WO2023026303A1

WO2023026303A1 - An information processing apparatus and method for continuous estimation of arterial blood pressure

Info

Publication number: WO2023026303A1
Application number: PCT/IN2022/050756
Authority: WO
Inventors: Thalansh Batra
Original assignee: Thalansh Batra
Priority date: 2021-08-26
Filing date: 2022-08-26
Publication date: 2023-03-02

Abstract

The invention provides an apparatus for estimation of Arterial Blood Pressure continuously and non-invasively using Photoplethysmography signals. The apparatus contains a wearable device (101) configured to acquire bio-information data includes Photoplethysmography (PPG) signal data and accelerometer data, an electronic device (103) configured to receive the bio-information data, and a server (105) communicatively coupled with the wearable device (101) and the electronic device (103). The server (105) is configured to receive the bio-information data from at least one of the wearable device (101) or the electronic device (103), extract features from the bio-information data, normalize the extracted features, input the normalized features and corresponding Arterial Blood Pressure (ABP) signals to a Machine learning (ML) model and output, predicted by the ML, ABP waveform for each input point value of PPG signal or multiple ABP point value outputs for multiple PPG point value inputs (MIMO model).

Description

AN INFORMATION PROCESSING APPARATUS AND

METHOD FOR CONTINUOUS ESTIMATION OF

ARTERIAL BLOOD PRESSURE

FIELD OF INVENTION

[001] The present invention generally relates to estimation of Arterial Blood

Pressure continuously and non-invasively using Photoplethysmography signals.

BACKGROUND OF THE INVENTION

[002] Blood pressure (BP) is one of the basic medical parameters used to diagnose human health condition. The most accurate methods for BP measurements involve insertion of a catheter into a human artery. However, the BP measurements using a catheter are invasive and costly since they require a medical professional to perform the measurements and, typically, can only be performed in a medical facility environment.

[003] Less accurate methods for BP measurements include use of an inflatable cuff to pressurize a blood artery. There are numerous cuff -based portable devices for BP measurements that patients can use at home and do not require assistance of a medical professional. However, cuff-based measurements require inflation and deflation of the inflatable cuff. Therefore, such devices are cumbersome to use and not suitable for continuous BP measurements.

[004] Some cuff-less devices for BP measurements use an electrical sensor to measure an electrocardiogram (ECG) signal and optical sensor to measure a photoplethysmogram signal (PPG). The ECG and PPG can be analysed to determine Pulse Transit Time (PTT). Because PTT is in-part inversely proportional to BP, the BP can in some cases be determined from PTT using a pre-defined relationship. However, changes in a cardio-vascular status of a patient require often re-calibration of PTT based blood pressure measurements. Cuff-less devices can potentially provide continuous monitoring of BP while imposing a minimal burden on normal activities when worn on various body parts such as a finger, a wrist, or an ankle.

[005] Determining BP based on the PTT alone may not be sufficiently accurate because of other cardiovascular parameters affecting hemodynamics such as vascular resistance, cardiac output, pulse rate (PR), temperature of a finger (if PPG is measured at the finger), and so forth. To compensate for influences of other parameters, some existing techniques for measuring of BP using PPG include applying correction factors to account for the vascular resistance and age of patient. The correction factors can be determined by an empirical formula. Some other techniques attempt to determine compensation factors to compensate for various additional influences (for example, contacting force to sensors, nervous activity and cardiac output of patient, and ambient temperature). The compensation factors can be determined using a calibration process.

[006] The patent document WO2016138965A1 describes a method for determining a blood pressure value that comprises the steps of: providing a pulsatile signal (1), determining a time -related feature and a normalized amplitude -related feature on the basis of the pulsatile signal; and calculating a blood pressure value on the basis of a blood pressure function depending on the time -related feature, the normalized amplitude-related feature and function parameters. However, method does not provide continuous and accurate values of the Arterial Blood Pressure.

[007] The patent document US9451920B2 describes a system for measuring of arterial and venous blood constituent concentration based first on measuring cardiac blood flow balance parameter between the right chamber of the heart and the left chamber of the heart, which includes a sensor device for measuring one of blood pressure and blood flow rate and blood constituent concentration of a patient so as to generate an arterial pulse signal. A processing unit is responsive to the arterial pulse signal for generating full arterial pulse plethysmography waveforms, arterio-venous pulse plethysmography waveforms, and balance parameters. A computational device that is responsive to plethysmography waveforms generating a plurality of state space linear transfer functions by applying system identification between plethysmography waveforms at various wavelengths representing a plurality of models of the blood constituent concentration, including oxygen, carbon dioxide, haemoglobin, and glucose, and displaying related useful information. However, the method and complex arrangement of the system leads to several drawbacks.

[008] Therefore, in order to overcome these problems, there is a need for a device or system for estimating Arterial Blood Pressure continuously and non-invasively using Photoplethysmography signals.

OBJECT OF THE INVENTION

[009] The principal object of the invention is to introduce different methods for non-invasive and continuous estimation of arterial blood pressure waveform from photoplethysmography (PPG) signal.

[0010] Another object of the invention is to apply the method to any pulse oximetry system which will allow for low-cost measurement of blood pressure parameters in a small form factor.

[0011] Another object of the invention is to reduce healthcare expenditures on separate blood pressure monitoring machines and could also easily allow for remote outpatient monitoring of blood pressure.

[0012] These and other objects and characteristics of the present invention will become apparent from the further disclosure to be made in the detailed description given below.

SUMMARY OF THE INVENTION

[0013] This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0014] The invention provides an information processing apparatus for continuous estimation of arterial blood pressure. The apparatus contains a wearable device configured to acquire bio -information data, wherein the bio -information data includes Photoplethysmography (PPG) signal data and accelerometer data, an electronic device configured to receive the bio -information data; a server communicatively coupled with the wearable device and the electronic device. The server is configured to receive the bio-information data from at least one of the wearable device or the electronic device. The server is further configured to extract features from the bio-information data, normalize the extracted features, input the normalized features and corresponding Arterial Blood Pressure (ABP) signals to a Machine Learning (ML) model or a Deep learning model. The server is further configured to generate outputs, based on a result predicted by the ML model, at least one of ABP point value/data point/sample for each input point value of PPG signal or multiple ABP point value outputs for multiple PPG point value inputs (MIMO model).

[0015] These together with other objects of the invention, along with the various features of novelty which characterize the invention, are pointed out with particularity in the disclosure. For a better understanding of the invention, its operating advantages and the specific objects attained by its uses, reference should be had to the accompanying drawings and descriptive matter in which there are illustrated preferred embodiments of the invention.

BRIEF DESCRIPTION OF DRAWINGS

[0016] The foregoing and other features of embodiments will become more apparent from the following detailed description of embodiments when read in conjunction with the accompanying drawings. In the drawings, like reference numerals refer to like elements. [0017] In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the invention. It is apparent, however, to one skilled in the art that the embodiments of the invention may be practiced without these specific details or with an equivalent arrangement. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the embodiments of the invention.

[0018] FIG. 1 illustrates network environment of a system for estimating Arterial blood pressure continuously and non-invasively using Photoplethysmography signals, according to one embodiment of the invention.

[0019] FIG. 2 illustrates a block diagram of a system for estimating Arterial blood pressure continuously and non-invasively using Photoplethysmography signals, accordance with an example embodiment of the present invention.

[0020] FIG. 3 illustrates a flow diagram of estimating arterial blood pressure using Stationary Wavelet Transform (SWT) Decomposition Method, according to one embodiment of the invention.

[0021] FIG. 4 illustrates estimating accurate Arterial blood pressure using Overlap-Windowing Technique, according to one embodiment of the invention.

[0022] FIG. 5 illustrates Arterial blood pressure (ABP) waveform estimation using Random Forest Regressor on MIMIC dataset using SWT decomposition method.

[0023] FIG. 6 illustrates ABP waveform estimation using AdaBoost Regressor on MIMIC dataset using SWT decomposition method.

[0024] FIG. 7 illustrates ABP waveform estimation using Decision Tree Regressor on MIMIC data using SWT decomposition method.

[0025] FIG. 8 illustrates Flowchart of ABP Waveform estimation using 15-

Features Method. [0026] FIG. 9 illustrates ABP estimation using Random Forest Regressor on UQVS Dataset by 15-Feature Method.

[0027] FIG. 10 illustrates ABP estimation using Decision Tree Regressor on UQVS Dataset by 15-Feature Method.

[0028] FIG. 11 illustrates ABP estimation using AdaBoost Regressor on UQVS Dataset by 15-Feature Method.

[0029] FIG. 12 illustrates flowchart of ABP Waveform estimation using Resampling as pre-processing and 15-Features Method.

[0030] FIG. 13, FIG. 14 and FIG. 15 illustrates smooth ABP waveform estimation using Random Forest, AdaBoost and Decision Tree Regressor respectively on smooth resampled input PPG from UQVS Dataset.

[0031] FIG. 16 illustrates Flow Chart of Training ABP Signal estimation using 45-Features.

[0032] FIG. 17 illustrates Flow Chart of Testing ABP Signal estimation using 45- Features.

[0033] FIG. 18 illustrates Flow Chart of Training ABP Signal estimation using 50-Features.

[0034] FIG. 19 illustrates Flow Chart of Testing ABP Signal estimation using 50- Features.

[0035] FIG. 20 illustrates the estimation of continuous ABP using 50 feature MIMO model.

DETAILED DESCRIPTION OF INVENTION

[0036] The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and / or detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practised and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.

[0037] Reference in this specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present disclosure. The appearance of the phrase “in an embodiment” in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Moreover, various features are described which may be exhibited by some embodiments and not by others. Similarly, various requirements are described which may be requirements for some embodiments but not for other embodiments.

[0038] Moreover, although the following description contains many specifics for the purposes of illustration, anyone skilled in the art will appreciate that many variations and/or alterations to said details are within the scope of the present disclosure. Similarly, although many of the features of the present disclosure are described in terms of each other, or in conjunction with each other, one skilled in the art will appreciate that many of these features can be provided independently of other features. Accordingly, this description of the present disclosure is set forth without any loss of generality to, and without imposing limitations upon the present disclosure.

[0039] As used in the application, the term ‘circuitry’ or ‘circuit’ refers to all of the following: (a) hardware-only circuit implementations (such as implementations in only analog and/or digital circuitry) and (b) to combinations of circuits and software (and/or firmware), such as (as applicable): (i) to a combination of processor(s) or (ii) to portions of processor(s)/software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and (c) to circuits, such as a microprocessor(s) or a portion of a microprocessor(s), that require software or firmware for operation, even if the software or firmware is not physically present and d) the interconnected brain and spinal cord circuits (both anatomical and functional circuits).

[0040] This definition of ‘circuitry’ applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term “circuitry” would also cover an implementation of merely a processor (or multiple processors) or portion of a processor and its (or their) accompanying software and/or firmware. The term “circuitry” would also cover, for example and if applicable to the particular claim element, a baseband integrated circuit or applications processor integrated circuit for a mobile phone or a similar integrated circuit in server, a cellular network device, or other network device.

[0041] The word “exemplary” is used herein to mean “serving as an example, instance, or illustration.” Any aspect described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects.

[0042] In this description, the term “application” may also include files having executable content, such as: object code, scripts, byte code, markup language files, and patches. In addition, an “application” referred to herein, may also include files that are not executable in nature, such as documents that may need to be opened or other data files that need to be accessed.

[0043] The term “content” may also include files having executable content, such as: object code, scripts, byte code, markup language files, and patches. In addition, “content” referred to herein, may also include files that are not executable in nature, such as documents that may need to be opened or other data files that need to be accessed. [0044] As used in this description, the terms “component,” “database,” “module,” “system,” and the like are intended to refer to a computer-related entity, either hardware, firmware, a combination of hardware and software, software, or software in execution. For example, a component may be, but is not limited to being, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and/or a computer. By way of illustration, both an application running on a computing device and the computing device may be a component. One or more components may reside within a process and/or thread of execution, and a component may be localized on one computer and/or distributed between two or more computers. In addition, these components may execute from various computer readable media having various data structures stored thereon. The components may communicate by way of local and/or remote processes such as in accordance with a signal having one or more data packets (e.g., data from one component interacting with another component in a local system, distributed system, and/or across a network such as the Internet with other systems by way of the signal).

[0045] In this description, the terms “communication device,” “wireless device,” “wireless telephone,” “wireless communication device,” and “wireless handset” are used interchangeably. With the advent of third generation (“3G”) wireless technology and four generation (“4G”), greater bandwidth availability has enabled more portable computing devices with a greater variety of wireless capabilities. Therefore, a portable computing device may include a cellular telephone, a pager, a PDA, or wearable device, a Smartphone, a navigation device, or a hand-held computer with a wireless connection or link.

[0046] FIG. 1 illustrates network environment of a system for estimating Arterial blood pressure continuously and non-invasively using Photoplethysmography signals, according to one embodiment of the invention. Referring to FIG. 1, the network environment may include a wearable device (101), an electronic device (103), and a server (105). The network environment may be communicatively coupled via network (107). [0047] In an example embodiment, the wearable device (101) may be communicatively coupled to the electronic device (103), and the server (105). The wearable device (103) may be worn on the wrist of a user. The wearable device (101) may be used to acquire Photoplethysmography (PPG) signal data and accelerometer data.

[0048] In an example embodiment, the electronic device (103) is communicatively coupled to the server (107) and wearable device (101). The electronic device (103) may be used to receive Photoplethysmography (PPG) signal data along with accelerometer data from the wearable device (101) and communicate the data collected at the wearable device (101) to the server (105).

[0049] In an example embodiment, the server (105) may be communicatively coupled to the electronic device (103) and the wearable device (101). The server (105) may be used to receive the bio-information data from at least one of the wearable device (101) or the electronic device (103), extract features from the bio -information data, normalize the extracted features, input the normalized features and corresponding Arterial Blood Pressure (ABP) signals to a Machine learning (ML) model, and output, based on a result predicted by the ML model, at least one of ABP point value/data point/sample for each input point value of PPG signal or multiple ABP point value outputs for multiple PPG point value inputs (MIMO model).

[0050] The network (107) may include the Internet or any other network capable of communicating data between devices. Suitable networks may include or interface with any one or more of, for instance, a local intranet, a PAN (Personal Area Network), a LAN (Local Area Network), a WAN (Wide Area Network), a MAN (Metropolitan Area Network), a Virtual Private Network (VPN), a Storage Area Network (SAN), a frame relay connection, an Advanced Intelligent Network (AIN) connection, a Synchronous Optical Network (SONET) connection, a digital Tl, T3, El or E3 line, Digital Data, Service (DDS) connection, DSL (Digital Subscriber Line) connection, an Ethernet connection, an ISDN (Integrated Services Digital Network) line, a dial-up port such as a V.90, V.34 or V.34bis analog modem connection, a cable modem, an ATM 13 (Asynchronous Transfer Mode) connection, or an FDDI (Fiber Distributed Data Interface) or CDDI (Copper Distributed Data Interface) connection. Furthermore, communications may also include links to any of a variety of wireless networks, including WAP (Wireless Application Protocol), GPRS (General Packet Radio Service), GSM (Global System for Mobile Communication), CDMA (Code Division Multiple Access) or TDMA (Time Division Multiple Access), cellular phone networks, GPS (Global Positioning System), CDPD (Cellular Digital Packet Data), RIM (Research in Motion, Limited) duplex paging network, Bluetooth radio, or an IEEE 802.11 -based radio frequency network. The network (107) can further include or interface with any one or more of an RS-232 serial connection, an IEEE- 1394 (Firewire) connection, a Fiber Channel connection, an IrDA (infrared) port, a SCSI (Small Computer Systems Interface) connection, a Universal Serial Bus (USB) connection or other wired or wireless, digital or analog interface or connection, mesh or Digi® networking.

[0051] FIG. 2 illustrates a block diagram of the server for estimating Arterial blood pressure continuously and non-invasively using Photoplethysmography signals, accordance with an example embodiment of the present invention. To execute activities associated with server for estimating Arterial blood pressure, the server has embedded a processor (201), a memory (203) and a communication interface (205).

[0052] In accordance with an embodiment, the processor (201) may be of any type of processor, such as 32-bit processors using a flat address space, such as a Hitachi SHI, an Intel 80386, an Intel 960, a Motorola 68020 (or other processors having similar or greater addressing space). Processor types other than these, as well as processors that may be developed in the future, are also suitable. The processor may include general processor, Digital Signal Processing (DSP) chip, an Application Specific Integrated Circuit (ASIC), Field Programmable Gate Arrays (FPGAs), AT89S52 microcontroller firmware or a combination thereof.

[0053] Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and anyone or more processors of any kind of digital computer. Generally, a processor receives instructions and data from a read only memory or a random-access memory or both. The essential elements of a computer are a processor for performing instructions and one or more memory devices for storing instructions and data. Generally, a computer also includes, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto optical disks, or optical disks. However, a computer need not have such devices. Moreover, a computer can be embedded in another device, e.g., a mobile telephone, a Personal Digital Assistant (PDA), a mobile audio player, a GPS receiver, to name just a few. Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media, and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto optical disks; and CD ROM and DVD-ROM disks. The memory may be a non-transitory medium such as a ROM, RAM, flash memory, etc. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

[0054] The processes and logic flows described in the specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit).

[0055] In accordance with an embodiment, the memory (203) includes both dynamic memory (e.g., RAM, magnetic disk, writable optical disk, etc.) and static memory (e.g., ROM, CD-ROM, etc.) for storing executable instructions that when executed perform the inventive steps described herein for continuous estimation of arterial blood pressure.

[0056] In accordance with an embodiment, network includes one or more networks such as a data network, a wireless network, a telephony network, or any combination thereof. It is contemplated that the data network may be any Local Area Network (LAN), Metropolitan Area Network (MAN), Wide Area Network (WAN), a public data network (e.g., the Internet), short range wireless network, or any other suitable packet- switched network, such as a commercially owned, proprietary packet-switched network, e.g., a proprietary cable or fiber-optic network, and the like, or any combination thereof. In addition, the wireless network may be, for example, a cellular network and may employ various technologies including enhanced data rates for global evolution (EDGE), general packet radio service (GPRS), global system for mobile communications (GSM), Internet protocol Multimedia Subsystem (IMS), Universal Mobile Telecommunications System (UMTS), etc., as well as any other suitable wireless medium, e.g., worldwide interoperability for microwave access (WiMAX), Long Term Evolution (LTE) networks, Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Wireless Fidelity (Wi-Fi), Wireless LAN (WLAN), Bluetooth®, Internet Protocol (IP) data casting, ZigBee satellite, mobile ad- hoc network (MANET), and the like, or any combination thereof.

[0057] Although the present specification describes components and functions that may be implemented in particular embodiments with reference to particular standards and protocols, the invention is not limited to such standards and protocols. For example, The ZigBee or ZigBee/IEEE 802.15.4 protocol is a specification created for wireless networking. It includes hardware and software standard design for WSN (Wireless Sensor Network) requiring high reliability, low cost, low power, scalability and low data rate. Accordingly, replacement standards and protocols having the same or similar functions as those disclosed herein are considered equivalents thereof.

[0058] In an example embodiment, the communication interface (205) may include but not limited to traditional interfaces which include no intelligence in the interface, only physical connection which could include changes in voltage levels and transformation from balanced to unbalanced signal, communication protocols which may use pre-programmed modules etc. Further, the communication interface may include Modem interfaces, which have a high level of intelligence in the interface where a high level of intelligence in the interface is employed to execute operations.

[0059] Further, at least one activity is executed by the server comprising the embedded processor (201), memory (203) and the communication interface (205), based on control and configuration of components associated with the system. Embodiments of the invention describes estimating Arterial Blood Pressure continuously and non- invasively using Photoplethysmography signals. The estimation of Arterial blood pressure is done using different methods with modification examples are explained below in different embodiments.

Embodiment 1:

[0060] The embodiment 1 describes estimation of Arterial Blood Pressure using Stationary Wavelet Transform (SWT) Decomposition Method. Referring to FIG. 3 in an example embodiment at step 301, Photoplethysmography (PPG) signal data along with accelerometer data is collected from a wearable device (101) worn on the wrist of a user. In some example embodiments, the electronic device (103) communicates the collected data to the server (105).

[0061] At step 303, the PPG data stream may be inputted by overlap-windowing technique (illustrated in FIG. 4) for at least 60 seconds i.e., as illustrated in Fig. 4, for initial Arterial blood pressure (ABP) estimation, a 60 seconds (one-minute) data stream is required post which a one second input data stream is provided by the overlapwindowing technique. In the overlap-windowing technique, the previous 59 seconds input data stream along with the newly arrived 1 second data stream total 60 second data stream is taken as input for estimation is shown in FIG. 4.

[0062] At step 305, denoising of the PPG signal is executed by pre-processing of the input data stream of PPG signal. In denoising, the induced noise due to motion artifacts and ambient light is removed by Discrete Wavelet Transform. The signal denoising is done by: a) The signal is decomposed to level- 1 by Daubechies mother wavelet in detailed coefficients corresponding to short- scale, high-frequency elements of the signal, and approximation coefficients corresponding to large-scale, low-frequency elements of the signal, b) A simple nonlinear technique, called thresholding, is applied on one wavelet coefficient at a time which is obtained in the decomposition step. Specifically, hard thresholding is used in which elements whose absolute value is less than a threshold value is set to zero. The threshold value is computed by the Mean Absolute Deviation method, c) The signal is reconstructed using Inverse Discrete Wavelet Transform.

[0063] At step 307, the denoised PPG signal is further used to extract features. The feature extraction process is done in a 3-level Stationary Wavelet Transform (SWT), in 3-level Stationary Wavelet Transform (SWT), the first extraction stage is time series decomposition using wavelet transform, a signal processing technique used to analyse the PPG signal in both the temporal and spectral domain. In this method, SWT is used to derive additional features with length equal to that of the PPG segment i.e., 60 second signal. The segment of PPG signal is decomposed up to 3-level using Daubechies(dblO) wavelets.

[0064] The SWT, which consists of a cascade of low and high pass filters. Ai and Di correspond to the approximate and detailed coefficients at level i. This cascade of filters allows for a multi-tiered coefficient representation of the PPG signal to be extracted. In some example embodiments, the feature extraction process may be carried out using Principal Component Analysis (PCA).

[0065] At step 309, the extracted features are normalized. The features include:

(i) 3-level Stationary Wavelet Decomposition is used to decompose the pre- processed PPG signal.

(ii) Total 6 features out of which 3 features are Detail coefficient at each level and 3 features are Approximate Coefficient. [0066] The PCA is applied on pre-processed PPG signal in order to extract one principal component as a 7th feature. The normalization is a scaling technique in which feature values are shifted and rescaled, so that they end up ranging between 0 and 1. It is also known as Min-Max scaling. The formula for normalization is as given below:

Here, Xmax and Xmin are the maximum and the minimum values of the feature respectively.

[0067] In some example embodiments, When the value of X is the minimum value in the column, the numerator will be 0, and hence X’ is 0. On the other hand, when the value of X is the maximum value in the column, the numerator is equal to the denominator and thus the value of X’ is 1. Further, If the value of X is between the minimum and the maximum value, then the value of X’ is between 0 and 1.

[0068] In an example embodiment, at step 311, the normalized features and corresponding ABP signal are used to train the machine learning model.

[0069] In an example embodiment, at step 313, the machine learning model predicts the ABP waveform for each and every input point value of PPG signal hence called it as a point-by-point estimation or construction. There are three machine learning algorithms used to construct the ABP Waveform: a) The Random Forest is an ensemble model, which trains multiple decision trees and averages the outputs of each of these trees when making a prediction. The ABP waveform estimation using Random Forest Regressor on MIMIC data is shown in FIG. 5. b) The AdaBoost is an ensemble technique that attempts to create a strong prediction from a number of weak predictions. The ABP waveform estimation using AdaBoost Regressor on MIMIC data is shown in FIG. 6. c) The Decision Tree is a decision-making tool that uses a flowchart-like tree structure or is a model of decisions. The ABP waveform estimation using Decision Tree on MIMIC data is shown in FIG. 7. Embodiment 2:

[0070] The Embodiment 2 describes process of estimating Arterial blood pressure using 15-Features Method. Referring to FIG. 8, for estimating Arterial blood pressure step 801, step 803, and step 805 is same as step 301, step 305, and step 305 respectively of embodiment 1.

[0071] At step 807, the Feature Extraction stage extracts the 15-features from the denoised PPG signal. The feature extraction may be done using: a. Derivative, 3-level Stationary Wavelet Transform (SWT), Empirical Wavelet Transform (EWT), Variational Mode Decomposition (VMD), Mel Frequency Cepstral Coefficient (MFCC), and Fractional Differentiation (FD):

[0072] The stationary wavelet transform (SWT) is used to derive additional features with length equal to that of the PPG segment i.e., 60 second signal. The segment of PPG signal is decomposed up to 3-level using Daubechies (dblO) wavelets. Figure shows a filter representation of the SWT, which consists of a cascade of low and high pass filters. Ai and Di and corresponds to the approximate and detailed coefficients at level i. This cascade of filters allows for a multi-tiered coefficient representation of the PPG signal to be extracted.

[0073] The Empirical Wavelet Transform was proposed by Gilles. This method builds an adaptive wavelet filter bank based on the spectrum information contained in the PPG signal. The EWT process contains two important aspects that includes segmenting the spectrum of the signal and constructing the empirical wavelets and decomposing the signal into different components. Here 4 components are extracted as feature.

[0074] The Variational Mode Decomposition (VMD) is a fully intrinsic and non- recursive method to process non-stationary signals. The VMD approach includes mainly three steps:

1. Computation of one-sided frequency spectrum pertaining to the analytical representation of the input signal by means of the Hilbert transform. 2. Each mode is multiplied with an exponential function in order to shift its frequency spectrum to the baseband where the frequency of exponential function is fixed according to the centre frequency of the mode; and

3. The bandwidth of the mode is estimated using the Gaussian smoothness applied to the demodulated signal.

[0075] The pre-processed PPG signal is decomposed into five sub-signals also known as 5-modes. The value of input parameters is set as follows: alpha=2000, tau=0, tol=10^A-7, and dc=0. The 5-modes represents the Cardiac component and Respiratory component.

[0076] At step 809, normalization is executed from the extracted features. Normalisation is a scaling technique in which feature values are shifted and rescaled so that they end up ranging between 0 and 1.

[0077] At step 811, The machine learning model predicts the ABP waveform for each and every input point value of PPG signal hence called it as a point-by-point estimation or estimation. There are three machine learning algorithms are used to construct the ABP waveform: a. The Random Forest is an ensemble model, which trains multiple decision trees and averages the outputs of each of these trees when making a prediction. The ABP waveform estimation using Random Forest Regressor on UQVS dataset is show in the FIG. 9. b. The AdaBoost is an ensemble technique that attempts to create a strong prediction from a number of weak predictions. The ABP waveform estimation using AdaBoost Regressor on UQVS dataset is shown in FIG.10. c. The Decision Tree is a decision-making tool that uses a flowchart-like tree structure or is a model of decisions. The ABP waveform estimation using AdaBoost Regressor on UQVS dataset is shown in FIG. 11.

Embodiment 3: [0078] The embodiment 3 describes estimation of Arterial Blood Pressure using Resampling as pre-processing and 15-Features Method. Referring to FIG. 12, the step 1201, step 1203 and step 1205 are same as step 301, step 303, and step 305 respectively. [0079] In an example embodiment at step 1207, resampling of the PPG signal that is obtained after denoising is executed. Resampling is the process which modifies the sample rate of the signal. The need for resampling in ABP estimation is that the input training sample is from a MIMIC database which has 125Hz sample rate and it is not necessary that the input test PPG signal has the same sample rate. For instance, the test PPG signal is from UQVS dataset having 100Hz sample rate.

[0080] The steps for resampling are as follows: when P > Q, here, P represents Train Sample Rate, and Q represents Test Sample Rate.

■ determine correction factor(cf) = P-Q

■ determine Interpolation value = Q/cf

■ Add a sample at a point interval given by Interpolation value

■ The added sample value is average of adjacent sample.

[0081] when. P < Q: the method includes:

■ determine correction factor(cf) = Q-P

■ determine Decimation Value = Q/cf

■ Remove a sample at a point interval given by Decimation value.

[0082] Savitzky_Golay filter is used for smoothing of resampled signals. The Savitzky-Golay filter is a digital filter that can be applied to a set of digital data points for the purpose of smoothing the data, that is, to increase the precision of the data without distorting the signal tendency. The length of the filter window is set to 11 for smoothing of resampled signal, which represents the number of coefficients and 4th order polynomial is used to fit the samples.

[0083] The feature extraction, normalization and machine learning process are the same as explained in embodiment 2. [0084] A Savitzky_Golay filter is used for smoothing of ABP waveform. The length of the filter window is set to 11 and 4th order polynomial is used to fit the samples. The Smooth ABP waveform estimation using Random Forest, AdaBoost and Decision Tree Regressor respectively on smooth resampled input PPG from UQVS Dataset is shown is illustrated in FIG. 13, FIG.14 and FIG.15.

Embodiment 4

[0085] The embodiment 4 describes estimation of Arterial Blood Pressure using 45-features method. FIG. 16 illustrates flow Chart of Training ABP Signal estimation using 45-Features.

[0086] In an example embodiment, at step 1601, the Photoplethysmography (PPG) signal data along with accelerometer data is collected from the wrist-worn sensor (101) connected with electronic device (103) via network. The collected data is communicated to the server.

[0087] In an example embodiment, at step 1603 pre-processing of the input data stream of PPG segment is done by a method called denoising. In denoising the induced noise due to motion artifacts and ambient light is removed by Discrete Wavelet Transform. The steps of signal denoising is as follows:

1. The signal is decomposed to level- 1 by Daubechies mother wavelet in detailed coefficients corresponding to short- scale, high-frequency elements of the signal, and approximation coefficients corresponding to large-scale, low-frequency elements of the signal.

2. A simple nonlinear technique, called thresholding, is applied on one wavelet coefficient at a time which is obtained in the decomposition step. Specifically, hard thresholding is used as thresholding is used in which elements whose absolute value is less than a threshold value are set to zero. The threshold value is computed by the Mean Absolute Deviation method.

3. The signal is reconstructed using Inverse Discrete Wavelet transform. [0088] In an example embodiment, at step 1605, for initial ABP estimation, PPG segment having one trough is taken as input for estimation.

[0089] In an example embodiment, at step 1607 the time domain and frequency domain features are extracted from the input PPG Segment. The total 45 features are given below:

1. At 10%: Systolic Width (SW10), Diastolic Width (DW10), SW10+DW10, DW10/SW10

2. At 25%: SW25, DW25, SW25+DW25, DW25/SW25

3. At 33%: SW33, DW33, SW33+DW33, DW33/SW33

4. At 50%: SW50, DW50, SW50+DW50, DW50/SW50

5. At 66%: SW66, DW66, SW66+DW66, DW10/SW66

6. At 75%: SW75, DW75, SW75+DW75, DW75/SW75

7. Maxima Index

8. Minima Index

9. Derivative- 1

10. Derivative- 1 Maxima Index

11. Derivative- 1 Minima Index

12. Derivative-2

13. Derivative-2 Maxima Index

14. Derivative-2 Minima Index

15. fbase - Fundamental Component Frequency

16. |sbase| - Fundamental Component Magnitude

17. f2 - Second-harmonic frequency

18. |s2| - Second-harmonic magnitude

19. f3 - Third harmonic frequency

20. |s31 - Third harmonic frequency

[0090] In an example embodiment, at step 1609, the input PPG and ABP signal to the training model is extracted from the MIMIC dataset.

[0091] In an example embodiment, at step 1611, Multiple Input Multiple Output (MIMO) Regression Machine Learning Model construct the ABP waveform from extracted features. Generally, a regression analysis involves predicting a single numeric value. Some tasks require predicting more than one numeric value. These tasks are referred to as multiple-output regression.

[0092] FIG. 17 illustrates flow Chart of testing ABP Signal estimation using 45- Features. Referring to FIG. 17, at step 1707 and step 1709 the Resampling of ABP waveform depends upon the input PPG segment points. MIMO regression can be implemented by a neural network by specifying the number of target variables as the number of nodes in the output layer. Each segment has 45 input features and 45 outputs, hence, the network requires an input that takes 45 input in the first hidden layer and 45 nodes in the output layer. The Rectified Linear Unit (ReLU) activation function is used in the hidden layer. The model fit using Mean Absolute Error (MAE) loss and the Adam optimizer.

[0093] In an example embodiment, at step 1711, the Interpolation and Decimation process for ABP signal construction is done. The Interpolation and Decimation process is as follows:

1. If P > Q: (here P - Input PPG Segment, and Q - Output Sample - 45) correction factor(cf) = P-Q;

Interpolation value = Q/cf;

Add a sample at a point interval given by Interpolation value;

The added sample value is average of adjacent sample;

2. If P < Q: correction factor(cf) = Q-P;

Decimation Value = Q/cf;

Remove a sample at a point interval given by Decimation value;

Embodiment 5

[0094] The embodiment 5 describes estimation of Arterial Blood Pressure using 50-features method. FIG. 18 illustrates, flow Chart of Training ABP Signal estimation using 50-Features. [0095] In an example embodiment, at step 1801, The Photoplethysmography (PPG) signal data along with accelerometer data is collected from the wrist-worn sensor (101) connected with the electronic device (103) via network (107) which is communicated to the server (105).

[0096] In an example embodiment, at step 1803, pre-processing of the input data stream of PPG segment is done by a method called denoising, in denoising the induced noise due to motion artifacts and ambient light is removed by Discrete Wavelet Transform. The steps of signal denoising is as follows:

2. A simple nonlinear technique, called thresholding, is applied on one wavelet coefficient at a time which is obtained in the decomposition step. Specifically, hard thresholding is used as thresholding is used in which elements whose absolute value is less than a threshold value is set to zero. The threshold value is computed by the Mean Absolute Deviation method.

3. The signal is reconstructed using Inverse Discrete Wavelet transform.

[0097] In an example embodiment, at step 1805, for initial ABP estimation, PPG segment having one trough is taken as input for estimation.

[0098] In an example embodiment at step 1807, the total 50 features are extracted from the PPG segment. The time domain and frequency domain features are the same as in embodiment 4 and new five statistical features added in the feature vector which are extracted from the input PPG Segment.

The newly added statistical features are as follows:

1. Mean

2. Maximum

3. Kurtosis

4. Median 5. Standard Deviation.

[0099] In an example embodiment, at step 1809 and step 1811, Multiple Input Multiple Output (MIMO) Regression Machine Learning Model construct the ABP waveform from extracted features. MIMO regression can be implemented by a neural network by specifying the number of target variables as the number of nodes in the output layer. Each segment has 50 input features and 50 outputs; hence, the network requires an input that takes 50 inputs in the first hidden layer and 50 nodes in the output layer. The Rectified Linear Unit (ReLU) activation function is used in the hidden layer. The model fit using Mean Absolute Error (MAE) loss and the Adam optimizer.

[00100] FIG. 19 illustrates the flow of the Testing MIMO Regression Machine Learning Model by using 50 features. Referring to FIG. 19, at step 1911, Interpolation and Decimation process for ABP signal construction is done. The Resampling of ABP waveform depends upon the input PPG segment points.

[00101] The Interpolation and Decimation process is as follows:

1. If P > Q (Here, P - Input PPG Segment, and Q - Output Sample - 50): correction factor(cf) = P-Q;

Interpolation value = Q/cf;

Add a sample at a point interval given by Interpolation value;

The added sample value is average of adjacent sample;

2. P < Q: correction factor(cf) = Q-P;

Decimation Value = Q/cf;

Remove a sample at a point interval given by Decimation value;

[00102] The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the spirit and scope of the embodiments as described herein.

Claims

WE CLAIM

1. An information processing apparatus for continuous estimation of arterial blood pressure, comprising: a wearable device (101) configured to acquire bio-information data, wherein the bio-information data includes Photoplethysmography (PPG) signal data and accelerometer data; an electronic device (103) configured to receive the bio-information data; a server communicatively coupled with the wearable device (101) and the electronic device (103), wherein the server is configured to: receive the bio-information data from at least one of the wearable device or the electronic device; extract features from the bio-information data; normalize the extracted features; input the normalized features and corresponding Arterial Blood Pressure (ABP) signals to one of a Machine learning (ML) model or Deep learning model; and output, based on a result predicted by the ML model or Deep learning model, at least one of ABP point value/data point/sample for each input point value of PPG signal or multiple ABP point value outputs for multiple PPG point value inputs (MIMO model).

2. The information processing apparatus (100) as claimed in claim 1, wherein the server (105) is further configured to execute smooth resampling of the received bio-information data.

3. The information processing apparatus (100) as claimed in claim 1, wherein the server (105) is further configured to extract PPG signal segment from the acquired bio-information data.

26

4. The information processing apparatus (100) as claimed in claim 1, wherein the server (105) is further configured to execute tracing and extracting ABP signal value at the same time instant as in input PPG signal segment.

5. The information processing apparatus (100) as claimed in claim 1, wherein the server (105) is further configured to execute Interpolation and Decimation of the output ABP waveform from the Machine learning (ML) model output.

6. The information processing apparatus (100) as claimed in claim 1, wherein the bio information is acquired based on overlap-windowing technique.

7. The information processing apparatus (100) as claimed in claim 1, wherein the denoising of acquired bio information include the steps of: decomposition of the acquired bio-information data to level- 1 by Daubechies mother wavelet in detailed coefficients corresponding to short-scale, high-frequency elements of the signal, and approximation coefficients corresponding to large-scale, low-frequency elements of the signal; thresholding on one wavelet coefficient at a time which is obtained in the decomposition step; reconstructing the bio-information data using Inverse Discrete Wavelet transform.

8. The information processing apparatus as claimed in 1, wherein the server is configured to reconstruct the ABP signal based on the ABP data point value outputs for multiple PPG point value inputs (MIMO model).

9. The information processing apparatus as claimed in 9, wherein the SWT decomposition method includes at least one of stationary wavelet transform or Principal component analysis.

10. The information processing apparatus as claimed in 9, wherein in the 15- features method includes at least one of Derivative, 3-level Stationary Wavelet Transform, Empirical Wavelet Transform (EWT), Variational Mode Decomposition, Mel Frequency Cepstral Coefficient (MFCC), or Fractional Differentiation (FD).

11. The information processing apparatus as claimed in 1, wherein the ME model includes at least one of a Random Forest, AdaBoost, or Decision Tree.

12. An information processing method for continuous estimation of arterial blood pressure, the method comprising: receiving the bio -information data from at least one of the wearable device or the electronic device; extracting features from the bio -information data; normalizing the extracted features; inputting the normalized features and corresponding Arterial Blood Pressure (ABP) signals to a Machine learning (ML) model or Deep learning model; and outputting, based on result predicted by the ML model, at least one of ABP ABP point value/data point/sample for each input point value of PPG signal or multiple ABP point value outputs for multiple PPG point value inputs (MIMO model).