAs shown in the mark 135 in, the estimation device can determine a BPL estimate for the current subject based on the transmitted current subject set and using a BPL estimation model. For example, the estimation device can provide information associated with the transmitted current subject set (e.g., information associated with one or more transmitted heartbeat profiles of the current subject) as input to the BPL estimation model, and can receive a BPL estimate associated with the current subject as output. In some embodiments, as described above, the BPL estimate can be a quantitative BPL estimate (e.g., an estimated BPL) or a qualitative BPL estimate (e.g., an indication of whether the BPL of the current subject is increasing, decreasing, or unchanged, or an indication of whether the current subject is in hypotension, a normal state, prehypertension, hypertension stage 1, hypertension stage 2, or a hypertensive crisis).

[0108] The estimation device may provide information associated with the current subject's BPL estimate, as indicated by reference numeral 140. For example, in some embodiments, the estimation device may provide information associated with the BPL estimate for display (e.g., via a display screen of a multispectral sensor device, via a display screen of a monitoring device, etc.).

[0109] In some embodiments, the estimation device may repeat the above process (eg, automatically, periodically, etc. based on a user's instruction) to enable monitoring of the current subject's BPL (eg, in real time or near real time).

[0110] In this way, the estimation device can non-invasively estimate and / or monitor the BPL of the current subject based on a heartbeat profile generated from the PPG signal (e.g., based on full-spectrum NIR). In some cases, beat-by-beat blood pressure monitoring can be performed. It is worth noting that the above-mentioned technology can be used for single-channel PPG and multi-channel PPG.

[0111] As mentioned above, Figure 1A and Figure 1B It is provided as an example only. Other examples may be related to Figure 1A and Figure 1B The examples described are different.

[0112] Figure 5 is an illustration of an example environment 500 in which the systems and / or methods described herein may be implemented. Figure 5 As shown, environment 500 may include a multispectral sensor device 505, an estimation device 510, and a network 515. The devices in environment 500 may be interconnected via wired connections, wireless connections, or a combination of wired and wireless connections. Multispectral sensor device 505 and estimation device 510 may correspond to the above combined devices. Figure 1A and Figure 1B The described multi-spectral sensor device and estimation device.

[0113] As described herein, the multi-spectral sensor device 505 includes a device capable of measuring, collecting, gathering, or otherwise determining PPG data associated with a plurality of wavelength channels. For example, the multi-spectral sensor device 505 can include a multi-spectral sensing device capable of determining PPG data (in the form of multi-variate time series data) on each of 64 wavelength channels. In some implementations, the multi-spectral sensor device 505 can operate in the visible spectrum, near-infrared spectrum, infrared spectrum, etc. In some implementations, the multi-spectral sensor device 505 can be a wearable device (e.g., a device that can be worn on a wrist, finger, arm, leg, head, ear, etc.). In some implementations, the multi-spectral sensor device 505 can be integrated with the estimation device 510 (e.g., such that the multi-spectral sensor device 505 and the estimation device 510 are on the same chip, in the same package, in the same housing, etc.). Alternatively, in some implementations, the multi-spectral sensor device 505 can be separate from the estimation device 510. In some implementations, the multi-spectral sensor device 505 can receive information from and / or send information to another device in the environment 500, such as the estimation device 510.

[0114] As described herein, the estimation device 510 includes a device capable of performing one or more operations associated with blood pressure estimation based on PPG data. For example, the estimation device 510 can include an application specific integrated circuit (ASIC), an integrated circuit, a server, a group of servers, etc., and / or another type of communication and / or computing device. In some implementations, the estimation device 510 can be integrated with the multi-spectral sensor device 505 (e.g., such that the multi-spectral sensor device 505 and the estimation device 510 are on the same chip, in the same package, in the same housing, etc.). Alternatively, in some implementations, the estimation device 510 can be separate from the multi-spectral sensor device 505. In some implementations, the estimation device 510 can receive information from and / or send information to another device in the environment 500, such as the multi-spectral sensor device 505.

[0115] The network 515 may include one or more wired and / or wireless networks. For example, the network 515 may include a wired network (e.g., when the multispectral sensor device 505 and the estimation device 510 are included in the same package and / or the same chip). As another example, the network 515 may include a cellular network (e.g., a long-term evolution (LTE) network, a code division multiple access (CDMA) network, a 3G network, a 4G network, a 5G network, or another type of next-generation network), a public land mobile network (PLMN), a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a telephone network (e.g., a public switched telephone network (PSTN)), a private network, an ad hoc network, an intranet, the Internet, a fiber-optic-based network, a cloud computing network, etc., and / or combinations of these or other types of networks. In some embodiments, the multispectral sensor device 505 and the estimation device 510 may communicate wirelessly, such as via Bluetooth, NFC, RF, etc.

[0116] Figure 5 The number and arrangement of devices and networks shown are provided as examples. Figure 5 There may be additional devices and / or networks, fewer devices and / or networks, different devices and / or networks, or differently arranged devices and / or networks than those shown. Figure 5 Two or more of the devices shown may be implemented in a single device, or Figure 5 The single device shown may be implemented as multiple distributed devices. Additionally or alternatively, one or more devices of the environment 500 may perform one or more functions described as being performed by another set of devices in the environment 500.

[0117] Figure 6 is a diagram of example components of device 600. Device 600 may correspond to multispectral sensor device 505 and / or estimation device 510. In some embodiments, multispectral sensor device 505 and / or estimation device 510 may include one or more devices 600 and / or one or more components of device 600. Figure 6 As shown, device 600 may include a bus 610 , a processor 620 , a memory 630 , a storage component 640 , an input component 650 , an output component 660 , and a communication interface 670 .

[0118] The bus 610 includes components that allow communication among the multiple components of the device 600. The processor 620 is implemented in hardware, firmware, and / or a combination of hardware and software. The processor 620 is a central processing unit (CPU), a graphics processing unit (GPU), an accelerated processing unit (APU), a microprocessor, a microcontroller, a digital signal processor (DSP), a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or another type of processing component. In some embodiments, the processor 620 includes one or more processors that can be programmed to perform functions. The memory 630 includes a random access memory (RAM), a read-only memory (ROM), and / or another type of dynamic or static storage device (e.g., flash memory, magnetic memory, optical memory, etc.) that stores information and / or instructions for use by the processor 620.

[0119] Storage component 640 stores information and / or software related to the operation and use of device 600. For example, storage component 640 may include a hard disk (e.g., a magnetic, optical, and / or magneto-optical disk), a solid-state drive (SSD), a compact disk (CD), a digital versatile disk (DVD), a floppy disk, a cassette, a magnetic tape, and / or another type of non-transitory computer-readable medium along with a corresponding drive.

[0120] Input components 650 include components that allow device 600 to receive information, for example, via a user input (e.g., a touch screen display, a keyboard, a keypad, a mouse, buttons, switches, and / or a microphone). Additionally or alternatively, input components 650 may include components for determining location (e.g., a global positioning system (GPS) component) and / or sensors (e.g., an accelerometer, a gyroscope, an actuator, another type of positioning or environmental sensor, etc.). Output components 660 include components that provide output information from device 600 (e.g., via a display, a speaker, a tactile feedback component, an audio or visual indicator, etc.).

[0121] The communication interface 670 includes transceiver-like components (e.g., a transceiver, a separate receiver, a separate transmitter, etc.) that enable the device 600 to communicate with other devices, such as via a wired connection, a wireless connection, or a combination of wired and wireless connections. The communication interface 670 can allow the device 600 to receive information from another device and / or provide information to another device. For example, the communication interface 670 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, a Wi-Fi interface, a cellular network interface, etc.

[0122] Device 600 can perform one or more of the processes described herein. Device 600 can perform these processes based on processor 620 executing software instructions stored by non-transitory computer-readable media (e.g., memory 630 and / or storage component 640). As used herein, the term "computer-readable medium" refers to a non-transitory storage device. A memory device includes storage space within a single physical storage device or storage space spread across multiple physical storage devices.

[0123] The software instructions may be read from another computer-readable medium or from another device into the memory 630 and / or storage component 640 via the communication interface 670. When executed, the software instructions stored in the memory 630 and / or storage component 640 may cause the processor 620 to perform one or more processes described herein. Additionally or alternatively, hardware circuitry may be used in place of or in combination with software instructions to perform one or more processes described herein. Thus, the implementations described herein are not limited to any specific combination of hardware circuitry and software.

[0124] Figure 6 The number and arrangement of components shown are provided as examples. Figure 6 Device 600 may include additional components, fewer components, different components, or differently arranged components than those shown. Additionally or alternatively, one or more components of device 600 may perform one or more functions described as being performed by another group of components of device 600.

[0125] Figure 7 is a flow chart of an example process 700 for determining a BPL estimate based on PPG data as described herein. In some embodiments, Figure 7 One or more process blocks of may be performed by an estimation device (eg, estimation device 510). In some embodiments, Figure 7 One or more process blocks of may be performed by another device or group of devices, such as a multispectral sensor device (eg, multispectral sensor device 505 ), separate from or including the estimation device.

[0126] like Figure 7 As shown, process 700 may include obtaining a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels (block 710). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may obtain a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels.

[0127] like Figure 7 As further shown in FIG, process 700 may include determining a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier (block 720). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may determine a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier.

[0128] like Figure 7 As further shown in FIG, process 700 may include identifying one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject (block 730). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject.

[0129] like Figure 7 As further shown in , process 700 may include identifying a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category (block 740). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category.

[0130] like Figure 7 As further shown in FIG, process 700 may include selecting at least two categories associated with the local subject for use in generating a BPL estimation model for the current subject (block 750). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may select at least two categories associated with the local subject for use in generating a BPL estimation model for the current subject. In some embodiments, the at least two categories include a closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier that is different from the BPL identifier of the closest category.

[0131] like Figure 7As further shown in process 700 can include generating a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject (block 760). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can generate a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject.

[0132] As further shown in process 700 can include generating a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject (block 760). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can generate a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject. Figure 7 As further shown in process 700 can include generating a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject (block 760). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can generate a BPL estimation model based on the heartbeat profile data for the at least two classes associated with the local subject.

[0133] Process 700 can include additional implementations, such as any single implementation of or any combination of the one or more other processes described below and / or elsewhere described herein.

[0134] In some implementations, determining the global model includes obtaining a plurality of heartbeat profiles associated with a plurality of local subjects (each heartbeat profile of the plurality of heartbeat profiles is a heartbeat profile of one of the plurality of local subjects collected at a known BPL), and generating the global model based on the plurality of heartbeat profiles associated with the plurality of local subjects.

[0135] In some implementations, the global model is generated using a non-linear classifier.

[0136] In some implementations, selecting the at least two classes for generating the BPL estimation model for the current subject includes: identifying a linear region of a set of classes associated with the local subject, the set of classes including the closest class and at least one other class; and selecting the at least two classes based on the linear region of the set of classes associated with the local subject.

[0137] In some implementations, the BPL estimation model is a quantitative model generated using a least squares regression method.

[0138] In some implementations, the BPL estimation model is a qualitative model generated using a classifier.

[0139] In some implementations, the BPL estimation includes an estimated BPL of the current subject.

[0140] In some embodiments, process 700 includes providing information associated with a current subject's BPL estimate.

[0141] Although Figure 7 Example blocks of process 700 are shown, but in some implementations, Figure 7 Process 700 may include additional blocks, fewer blocks, different blocks, or differently arranged blocks than those depicted in . Additionally or alternatively, two or more blocks of process 700 may be performed in parallel.

[0142] Figure 8 is a flow chart of an example process 800 for determining a BPL estimate based on PPG data as described herein. In some embodiments, Figure 8 One or more process blocks of may be performed by an estimation device (eg, estimation device 510). In some embodiments, Figure 8 One or more process blocks of may be performed by another device or group of devices, such as a multispectral sensor device (eg, multispectral sensor device 505), separate from or including the estimation device.

[0143] like Figure 8 As shown, process 800 may include obtaining a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels (block 810). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may obtain a heartbeat profile of a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels.

[0144] like Figure 8 As further shown in FIG, process 800 may include determining a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier (block 820). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may determine a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier.

[0145] like Figure 8As further shown in FIG, process 800 may include identifying one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject (block 830). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject.

[0146] like Figure 8 As further shown in , process 800 may include identifying a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category (block 840). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category.

[0147] like Figure 8 As further shown, process 800 may include determining a local subject transmission set associated with the local subject based on identifying the local subject, the local subject transmission set including one or more heartbeat profiles of the local subject collected at a reference BPL (block 850). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may determine a local subject transmission set associated with the local subject based on identifying the local subject, the local subject transmission set including one or more heartbeat profiles of the local subject collected at or near the reference BPL (e.g., at the reference BPL or at the BPL closest to the reference BPL).

[0148] like Figure 8 As further shown, process 800 may include obtaining a current subject transfer set associated with the current subject, the current subject transfer set including one or more heartbeat profiles of the current subject collected at a reference BPL (block 860). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may obtain a current subject transfer set associated with the current subject, the current subject transfer set including one or more heartbeat profiles of the current subject collected at a reference BPL.

[0149] As Figure 8 Further, process 800 can include creating a set of current subjects of transmissions based on the set of current subject transmissions and the set of local subject transmissions (block 870). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can create a set of current subjects of transmissions based on the set of current subject transmissions and the set of local subject transmissions. In some implementations, the set of current subjects of transmissions includes a plurality of transmission heartbeat profiles associated with the current subject.

[0150] As Figure 8 Further, process 800 can include generating a BPL estimation model based on the set of local subjects associated with the identified local subject (block 880). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can generate a BPL estimation model based on the set of local subjects.

[0151] As Figure 8 Further, process 800 can include determining a BPL estimate for the current subject based on the set of current subjects of transmissions and using the BPL estimation model (block 890). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) can determine a BPL estimate for the current subject based on the set of current subjects of transmissions and using the BPL estimation model.

[0152] Process 800 can include additional implementations, such as any single implementation of or any combination of the implementations described below and / or in one or more other processes described elsewhere herein.

[0153] In some implementations, determining the global model includes: obtaining a plurality of heartbeat profiles associated with a plurality of local subjects, wherein each heartbeat profile of the plurality of heartbeat profiles is a heartbeat profile of one of the plurality of local subjects collected at a known BPL; and generating the global model based on the plurality of heartbeat profiles associated with the plurality of local subjects.

[0154] In some implementations, generating the global model uses a non-linear classifier.

[0155] In some embodiments, the transferred current subject set is created using mean difference correction, piecewise direct normalization, generalized least squares, orthogonal signal correction, and / or one or more additional calibration transfer methods.

[0156] In some embodiments, the BPL estimation model is a quantitative model generated using partial least squares regression.

[0157] In some embodiments, the BPL estimate comprises the estimated BPL of the current subject.

[0158] In some embodiments, a BPL estimation model is a qualitative model generated using a classifier.

[0159] In some embodiments, process 800 includes providing information associated with a current subject's BPL estimate.

[0160] Although Figure 8 Example blocks of process 800 are shown, but in some implementations, Figure 8 Process 800 may include additional blocks, fewer blocks, different blocks, or differently arranged blocks compared to those depicted in . Additionally or alternatively, two or more blocks of process 800 may be performed in parallel.

[0161] Figure 9 is a flow chart of an example process 900 for determining a BPL estimate based on PPG data as described herein. In some embodiments, Figure 9 One or more process blocks of can be performed by an estimation device (eg, estimation device 510). In some embodiments, Figure 9 One or more process blocks of may be performed by another device or group of devices, such as a multispectral sensor device (eg, multispectral sensor device 505 ), separate from or including the estimation device.

[0162] like Figure 9 As shown, process 900 may include obtaining a heartbeat profile for a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels (block 910). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may obtain a heartbeat profile for a current subject, the heartbeat profile being based on PPG data associated with a set of wavelength channels.

[0163] like Figure 9As further shown in FIG, process 900 may include determining a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier (block 920). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may determine a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a BPL identifier.

[0164] like Figure 9 As further shown in FIG, process 900 may include identifying one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject (block 930). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify one of the multiple categories included in the global model as a closest category for the current subject, the closest category identified based on the heartbeat profile of the current subject.

[0165] like Figure 9 As further shown in , process 900 may include identifying a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category (block 940). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may identify a local subject associated with the closest category, the local subject being identified based on a subject identifier associated with the closest category.

[0166] like Figure 9 As further shown in FIG9 , process 900 may include generating a BPL estimation model for the current subject based on the heartbeat profile data for the local subject (block 950). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may generate a BPL estimation model for the current subject based on the heartbeat profile data for the local subject.

[0167] like Figure 9As further shown in FIG, process 900 may include determining a BPL estimate for the current subject based on the heartbeat profile of the current subject and using a BPL estimation model (block 960). For example, as described above, the estimation device (e.g., using processor 620, memory 630, storage component 640, input component 650, output component 660, communication interface 670, etc.) may determine a BPL estimate for the current subject based on the heartbeat profile of the current subject and using a BPL estimation model.

[0168] Process 900 may include additional implementations, such as any single implementation or any combination of implementations described below and / or in combination with one or more other processes described elsewhere herein.

[0169] In some embodiments, generating the BPL estimation model includes selecting at least two categories associated with the local subject for generating the BPL estimation model of the current subject, the at least two categories including a closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier that is different from the BPL identifier of the closest category, and generating the BPL estimation model based on the heartbeat profile data of the at least two categories associated with the local subject.

[0170] In some embodiments, generating a BPL estimation model includes determining a local subject transmission set associated with the local subject, the local subject transmission set including one or more heartbeat profiles of the local subject collected at or near a reference BPL (e.g., at the reference BPL or at a BPL closest to the reference BPL); obtaining a current subject transmission set associated with the current subject, the current subject transmission set including one or more heartbeat profiles of the current subject collected at the reference BPL; creating a transmitted current subject set based on the current subject transmission set and the local subject transmission set, the transmitted current subject set including multiple transmitted heartbeat profiles associated with the current subject, and, based on the local subject set associated with the identified local subject, generating a BPL estimation model, wherein a BPL estimate for the current subject is determined based on the transmitted current subject set and using the BPL estimation model.

[0171] In some embodiments, determining a global model includes: obtaining a plurality of heartbeat profiles associated with a plurality of local subjects, each of the plurality of heartbeat profiles being a heartbeat profile of a local subject among the plurality of local subjects collected at a known BPL, and generating a global model based on the plurality of heartbeat profiles associated with the plurality of local subjects, and generating the global model using a nonlinear classifier.

[0172] In some embodiments, the BPL estimate comprises the estimated BPL of the current subject.

[0173] Some embodiments described herein provide devices (e.g., estimation device 510, multispectral sensor device 505, etc.) that provide improved PPG-based BPL estimation. In some embodiments, the device generates a BPL estimation model for a current subject (i.e., the subject for which a BPL estimate is determined) based on heartbeat profile data for a local subject (i.e., a specific subject identified from a global model associated with multiple subjects), and determines a BPL estimate for the current subject based on the current subject's heartbeat profile and using the BPL estimation model.

[0174] The foregoing disclosure provides illustration and description, but is not intended to be exhaustive or to limit the embodiments to the precise form disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from practice of the embodiments.

[0175] As used herein, the term "component" is intended to be broadly interpreted as hardware, firmware, and / or a combination of hardware and software.

[0176] Some embodiments are described herein in conjunction with threshold values. As used herein, satisfying a threshold value may refer to a value being greater than a threshold value, being more than a threshold value, being higher than a threshold value, being greater than or equal to a threshold value, being less than a threshold value, being less than a threshold value, being lower than a threshold value, being less than or equal to a threshold value, being equal to a threshold value, etc., depending on the context.

[0177] It will be apparent that the systems and / or methods described herein can be implemented in various forms of hardware, firmware, or a combination of firmware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not a limitation of implementation. Thus, the operation and behavior of the systems and / or methods are described herein without reference to specific software code, with the understanding that software and hardware can be designed to implement the systems and / or methods based on the description herein.

[0178] Even though particular combinations of features are recited in the claims and / or disclosed in the specification, these combinations are not intended to limit the disclosure of the various embodiments. In fact, many of these features can be combined in ways not specifically recited in the claims and / or disclosed in the specification. Although each dependent claim listed below may be directly dependent on only one claim, the disclosure of the various embodiments includes each dependent claim in combination with every other claim in the claim set.

[0179] Unless so clearly described, the elements, actions or instructions used herein should not be interpreted as key or necessary. In addition, as used herein, the articles "a" and "an" are intended to include one or more projects and can be used interchangeably with "one or more". In addition, as used herein, the article "the" is intended to include one or more projects mentioned in association with the article "the" and can be used interchangeably with "the one or more". In addition, as used herein, the term "set" is intended to include one or more projects (for example, related projects, unrelated projects, the combination of related projects and unrelated projects, etc.), and can be used interchangeably with "one or more". In the case of intending only one project, the phrase "only one" or similar language is used. In addition, as used herein, the term "has", "have", "having" or similar terms are intended to be open terms. In addition, the phrase "based on" is intended to mean "at least partially based on", unless otherwise clearly stated. Furthermore, as used herein, the term "or" when used in a series is intended to be inclusive and can be used interchangeably with "and / or" unless expressly stated otherwise (e.g., if used in combination with "either" or "only one of...").

Claims

1. A method for estimating blood pressure, comprising: obtaining, by the device, a heartbeat profile of a current subject, the heartbeat profile being based on photoplethysmography (PPG) data associated with a set of wavelength channels; determining, by the apparatus, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a blood pressure level (BPL) identifier; identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on a heartbeat profile of the current subject; identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category; selecting, by the device, at least two categories associated with the local subject for use in generating a BPL estimation model for the current subject, wherein the at least two categories include the closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier different from the BPL identifier of the closest category; generating, by the device, the BPL estimation model based on the heartbeat profile data regarding the at least two categories associated with the local subject; and A BPL estimate for the current subject is determined, by the apparatus, based on the heartbeat profile of the current subject and using the BPL estimation model.

2. The method according to claim 1, wherein Determining the global model includes: obtaining a plurality of heartbeat profiles associated with a plurality of local subjects, wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects.

3. The method according to claim 1, wherein The global model is generated using a nonlinear classifier.

4. The method according to claim 1, wherein Selecting the at least two categories for generating a BPL estimation model for the current subject includes: identifying a linear region of a set of categories associated with the local subject, the set of categories including the closest category and the at least one other category; and The at least two categories are selected based on linear regions of the set of categories associated with the local subject.

5. The method according to claim 1, wherein The BPL estimation model is a quantitative model generated using the partial least squares regression method.

6. The method according to claim 1, wherein The BPL estimation model is a qualitative model generated using a classifier.

7. The method according to claim 1, wherein The BPL estimate includes an estimated BPL for the current subject.

8. The method according to claim 1, further comprising: Information associated with the BPL estimate for the current subject is provided.

9. A method for estimating blood pressure, comprising: obtaining, by the device, a heartbeat profile of a current subject, the heartbeat profile being based on photoplethysmography (PPG) data associated with a set of wavelength channels; determining, by the apparatus, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a blood pressure level (BPL) identifier; identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on a heartbeat profile of the current subject; identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category; determining, by the device and based on identifying the local subject, a local subject transmission set associated with the local subject, the local subject transmission set comprising one or more heartbeat profiles of the local subject collected at or near a reference BPL; obtaining, by the device, a current subject delivery set associated with the current subject, the current subject delivery set comprising one or more heartbeat profiles of the current subject collected at the reference BPL; creating, by the device, a delivered current subject set based on the current subject delivered set and the local subject delivered set, wherein the transferred current subject set comprises a plurality of transferred heartbeat profiles associated with the current subject; generating, by the apparatus, a BPL estimation model based on a set of local subjects associated with the identified local subjects; and A BPL estimate for the current subject is determined, by the device, based on the communicated set of current subjects and using the BPL estimation model.

10. The method according to claim 9, wherein: Determining the global model includes: obtaining a plurality of heartbeat profiles associated with a plurality of local subjects, wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects.

11. The method according to claim 9, wherein The global model is generated using a nonlinear classifier.

12. The method according to claim 9, wherein The current subject set for the transfer is created using one of the following: Mean difference correction, Segment-wise direct standardization, Generalized least squares method, Quadrature signal correction method, or One or more additional model transfer methods.

13. The method according to claim 9, wherein: The BPL estimation model is a quantitative model generated using the partial least squares regression method.

14. The method according to claim 9, wherein The BPL estimate includes an estimated BPL for the current subject.

15. The method according to claim 9, wherein The BPL estimation model is a qualitative model generated using a classifier.

16. The method according to claim 9, further comprising: Information associated with the BPL estimate for the current subject is provided.

17. A method for performing blood pressure estimation, comprising: obtaining, by the device, a heartbeat profile of a current subject, the heartbeat profile being based on photoplethysmography (PPG) data associated with a set of wavelength channels and determined by the sensor device; determining, by the apparatus, a global model comprising a plurality of categories, each category in the plurality of categories being associated with a subject identifier and a blood pressure level (BPL) identifier; identifying, by the device, one of the plurality of categories included in the global model as a closest category for the current subject, the closest category being identified based on a heartbeat profile of the current subject; identifying, by the device, a local subject associated with the closest category, the local subject being identified based on the subject identifier associated with the closest category; generating, by the device, a BPL estimation model for the current subject based on the heartbeat profile data of the local subject; and A BPL estimate for the current subject is determined, by the apparatus, based on the heartbeat profile of the current subject and using the BPL estimation model.

18. The method according to claim 17, wherein Generating the BPL estimation model includes: selecting at least two categories associated with the local subject for use in generating the BPL estimation model for the current subject, wherein the at least two categories include the closest category and at least one other category associated with the local subject, the at least one other category being associated with a BPL identifier different from the BPL identifier of the closest category; and The BPL estimation model is generated based on heartbeat profile data about the at least two categories associated with the local subject.

19. The method according to claim 17, wherein Generating the BPL estimation model includes: determining a local subject transfer set associated with the local subject, the local subject transfer set comprising one or more heartbeat profiles of the local subject collected at a reference BPL; obtaining a current subject delivery set associated with the current subject, the current subject delivery set comprising one or more heartbeat profiles of the current subject collected at the reference BPL; creating a delivered current subject set based on the current subject delivery set and the local subject delivery set, wherein the transferred current subject set comprises a plurality of transferred heartbeat profiles associated with the current subject; and generating the BPL estimation model based on a set of local subjects associated with the identified local subjects, Wherein the BPL estimate for the current subject is determined based on the transferred current subject set and using the BPL estimation model.

20. The method according to claim 17, wherein Determining the global model includes: obtaining a plurality of heartbeat profiles associated with a plurality of local subjects, wherein each of the plurality of heartbeat profiles is a heartbeat profile of a local subject of the plurality of local subjects collected at a known BPL; and The global model is generated based on the plurality of heartbeat profiles associated with the plurality of local subjects, the global model being generated using a nonlinear classifier.

Images