Non-invasive continuous blood pressure estimation device and method thereof

Abstract

The present invention relates to a non-invasive continuous blood pressure estimation device and a method thereof. The non-invasive continuous blood pressure estimation device according to an embodiment of the present invention comprises: a sensor unit including a photoplethysmography (PPG) sensor for detecting a PPG signal of a subject; a data collection unit for collecting the PPG signal from the sensor unit and storing the PPG signal together with time information; a feature extraction unit for extracting one or more biosignal features from among a systolic timespan (ST), a systolic timespan variation based on intersecting tangent point variation (STVIT), and a systolic timespan variation based on maximum slope point variation (STVMD), which are derived from a specific point of the PPG signal; and a blood pressure estimation unit for estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) by using the biosignal features extracted by the feature extraction unit.

Description

Non-invasive continuous blood pressure estimation device and method

[0001] The present invention relates to a biosignal-based non-invasive blood pressure measurement technology, and more specifically, to a non-invasive continuous blood pressure estimation device and method that estimates blood pressure using PPG signals.

[0002] Conventional blood pressure measurement methods typically involved intermittent measurement using a cuff. However, this method cannot reflect continuous blood pressure changes in real time, and because it applies pressure to the user's body with every measurement, it is difficult to use for extended periods and causes discomfort.

[0003] Accordingly, research aimed at continuously estimating blood pressure using biosignals in a non-invasive manner has been actively conducted. Among these, photoplethysmography (PPG) signals are attracting attention as signals with high potential for non-pressurized continuous blood pressure estimation, as they can be conveniently measured from the fingers, ears, or wrists using only a sensor.

[0004] Conventional non-invasive blood pressure estimation studies have primarily used methods to estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP) by simultaneously utilizing electrocardiograms (ECGs) and photoplethysmography (PPGs).

[0005] However, this ECG-PPG combined method had a problem in that the accuracy of blood pressure estimation was reduced due to time delay or time discrepancy between the two signals. The R-peak of the ECG signal represents the electrical contractile activity of the heart, and the waveform of the PPG signal reflects changes in blood flow in peripheral blood vessels, but depending on the sensor attachment location, the characteristics of the measurement equipment, and the physical condition of the subject, the two signals may not be collected simultaneously and a delay may occur.

[0006] According to an embodiment of the present invention, a non-invasive continuous blood pressure estimation device and a method are provided that improve the accuracy and reliability of non-invasive continuous blood pressure estimation using a photoplethysmography (PPG) signal.

[0007] A non-invasive continuous blood pressure estimation device according to an embodiment of the present invention comprises: a sensor unit including a PPG sensor that detects a photoplethysmography (PPG) signal of a subject; a data collection unit that collects the PPG signal from the sensor unit and stores it along with time information; a feature extraction unit that extracts one or more biosignal features among a systolic timespan (ST), a systolic timepan based on cross-tangential variation (STVIT), and a systolic timepan based on maximum slope variation (STVMD) derived from a specific point of the PPG signal; and a blood pressure estimation unit that estimates systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the biosignal features extracted from the feature extraction unit.

[0008] A non-invasive continuous blood pressure estimation method according to an embodiment of the present invention comprises the steps of detecting a photoplethysmography (PPG) signal of a subject, collecting the PPG signal and storing it together with time information, and systolic timespan (ST) derived from a specific point of the PPG signal, and systolic timespan variation based on cross-tangential variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMDThe method includes the step of extracting one or more biosignal features and the step of estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the extracted biosignal features.

[0009] According to an embodiment, the data collection unit can collect the PPG signal at the same sampling frequency and align the collected PPG signal based on the time axis.

[0010] According to an embodiment, the feature extraction unit can extract one or more of a peak (P), a valley (V), an intersecting tangent (IT), a maximal derivative (MD), photoplethysmography intensity (PI), branch width (BW), systolic branch width (SBW), and diastolic branch width (DBW) from the PPG signal.

[0011] According to an embodiment, the blood pressure estimation unit can estimate the systolic blood pressure and the diastolic blood pressure through a blood pressure estimation model including a Temporal Convolutional Network (TCN) using the extracted biosignal feature value as input.

[0012] According to an embodiment, the blood pressure estimation model includes a plurality of temporal blocks, each temporal block includes two convolutional layers, and the dilation factor may increase exponentially depending on the depth of the convolutional layers.

[0013] According to an example, the blood pressure estimation model may prevent overfitting by configuring the input sequence with 16 consecutive cardiac cycles, including three hidden layers, and applying dropout.

[0014] According to an embodiment, the blood pressure estimation model can be trained by using the Mean Squared Error (MSE) as the loss function and applying the Rectified Adam (RAdam) algorithm as the optimizer.

[0015] According to an embodiment, the blood pressure estimation unit can calculate the mean arterial pressure (MAP) using systolic blood pressure and diastolic blood pressure values ​​calculated through a learned blood pressure estimation model.

[0016] According to an embodiment, the blood pressure estimation model can be trained by dividing the training dataset into training, validation, and test sets in a 2:1:1 ratio, and can save the model at the point where validation loss is minimized.

[0017] According to an embodiment, the blood pressure estimation unit is the ST VIT and the above ST VMD A set of PPG signal-based features including can be learned.

[0018] According to an embodiment of the present invention, systolic blood pressure (SBP) and diastolic blood pressure (DBP) can be continuously estimated using only photoplethysmography (PPG) signals.

[0019] In addition, a non-invasive continuous blood pressure estimation device according to one embodiment of the present invention includes a cross-tangential point variation-based systolic interval variation (ST VIT) and systolic interval variation based on maximum slope point variation (ST VMD By adding new morphological features such as ), the accuracy of blood pressure estimation can be further improved compared to existing blood pressure estimation models that use only the systolic interval time (ST).

[0020] In addition, since it uses only PPG signals, it can be implemented in the form of a simple sensor or wearable device attached to the finger, wrist, or ear, and allows for easy non-invasive blood pressure monitoring in a home environment or medical institution.

[0021] FIG. 1 is a block diagram showing a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0022] FIG. 2 is a diagram showing features extracted from a PPG signal in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0023] FIG. 3 is a diagram showing the structure of a time-series convolutional neural network used in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0024] FIG. 4 is a diagram showing the overall operation flow of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0025] Figure 5 is a diagram showing the number of feature sets by age group of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0026] FIG. 6 is a histogram showing the blood pressure distribution according to age percentiles of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0027] FIG. 7 is a scatter plot showing the actual blood pressure (BP) and the blood pressure estimated by a blood pressure estimation model in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0028] FIG. 8 shows a Bland-Altman plot indicating the degree of agreement between actual blood pressure (BP) and blood pressure estimated by a blood pressure estimation model in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0029] Figure 9 is a figure showing the results of analyzing the importance and influence of each feature used for blood pressure estimation in a blood pressure estimation model according to an embodiment of the present invention as SHAP values ​​(Shapley Additive Explanations).

[0030] FIG. 10 is a flowchart of a non-invasive continuous blood pressure estimation method according to an embodiment of the present invention.

[0031] FIG. 11 illustrates a non-invasive continuous blood pressure estimation device according to one embodiment of the present invention.

[0032] Hereinafter, the present invention will be described in detail with reference to the attached drawings. However, this is merely illustrative and the present invention is not limited to the specific embodiments described illustratively.

[0033] The terms used herein are for describing the embodiments and are not intended to limit the invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. As used herein, "comprises" and / or "made of" do not exclude the presence or addition of one or more other components, steps, actions, and / or elements to the mentioned components, steps, actions, and / or elements.

[0034] In addition, terms such as "~part," "~unit," "~block," "~part," and "~module" may refer to a unit that processes at least one function or operation. For example, the above terms may refer to at least one piece of hardware such as an FPGA (Field Programmable Gate Array) or ASIC (Application Specific Integrated Circuit), at least one piece of software stored in memory, or at least one process processed by a processor.

[0035] Unless otherwise defined, all terms used in this specification (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0036]

[0037] FIG. 1 is a block diagram showing a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0038] Referring to FIG. 1, a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention comprises a sensor unit (100) including a PPG sensor (120) that detects a photoplethysmography (PPG) signal of a subject, a data collection unit (220) that collects the PPG signal from the sensor unit (100) and stores it together with time information, a feature extraction unit (240) that extracts one or more biosignal features among Systolic Timespan (ST), Systolic Timespan based on cross-tangential variation (STVIT), and Systolic Timespan based on maximum slope variation (STVMD) derived from a specific point of the PPG signal, and a blood pressure estimation unit (260) that estimates systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the biosignal features extracted from the feature extraction unit (240).

[0039] A sensor unit (100) according to an embodiment of the present invention is for detecting a subject's biological signal and includes a PPG sensor (120) that detects a PPG signal.

[0040] The above PPG sensor (120) is attached to or in contact with the subject's body (e.g., finger, earlobe, or wrist) to detect an optical signal corresponding to a change in blood flow in a peripheral blood vessel area. For example, it includes a light source (e.g., LED) and a photodetector, and utilizes the principle that the intensity of reflected light or transmitted light changes according to fluctuations in blood flow when light emitted from the light source passes through skin tissue.

[0041] In this case, the signal detected by the PPG sensor (120) includes waveform shapes of the systole and diastole sections and consists of a continuous waveform including the rising part, peak, falling part, and intersecting tangent of the pulse wave.

[0042] In the embodiment of the present invention, since it includes only a PPG sensor (120) and does not include an ECG sensor, the problem of synchronization between the two signals is fundamentally eliminated, and blood pressure is estimated based only on the morphological change of the PPG signal.

[0043] Accordingly, it is important for the sensor unit (100) according to the embodiment of the present invention to acquire the PPG signal stably, with low noise, and continuously, and the following data collection unit (220), feature extraction unit (240), and blood pressure estimation unit (260) all operate based on the PPG signal.

[0044]

[0045] The data collection unit (220) according to an embodiment of the present invention collects a PPG signal from the sensor unit (100) and stores it together with time information.

[0046] The data collection unit (220) collects the signal input from the PPG sensor (120) at a predetermined sampling frequency and assigns time-axis information to the collected PPG signal so that the subsequent feature extraction time can be accurately matched.

[0047] According to one embodiment, the data collection unit (220) includes a function to perform time axis alignment of the signal. Time axis alignment is performed based on the start and end times of the PPG signal and is aligned based on the time information (Time Stamp) assigned to each signal.

[0048] For example, in one embodiment of the present invention, 16 consecutive PPG segments constitute one input sequence, so the data collection unit (220) can sequentially store data based on the time axis so that these sequence units are not interrupted. In this case, the drop intervals of the PPG signal, sensor dropouts, abnormal values, etc., can also be filtered based on time information, thereby increasing the stability of subsequent steps.

[0049] According to one embodiment, the data collection unit (220) performs the function of storing the collected signal along with time information. The data stored at this time includes the voltage value, light intensity value, and time stamp at each sampling point, and subsequently, the feature extraction unit (240) uses the data to search for a reference point of the biosignal.

[0050] According to one embodiment, the data collection unit (220) may store a signal in an internal memory or an external storage device and may be configured in parallel with a buffer memory for real-time processing. Additionally, the data collection unit (220) may communicate with the sensor unit (100) via a wired interface (e.g., USB, RS-232) or a wireless interface (e.g., BLE, Wi-Fi).

[0051] In this way, the data collection unit (220) according to the embodiment of the present invention can improve the stability of the training data in the subsequent step by performing time-axis alignment of the PPG signal.

[0052]

[0053] A feature extraction unit (240) according to an embodiment of the present invention includes a systolic timespan (ST; Systolic Timespan) derived from a specific point of the PPG signal, and a systolic timespan variation based on cross-tangential variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD It is a part that extracts one or more biological signal features from ).

[0054] The above feature extraction unit (240) receives a PPG signal that is aligned and stored along the time axis by the data collection unit (220) as input, detects major morphological reference points within the waveform of the PPG signal, and quantifies the temporal or amplitude relationship between these reference points to calculate a feature value.

[0055] The feature extraction unit (240) according to the embodiment identifies significant points on the waveform, such as the starting point, valley (V), peak (P), intersecting tangent (IT), and maximal derivative (MD) within the waveform of the PPG signal, and interprets the intervals between them as systolic intervals to reveal the attributes of systolic blood flow change.

[0056] The feature extraction unit (240) according to the embodiment is ST VIT and ST VMD Including additionally, ST VIT represents the variation in the systolic period based on the intersection tangent point (IT), and ST VMD represents the variation in the systolic period based on the MD point where the rate of change is maximum in the rising part of the waveform.

[0057] For example, even if ST is constant in consecutive PPG heart rate cycles, there may be slight positional changes at the IT time or MD time, and the feature extraction unit (240) can quantify these slight fluctuations as time values ​​and output them as a single feature value. In this way, changes in arterial elasticity or peripheral resistance can be reflected solely by the time dynamics within the PPG waveform, so that the amount of information necessary for blood pressure estimation can be secured even in an environment without an ECG.

[0058] The feature extraction unit (240) according to the embodiment may extract only ST as needed, and ST and ST VIT You can also extract together, and ST, ST VIT , ST VMD All of them can also be extracted and organized into a multidimensional feature set.

[0059]

[0060] As described above, the feature extraction unit (240) according to an embodiment of the present invention can extract one or more of a peak (P), a valley (V), an intersecting tangent (IT), a maximal derivative (MD), photoplethysmography intensity (PI), branch width (BW), systolic branch width (SBW), and diastolic branch width (DBW) from the PPG signal.

[0061] As illustrated in FIG. 2, the feature extraction unit (240) of the present invention detects a morphological reference point of a biosignal from a PPG signal, extracts various biosignal features using this, and constructs input data for blood pressure estimation using these features.

[0062] According to the embodiment, the feature extraction unit (240) first receives a PPG signal, removes noise from the signal, and normalizes the baseline. At this time, the signal collected from the PPG sensor (120) attached to the subject's finger or ear is aligned with respect to the time axis, and the feature extraction unit (240) analyzes the rising and falling parts of the waveform based on this to detect key feature points.

[0063] The peak point (P) represents the highest point of the waveform and indicates the point of peak systole, while the trough point (V) represents the lowest point of the waveform and reflects the blood flow state at the end of diastole. The intersection tangent point (IT) represents the point where the tangent to the baseline intersects at a constant angle during the rising portion of the waveform and reflects the elasticity of the arterial wall and changes in blood flow velocity. The maximum slope point (MD) is the point where the rate of change is maximum during the rising portion of the waveform and is used as a key factor representing the acceleration of blood flow.

[0064] Photoplethysmography intensity (PI) is the intensity of an optical signal calculated based on the maximum amplitude of the waveform, quantifying changes in blood flow volume due to changes in blood light absorption. The bifurcation width (BW) refers to the width between the inflection point (V) and the cross-tangential point (IT) in the PPG signal, and the systolic bifurcation width (SBW) is defined as the width between the inflection point (V) and the maximum slope point (MD). The diastolic bifurcation width (DBW) represents the width from the cross-tangential point (IT) to the next inflection point (V) and reflects peripheral vasodilation characteristics.

[0065] The feature extraction unit (240) quantifies the temporal and amplitude changes of the biosignal based on reference points such as the peak point (P), trough point (V), intersection tangent point (IT), and maximum slope point (MD), and aligns the feature values ​​derived therefrom with the time information input from the data collection unit (220) and transmits them to the blood pressure estimation unit (260).

[0066] Specifically, in this embodiment, the feature extraction unit (240) can sequentially process 16 consecutive PPG segments and extract one or more of the eight reference points for each segment. For example, in one segment, features can be calculated centered on the peak point (P), the intersection tangent point (IT), and the maximum slope point (MD), and in another segment, the trough point (V), photoplethysmography intensity (PI), systolic bifurcation width (SBW), and diastolic bifurcation width (DBW) can be calculated together to obtain various morphological information.

[0067] The feature values ​​extracted in this way are converted into forms such as the relative time difference between each reference point, amplitude ratio, and slope change rate, and are utilized in the blood pressure estimation unit (260). The feature extraction unit (240) can reflect even minute fluctuations in the waveform according to changes in the subject's physiological signal, and can analyze not only the period between the peak and trough points but also the geometric pattern within the waveform.

[0068] According to one embodiment, the feature extraction unit (240) converts data into a format suitable for machine learning while maintaining the physiological meaning of the signal. To this end, each feature value undergoes a normalization process and is scaled to a certain range, and is then input to the blood pressure estimation unit (260).

[0069] Therefore, the feature extraction unit (240) can quantitatively analyze the dynamic characteristics of blood flow using only the photoplethysmography signal.

[0070] In this way, the non-invasive continuous blood pressure estimation device according to the embodiment of the present invention can secure diversity of input information for the blood pressure estimation model (262) by stably calculating various morphological features through the feature extraction unit (240), and consequently improve the prediction accuracy of systolic blood pressure and diastolic blood pressure.

[0071]

[0072] The blood pressure estimation unit (260) according to an embodiment of the present invention can estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP) using biosignal features extracted from the feature extraction unit (240).

[0073] The above blood pressure estimation unit (260) includes a computation unit equipped with a machine learning or deep learning-based blood pressure estimation model (262), and specifically, may include a structure such as a time series convolutional neural network (TCN).

[0074] According to an embodiment, the blood pressure estimation unit (260) is a feature value extracted from a PPG signal (e.g., ST, ST VIT , ST VMD Blood pressure can be calculated using a blood pressure estimation model (262) that includes a Temporal Convolutional Network (TCN) with inputs such as , PI, BW, etc.

[0075]

[0076] FIG. 3 is a diagram showing the structure of a Temporal Convolutional Network (TCN) used in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0077] As described above, the blood pressure estimation unit (260) according to an embodiment of the present invention can estimate the systolic blood pressure and the diastolic blood pressure through a blood pressure estimation model (262) including a Temporal Convolutional Network (TCN) using feature values ​​extracted from a PPG signal as input.

[0078] According to one embodiment, the blood pressure estimation unit (260) can predict blood pressure from the feature values ​​of the input biosignal. The blood pressure estimation unit (260) receives the feature values ​​of the PPG signal transmitted from the feature extraction unit (240) as input and applies a deep learning model for analyzing time series data.

[0079] According to one embodiment, the blood pressure estimation model (262) included in the blood pressure estimation unit (260) includes a Temporal Convolutional Network (TCN). The Temporal Convolutional Network has a structure capable of processing time-series input data in parallel while maintaining time dependency. The TCN applied in this embodiment processes the input sequence in chronological order and is designed so that the signal at each time point influences the prediction at subsequent times points.

[0080] The blood pressure estimation unit (260) according to one embodiment is composed of an input layer, a hidden layer, and an output layer. The input layer receives feature values ​​extracted from a corrected PPG signal, and these feature values ​​include systolic timespan (ST) and systolic timespan variation based on cross-tangential variation (ST VIT ) and systolic interval variation based on maximum slope point variation (STVMD It includes the peak (P), valley (V), intersecting tangent (IT), and maximal derivative (MD), photoplethysmography intensity (PI), branch width (BW), systolic branch width (SBW), and diastolic branch width (DBW).

[0081] The hidden layer consists of multiple convolutional layers, activation functions, and normalization layers, and converts input time series data into learnable feature maps. The output layer calculates systolic and diastolic blood pressure based on the learned feature maps.

[0082] A blood pressure estimation unit (260) according to one embodiment recognizes the temporal pattern of input data by applying a filter in each convolutional layer. At this time, the convolutional kernel is extended in the direction of the time axis and can extract multiple levels of temporal information depending on the length of the signal. For example, the initial layer learns short-term signal patterns, and the deep layer learns long-term patterns. Through this hierarchical learning structure, the blood pressure estimation unit (260) can reflect not only instantaneous fluctuations of individual heartbeats but also long-term blood pressure trends.

[0083] According to one embodiment, the blood pressure estimation unit (260) processes a fixed number of cardiac cycles as a learning unit by fixing the length of the input sequence. Specifically, the input sequence consists of 16 consecutive cardiac cycles, which is set to a range sufficient to capture the temporal context of blood pressure changes. Additionally, the blood pressure estimation unit (260) can prevent overfitting by applying a dropout technique.

[0084] The blood pressure estimation unit (260) according to an embodiment of the present invention may have a structure that gradually increases the dilation factor for input feature values ​​in each temporal block. This structure enables efficient learning of the long-term dependency of the signal and can more precisely reflect the correlation between consecutive heart rate cycles.

[0085] As such, the blood pressure estimation unit (260) according to the embodiment of the present invention can reflect both the temporal variation of the PPG signal and the morphological information of the waveform by estimating blood pressure through a time-series convolutional neural network using feature values ​​extracted from the PPG signal as input. Accordingly, changes in systolic blood pressure and diastolic blood pressure can be predicted continuously and non-invasively, and precise blood pressure estimation based on temporal continuity is possible.

[0086]

[0087] A blood pressure estimation model (262) according to an embodiment of the present invention includes a plurality of temporal blocks, each temporal block includes two convolutional layers, and the dilation factor may increase exponentially depending on the depth of the convolutional layers.

[0088] As illustrated in FIG. 3, the blood pressure estimation model (262) according to an embodiment of the present invention includes a plurality of temporal blocks, and each temporal block may be structured to include two convolutional layers. In addition, the model is configured so that the dilation factor increases exponentially according to the depth of each convolutional layer, thereby enabling efficient learning of the long-term dependency of the input temporal signal.

[0089] A blood pressure estimation model (262) according to one embodiment includes a structure for recognizing temporal patterns and maintaining long-term signal correlations. Specifically, a time series block is a minimum unit for processing feature values ​​extracted from an input PPG signal and can analyze the time series characteristics of the signal in a multi-layered manner. Each time series block consists of two consecutive convolutional layers, and a temporal feature map is extracted from the input data through a convolution operation and passed through an activation function to perform a non-linear transformation. The first convolutional layer learns short-term time series dependencies by applying a relatively small dilation factor, and the second convolutional layer reflects long-term dependencies by applying a larger dilation factor.

[0090] According to one embodiment, the dilation factor applied in the time series block increases exponentially with the depth of the convolutional layer. For example, the dilation factor may increase to 1 in the first time series block, 2 in the second block, 4 in the third block, and 8 in the fourth block. This dilation factor structure enables the blood pressure estimation model (262) to capture long-term dependencies of the input signal and to simultaneously learn blood pressure change patterns over multiple cardiac cycles as well as short-term fluctuations.

[0091] According to one embodiment, each time series block includes a normalization layer and an activation function layer (ReLU, Rectified Linear Unit, etc.) in addition to the convolutional layer to increase learning stability and prevent overfitting.

[0092] According to one embodiment, the time series block within the blood pressure estimation model (262) is designed to perform parallel operations, resulting in higher computational efficiency compared to a traditional recurrent neural network (RNN). That is, since calculations for all time points of the input sequence can be performed simultaneously, long-term time series data of the PPG signal can be processed quickly and efficiently.

[0093] According to one embodiment, the blood pressure estimation model (262) may be composed of a total of 4 to 8 time series blocks, and the filter size of each block may be set to a range of 3 to 5. This setting is a structure optimized to reflect both short-period characteristics (e.g., interval between waveform peaks, amplitude change) and long-period characteristics (e.g., mean systolic interval time (ST), heart rate variability pattern) of the PPG signal.

[0094] According to one embodiment, the blood pressure estimation model (262) may apply a padding technique to maintain a constant output length regardless of the length of the input signal. As a result, the output of the time series block is maintained at the same length as the input sequence, and the corresponding blood pressure value for each heart rate cycle can be stably predicted. In addition, the blood pressure estimation model (262) is suitable for real-time blood pressure estimation environments because it continuously predicts systolic blood pressure (SBP) and diastolic blood pressure (DBP) by arranging the feature values ​​of the input PPG signal in chronological order.

[0095] According to one embodiment, the blood pressure estimation model (262) can simultaneously learn short-term or long-term signal dependencies through an exponential increase in the latency rate within a time series block. That is, as the deeper layers of the blood pressure estimation model (262) become, signal patterns covering a wider time range are learned, which accurately reflect the complex interactions between PPG signals. Due to these structural features, the blood pressure estimation model (262) can achieve the high precision and stability required for non-invasive continuous blood pressure estimation.

[0096]

[0097] A blood pressure estimation model (262) according to an embodiment of the present invention can be trained by using the Mean Squared Error (MSE) as the loss function and applying a normalized Adam (RAdam; Rectified Adam) algorithm as the optimizer.

[0098] The blood pressure estimation model (262) can predict systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the feature values ​​of the input PPG signal. At this time, the blood pressure estimation model (262) can quantify the prediction error by using the mean squared error (MSE) as a loss function. The mean squared error is the average of the squared differences between the predicted value and the actual value, and is the most common regression loss function for evaluating the magnitude of the overall error that the blood pressure estimation model (262) causes in the prediction.

[0099] According to one embodiment, the mean squared error loss function used in the learning process of the blood pressure estimation model (262) can adjust the output of the blood pressure estimation model (262) by simultaneously considering the prediction accuracy of systolic blood pressure and diastolic blood pressure. That is, in the learning process, the difference between the predicted blood pressure value and the actual blood pressure value is calculated for each learning iteration (Epoch), and the average of the squared values ​​is calculated as the loss value. A smaller loss value means that the blood pressure estimation model (262) predicts the actual blood pressure more accurately, and the learning can be repeated until the loss value is minimized.

[0100] According to one embodiment, the blood pressure estimation model (262) may apply a Rectified Adam (RAdam) algorithm as an optimizer to improve the efficiency and stability of learning. Adaptive Moment Estimation (Adam) optimizes using an adaptive learning rate. However, the existing Adam algorithm may have a slow convergence speed or oscillations due to an unstable adaptive rate in the early stages of learning. To improve this problem, an embodiment of the present invention applies a Rectified Adam (RAdam) algorithm to automatically adjust the learning rate in the early stages of learning and ensure stable convergence.

[0101] This RAdam algorithm can solve the problem of excessive fluctuation in the learning rate by correcting the bias of the moment estimate during the early stages of learning. Through this, the blood pressure estimation model (262) can learn stably while responding to the diversity of the data and the variability of the biosignal. In addition, RAdam has improved learning speed and convergence stability compared to the existing Adam, so the learning efficiency of the blood pressure estimation model (262) from the PPG signal can be improved.

[0102] According to the embodiment, the training of the blood pressure estimation model (262) is repeated approximately 200 times (epochs), and the learning rate can be set within the range of 0.0005. At this time, the parameters of the blood pressure estimation model (262) are updated through the backpropagation process by the mean squared error loss function, and the optimizer finely adjusts the parameters at each training step to minimize the loss. In addition, a validation dataset is used in parallel during the training process to prevent overfitting and maintain the generalization performance of the blood pressure estimation model (262).

[0103] According to the embodiment, after the training is completed, the blood pressure estimation model (262) predicts blood pressure based on actual PPG signals during the verification phase, and the performance of the blood pressure estimation model (262) is evaluated through the mean squared error between the predicted value and the actual value. As in the embodiment, a structure that applies the mean squared error as a loss function and uses a normalized Adam algorithm as an optimizer enables stable training while reflecting both the characteristics of non-linear biological signals and time-series patterns.

[0104] As such, by using the mean squared error (MSE) as the loss function and applying a normalized Adam (RAdam) algorithm as the optimizer, the learning stability of the blood pressure estimation model (262) can be improved and the prediction error reduced. Accordingly, the accuracy of systolic and diastolic blood pressure non-invasively estimated from PPG signals is increased, and the reliability and reproducibility of continuous blood pressure estimation can be improved.

[0105]

[0106] The blood pressure estimation unit (260) according to an embodiment of the present invention can calculate the mean arterial pressure (MAP) using systolic blood pressure and diastolic blood pressure values ​​calculated through a learned blood pressure estimation model (262).

[0107] The blood pressure estimation unit (260) according to the embodiment may include a computation module for calculating a physiologically valid mean arterial pressure based on a blood pressure value estimated non-invasively. Mean arterial pressure refers to the average pressure over a certain period of time while blood circulates through the artery, and is used as a key indicator for evaluating the cardiovascular condition. Generally, mean arterial pressure is calculated using systolic blood pressure and diastolic blood pressure, and in the embodiment of the present invention, it is calculated by applying the two blood pressure values ​​estimated through a learned blood pressure estimation model (262) to a computation formula.

[0108] According to one embodiment, the mean arterial pressure is a representative physiological indicator that reflects changes in blood pressure during a heart rate cycle, and the blood pressure estimation unit (260) includes a structure for precisely calculating it. Specifically, the mean arterial pressure can be calculated according to the following relationship.

[0109] MAP=DBP+1 / 3(SBP-DBP)

[0110] In the above equation, SBP represents systolic blood pressure and DBP represents diastolic blood pressure, and the equation is a value calculated by weighting the intervals in which blood pressure changes during the heart rate cycle. The blood pressure estimation unit (260) can perform this calculation process to calculate the average arterial pressure in real time based on the SBP and DBP estimated from the PPG signal.

[0111] According to the embodiment, the blood pressure estimation unit (260) can update the mean arterial pressure at each point in time based on continuous time series data input, thereby reflecting the trend of blood pressure fluctuation over time in real time. In addition, the calculated mean arterial pressure can be provided as a reference signal for evaluating the user's condition, feedback control, or linking with medical equipment. This allows the non-invasive blood pressure estimation device of the present invention to go beyond simple blood pressure value prediction and produce a blood pressure indicator with physiological reliability.

[0112] According to an embodiment, the blood pressure estimation unit (260) includes a dedicated computation module for calculating the mean arterial pressure, and the computation module mathematically processes the blood pressure value output from the learned blood pressure estimation model (262). Additionally, a moving average or filtering technique may be additionally applied to remove signal outliers during the calculation of the mean arterial pressure. In this case, averaging is performed on the data of consecutive heart rate intervals so that distortion caused by outliers in a single heart rate cycle is minimized.

[0113] A blood pressure estimation unit (260) according to one embodiment may store the calculated mean arterial pressure together with other physiological signal indicators (e.g., heart rate, pulse wave transit time, etc.) or transmit it to an external display and control system. This allows the user to monitor the blood pressure status in real time and analyze the trend of blood pressure changes over the long term. Additionally, since mean arterial pressure is a clinically important indicator, the device of the present invention can also be applied to medical diagnosis and preventive health management.

[0114] In this way, the blood pressure estimation unit (260) according to the embodiment of the present invention calculates the average arterial pressure using the systolic blood pressure and diastolic blood pressure values ​​calculated through the learned blood pressure estimation model (262), thereby providing accurate and reliable average blood pressure information in a non-invasive manner.

[0115]

[0116] A blood pressure estimation model (262) according to an embodiment of the present invention can learn by dividing the learning dataset into training, validation, and test sets in a 2:1:1 ratio, and can save the model at the point where the validation loss is minimized.

[0117] According to one embodiment, the training dataset of the blood pressure estimation model (262) includes various biosignal feature values ​​extracted from PPG signals, and these feature values ​​are used as input data to learn the relationship with systolic blood pressure (SBP) and diastolic blood pressure (DBP). The dataset is organized into units of data windows of a certain length, taking into account temporal continuity, and each window consists of multiple biosignal samples. In the step of configuring the training dataset, about 50% of the total data is allocated as the training set, 25% as the validation set, and the remaining 25% as the test set.

[0118] The above training set is used to train the parameters of the blood pressure estimation model (262). The blood pressure estimation model (262) repeatedly learns the relationship between the input signals included in the training set and the actual blood pressure values, and updates the weights of the blood pressure estimation model (262) in a direction that minimizes the loss function.

[0119] The above validation set is used to evaluate the intermediate results of the model's training and serves as a criterion for determining whether overfitting has occurred. That is, to prevent the blood pressure estimation model (262) from fitting well to the training data but having a large error on new data, the validation loss is monitored at each training iteration (Epoch).

[0120] The above test set is used for final performance evaluation after the model training is completed. The data included in the test set consists of independent data that was not used during the training and validation process, and is used as a standard for evaluating the generalization ability of the blood pressure estimation model (262). In one embodiment of the present invention, by configuring the dataset in a ratio of 2:1:1 as described above, sufficient training data can be secured while maintaining the statistical reliability of the validation and test data.

[0121] According to one embodiment, the blood pressure estimation model (262) is designed to store the optimal blood pressure estimation model (262) based on the point in time when the validation loss is minimized. Specifically, during the learning process, the loss value of the validation set is calculated for each iteration (Epoch), and when a point is detected where the loss value no longer decreases compared to the previous Epoch, the blood pressure estimation model (262) at that point in time is stored. This prevents overfitting, where the blood pressure estimation model (262) is overfitted to the training set, and improves blood pressure prediction performance in a real environment.

[0122] As such, according to an embodiment of the present invention, the blood pressure estimation model (262) learns by dividing the learning dataset into training, verification, and test sets in a ratio of 2:1:1, and saves the blood pressure estimation model (262) at the point where verification loss is minimized, thereby improving the learning efficiency and prediction accuracy of the blood pressure estimation model (262). Accordingly, the reliability of systolic and diastolic blood pressure estimated in a non-invasive manner is increased, and stable performance can be maintained even in a long-term continuous blood pressure monitoring environment.

[0123]

[0124] The above blood pressure estimation unit (260) is the ST VIT and the above ST VMD The prediction accuracy of systolic and diastolic blood pressure can be improved by learning a set of PPG signal-based features that includes [the element].

[0125] According to the embodiment, the blood pressure estimation unit (260) is a cross-tangential variation-based systolic interval variation (ST) among the input PPG signal-based feature sets. VIT ) and systolic interval variation based on maximum slope point variation (ST VMD By performing learning including ), the sensitivity and accuracy of blood pressure estimation can be improved compared to existing systolic interval time (ST)-only-based models.

[0126] Specifically, the blood pressure estimation unit (260) includes a Temporal Convolutional Network (TCN) and can learn by configuring 16 consecutive PPG signal segments as an input sequence. In this case, ST VIT and ST VMD Through the combination of the subjects, the correlation between the systolic and diastolic intervals is learned within the blood pressure estimation model (262), and based on this, the blood pressure of each heart rate cycle can be accurately predicted.

[0127]

[0128] FIG. 4 is a diagram showing the procedure flow of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0129] As described above, the data used includes ECG, PPG, and ABP (Arterial Blood Pressure) signals collected from the public biosignal database VitalDB and the pediatric intensive care database (VitalDB PICUs Dataset). After selecting subject data containing signals necessary for blood pressure estimation from the said database, multiple feature sets are extracted for each subject based on ECG, PPG, and ABP signals.

[0130] In this embodiment, a total of 157,678,505 feature sets were generated from approximately 3,766 subjects, and a balanced data composition was secured by limiting the number of sets to a maximum of 200,000 per age interval (in 5-year increments). As a result, approximately 2,312,381 feature sets were retained, and among these, 71,559 outliers falling outside the age percentiles (1st to 99th percentile) were removed, resulting in a final total of 2,240,822 feature sets being included in the analysis. Subsequently, the entire feature set is divided into 1,115,681 sets for training, 574,967 sets for validation, and 550,174 sets for testing, which are then used for training a Temporal Convolutional Network (TCN).

[0131]

[0132] Figure 5 is a diagram showing the number of feature sets by age group of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0133] As illustrated, the results show that the entire data collected from the biosignal databases (VitalDB and VitalDB PICUs Dataset) is organized into 5-year intervals based on age ranges (0 years and older, but less than 95 years). The feature sets extracted for each age range are classified into the Open Dataset, the PICUs Dataset, and outliers, with each color in the bar graph representing one of these three categories. In the example, up to 200,000 feature sets were randomly extracted for each age range, and outliers are defined as cases where systolic blood pressure (SBP) or diastolic blood pressure (DBP) falls outside the age-specific percentile range (1st to 99th percentile). As illustrated, the PICUs Dataset has a high proportion in the age range of 0 years and older, while the Open Dataset has a relatively high proportion in the age range of 30 years and older. Furthermore, the total number of data points tends to decrease as age increases, which reflects the actual clinical data collection environment. Therefore, this indicates that the blood pressure estimation device according to the embodiment of the present invention is designed to ensure blood pressure estimation accuracy that reflects physiological differences according to age by configuring balanced training data considering age-specific characteristics.

[0134]

[0135] FIG. 6 is a histogram showing the blood pressure distribution according to age percentiles of a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0136] As described, Figure A shows the distribution of systolic blood pressure (SBP), and Figure B shows the distribution of diastolic blood pressure (DBP). In one embodiment, the blood pressure data of each subject was classified according to the 1st to 99th percentile intervals based on age, and if it falls within the above range, it is indicated as 'Within (normal range)', and if it falls outside the above range, it is indicated as 'Outside (outlier)'.

[0137] As can be seen in the illustrated histogram, most systolic and diastolic blood pressures are distributed within the normal range (Within), while outliers (Outside) exist in a relatively small proportion. These data distribution characteristics are utilized as a preprocessing step to ensure the stability and reliability of the training data in the blood pressure estimation device of the present invention. Therefore, this drawing visually demonstrates that the blood pressure estimation model (262) of the present invention can improve blood pressure prediction accuracy by systematically excluding data outliers based on the physiological blood pressure range by age.

[0138]

[0139] FIG. 7 is a scatter plot showing the actual blood pressure (BP) and the blood pressure estimated by the blood pressure estimation model (262) in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0140] As shown, Fig. 7A shows a scatter plot for systolic blood pressure (SBP), and Fig. 7B shows a scatter plot for diastolic blood pressure (DBP).

[0141] In Fig. 7, each point corresponds to an individual feature set, and the density of the points is represented by the color gradient shown on the right. A higher density of points indicates a higher degree of agreement between the actual blood pressure and the estimated blood pressure, and in the case of the systolic blood pressure in Fig. 7A, the coefficient of determination (R²) 2 ) is found to be approximately 0.931, confirming that the blood pressure estimation model (262) of the present invention has a very high correlation with the actual systolic blood pressure. In addition, in the case of the diastolic blood pressure in FIG. 7B, R 2 With a value of approximately 0.910, it can be seen that the prediction accuracy with respect to actual diastolic blood pressure is also excellent.

[0142] Accordingly, FIG. 7 demonstrates that a blood pressure estimation model (262) according to an embodiment of the present invention can predict actual blood pressure with high precision based on feature values ​​extracted from a PPG signal, and visually illustrates the reliability and prediction performance of a non-invasive continuous blood pressure estimation device.

[0143]

[0144] FIG. 8 shows a Bland-Altman plot indicating the degree of agreement between actual blood pressure (BP) and blood pressure estimated by a blood pressure estimation model (262) in a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention.

[0145] As illustrated, Figure 8A shows the relationship between actual systolic blood pressure (SBP) and estimated systolic blood pressure, and Figure 8B shows the relationship between actual diastolic blood pressure (DBP) and estimated diastolic blood pressure.

[0146] In each plot, the horizontal axis represents the mean value of actual blood pressure and estimated blood pressure, and the vertical axis represents the difference between the two blood pressure values; the color of the points indicates the density level of the data through the color gradient shown on the right. The points in the plot correspond to feature sets for each subject, and the horizontal dashed line represents the mean difference (bias) and the range of agreement (±1.96 standard deviations from the mean difference). This indicates that it falls within the range of the 95% confidence interval.

[0147] As illustrated, it can be seen that most points are distributed within the agreement range for both systolic and diastolic blood pressure, which indicates that the blood pressure estimation model (262) according to the embodiment of the present invention has a statistically significant level of consistency and accuracy with actual blood pressure. Thus, FIG. 8 indicates that a non-invasive continuous blood pressure estimation device can provide stable blood pressure estimation performance within a clinically acceptable error range, as well as a high correlation between the model prediction value and the actual measurement value.

[0148]

[0149] FIG. 9 is a diagram showing the results of analyzing the importance and influence of each feature used for blood pressure estimation in the blood pressure estimation model (262) according to an embodiment of the present invention as SHAP values ​​(Shapley Additive Explanations).

[0150] As described, the graph on the left represents the average absolute SHAP value of each feature to quantitatively express overall feature importance, while the graph on the right plots the distribution of SHAP values ​​to visually indicate the direction and magnitude of the influence of individual features on the model output (blood pressure estimate). The color gradient in the figure is distinguished according to the magnitude of each feature value, with blue representing low values ​​and red representing high values. Key features include Pulse Arrival Time (PAT), PI Ratio (PIR), Acceleration Plethysmogram Intensity (AI), Branch Width (BW), Systolic Branch Width (SBW), Diastolic Branch Width (DBW), Photoplethysmogram K Value (PPGK), Intersecting Tangent (IT), Systolic Timespan (ST), and Intersecting Tangent Variation-based Systolic Timespan (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD ...etc. are included.

[0151] According to the above analysis results, PAT P , PAT MD and PIR P Time-based and intensity-based features such as [the above] have the greatest influence on blood pressure estimation, which means that the blood pressure estimation model (262) according to the embodiment of the present invention accurately predicts blood pressure by utilizing the time difference and waveform shape change of the PPG signal as key indicators. Accordingly, FIG. 9 shows that the blood pressure estimation model (262) of the present invention secures both reliability and interpretability of non-invasive blood pressure estimation by systematically reflecting the relative importance of each biosignal feature.

[0152]

[0153] FIG. 10 is a flowchart of a non-invasive continuous blood pressure estimation method according to an embodiment of the present invention. Since it overlaps with the previously described embodiment, a detailed description of this embodiment will be omitted.

[0154] As described above, a non-invasive continuous blood pressure measurement device according to an embodiment of the present invention comprises the steps of: detecting a photoplethysmography (PPG) signal of a subject (S910); collecting the PPG signal and storing it together with time information (S920); extracting one or more biosignal features among a systolic timespan (ST), a systolic timepan based on cross-tangential variation (STVIT), and a systolic timepan based on maximum slope variation (STVMD) derived from a specific point of the PPG signal (S930); and estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the extracted biosignal features (S940).

[0155] According to an embodiment of the present invention, the step (S910) of detecting a photoplethysmography (PPG) signal of a subject is a process of detecting a change in blood flow volume within a blood vessel using a PPG sensor. The PPG signal is obtained by detecting a change in the amount of light reflected or transmitted from the surface of the skin and provides waveform information according to the change in blood flow volume within the blood vessel. The PPG signal can be collected at a constant sampling frequency and aligned with respect to the same time axis.

[0156] According to an embodiment of the present invention, in the step (S920) of collecting a PPG signal and storing it together with time information, a data collection unit acquires the PPG signal transmitted from the sensor unit at the same sampling frequency and records the time information at the time the signal was collected together to form a data set. The collected data is aligned with respect to the time axis to maintain the continuity of the signal and is stored as raw data so that time-series-based processing can be performed in the subsequent feature extraction and blood pressure estimation steps.

[0157] According to an embodiment of the present invention, a systolic timespan (ST; Systolic Timespan) derived from a specific point of the PPG signal, and a systolic timespan variation based on cross-tangential variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD In the step (S930) of extracting one or more biological signal features, the feature extraction unit detects waveform features such as peaks (P; Peak), valleys (V; Valley), intersecting tangents (IT; Intersecting Tangent), and maximal derivatives (MD; Maximal Derivative) within the PPG waveform and calculates temporal and morphological parameters.

[0158] For example, ST is defined as the time interval from the trough point (V) to the peak point (P) within one period, and ST VIT and ST VMD It can be obtained by calculating ST fluctuations between consecutive heart rate cycles based on the cross-tangential point and the maximum slope point. These features are key factors reflecting the subject's vascular elasticity, pulse wave conduction characteristics, and heart rate dynamics, and are used as inputs for a blood pressure estimation model.

[0159] According to an embodiment of the present invention, the step (S940) of estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the extracted biosignal features is performed by a blood pressure estimation unit.

[0160] The blood pressure estimation unit calculates blood pressure from input feature values ​​through a blood pressure estimation model comprising a Temporal Convolutional Network (TCN). The blood pressure estimation model includes multiple time-series blocks and receives continuous PPG signal segments as input to learn temporal dependencies and patterns, thereby predicting SBP and DBP corresponding to each heart rate cycle. The learned model reflects fine morphological changes in the PPG waveform, enabling continuous blood pressure estimation that can stably respond to changes in the subject's physiological state.

[0161]

[0162] The non-invasive continuous blood pressure estimation method according to the embodiment of the present invention is based solely on PPG signals without using ECG signals, and thus offers higher wearability compared to existing multi-sensor-based measurement methods and is suitable for long-term continuous monitoring. In addition, ST VIT and ST VMD By reflecting new morphological features such as [ ], the accuracy of blood pressure prediction is improved, and real-time blood pressure fluctuations of the subject can be tracked more precisely.

[0163]

[0164] Table 1 below shows the results of evaluating the blood pressure (BP) estimation performance of the blood pressure estimation model according to an embodiment of the present invention in accordance with international standards.

[0165]

[0166]

[0167] Table 1 shows the key indicators, such as Mean Error (ME), Standard Deviation Error (SDE), Mean Absolute Deviation (MAD), and Cumulative Error (CE), for the Systolic Blood Pressure (SBP) estimation model and the Diastolic Blood Pressure (DBP) estimation model, respectively. CE represents the percentage of predicted values ​​that fall within ±5 mmHg, ±10 mmHg, and ±15 mmHg of the actual blood pressure value.

[0168] According to the results in Table 1, the blood pressure estimation model according to the embodiment of the present invention satisfied a 'PASS' or 'A' grade in all indicators according to international blood pressure measurement standards (e.g., AAMI, BHS standards). In particular, high precision was secured with a mean absolute deviation (MAD) of 3 mmHg or less for both systolic and diastolic blood pressure, and recorded 91.5% and 97.2%, respectively, within the cumulative error range of 10 mmHg, demonstrating excellent agreement in terms of clinical reliability.

[0169] Therefore, Table 1 above is evidence showing that the blood pressure estimation model of the present invention maintains statistically high correlation and accuracy with actual blood pressure measurements, despite being a non-invasive method.

[0170]

[0171] Table 2 below shows the results of analyzing the blood pressure estimation performance of the blood pressure estimation model according to an embodiment of the present invention, based on age group and gender.

[0172]

[0173]

[0174] As shown in Table 2 above, the performance evaluation is based on the coefficient of determination (R²) for systolic blood pressure (SBP) and diastolic blood pressure (DBP). 2 It includes the mean error (ME), standard deviation error (SDE), mean absolute deviation (MAD), and cumulative error (CE). CE represents the percentage of estimates that fall within ±5 mmHg, ±10 mmHg, and ±15 mmHg of the actual blood pressure value.

[0175] According to the results in Table 2, the highest coefficient of determination (R²) for both systolic and diastolic blood pressure was found in the age group of 15 to under 20 years. 2 By exhibiting a low mean error (ME) and a low mean error (ME), it can be confirmed that the blood pressure estimation model shows the highest accuracy in the corresponding age group. On the other hand, in the age groups of 45 to under 55 and 55 to under 65, a tendency for the mean error and standard deviation error to increase slightly was observed due to physiological changes and a decrease in vascular elasticity.

[0176] Gender analysis showed that the female group had a mean absolute deviation (MAD) of 2.716 mmHg in systolic blood pressure prediction, which was slightly lower than the male group's 2.727 mmHg, indicating higher consistency. In diastolic blood pressure prediction as well, the standard deviation error (SDE) for the female group was 4.104 mmHg and for the male group was 3.913 mmHg, confirming a similar level of stability.

[0177] In addition, in the estimation of diastolic blood pressure, most age groups showed a cumulative accuracy of over 90% in the results of CE10 and CE15, and in particular, CE15 satisfied over 95% across all age groups, proving that the blood pressure estimation model according to the present invention operates stably in various population groups.

[0178] Therefore, despite the present invention being a blood pressure estimation method using only photoplethysmography (PPG) signals, it can secure prediction performance comparable to existing composite signal-based methods including ECG signals by maintaining high correlation and low error across various age and gender ranges.

[0179]

[0180] FIG. 11 illustrates a non-invasive continuous blood pressure estimation device according to an embodiment of the present invention. Referring to FIG. 11, the non-invasive continuous blood pressure estimation device (10) may include a sensor unit (100), a processing unit (200), a storage unit (300), and a communication unit (400).

[0181] The processing unit (200) can perform signal preprocessing, feature extraction, blood pressure estimation, and mean arterial pressure calculation operations based on the photoplethysmography (PPG) signal collected from the PPG sensor (120).

[0182] The processing unit (200) first aligns the time axis of the PPG signal input from the data collection unit, and through the feature extraction unit, obtains the systolic timespan (ST; Systolic Timespan), the systolic timespan variation based on the intersection tangent variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD It detects signal features such as peaks (P), valleys (V), intersecting tangents (IT), and maximal derivatives (MD).

[0183] The processing unit (200) transmits the detected PPG signal to the blood pressure estimation unit to execute a blood pressure estimation model configured based on a Temporal Convolutional Network (TCN). The processing unit (200) continuously estimates systolic blood pressure (SBP) and diastolic blood pressure (DBP) through the learned blood pressure estimation model, and calculates the mean arterial pressure (MAP) from the calculated blood pressure values.

[0184] The storage unit (300) is configured to store data necessary for the operation of the non-invasive continuous blood pressure estimation device (10), and may include, for example, PPG raw signal data, feature extraction results, learned model parameters, blood pressure estimation results, mean arterial pressure calculation results, time information, user identification information, etc. Additionally, the storage unit (300) may store the configurations of the training, validation, and test datasets used for model learning in a 2:1:1 ratio, and preserve the weight values ​​of the model with minimized validation loss so that they can be utilized for the retraining and updating of the device.

[0185] The communication unit (400) can transmit and receive data between the non-invasive continuous blood pressure estimation device (10) and a user terminal, an external analysis server, or a cloud-based data storage. The communication unit (400) can transmit measured PPG signals, blood pressure estimation results, and mean arterial pressure (MAP) data to the outside via wireless communication such as Wi-Fi, Bluetooth, LTE, 5G, or a wired network (Ethernet, LAN), or update measurement conditions and reference data set externally to the device.

[0186] The processing unit (200) can process the PPG signal in real time by referring to reference data and control conditions stored in the storage unit (300), calculate continuous blood pressure values ​​through a blood pressure estimation model, and then provide the results to the user interface (UI) through the communication unit (400). Additionally, measurement results such as the PPG signal waveform, systolic blood pressure, diastolic blood pressure, mean arterial pressure, heart rate cycle, and pulse wave transit time (PAT) can be visually displayed on the user interface (UI).

[0187] The non-invasive continuous blood pressure estimation device (10) according to an embodiment of the present invention can be applied to application environments such as hospital patient monitoring systems, home healthcare devices, smartwatches, and wearable devices, and can be used for biosignal-based health monitoring and data analysis by continuously and non-invasively measuring the blood pressure of a subject.

[0188]

[0189] As such, the non-invasive continuous blood pressure estimation device according to an embodiment of the present invention can continuously estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP) using only photoplethysmography (PPG) signals.

[0190] In addition, a non-invasive continuous blood pressure estimation device according to one embodiment of the present invention includes a cross-tangential point variation-based systolic interval variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD By adding new morphological features such as ), the accuracy of blood pressure estimation can be further improved compared to existing blood pressure estimation models that use only the systolic interval time (ST).

[0191] In addition, since it uses only PPG signals, it can be implemented in the form of a simple sensor or wearable device attached to the finger, wrist, or ear, and allows for easy non-invasive blood pressure monitoring in a home environment or medical institution.

[0192]

[0193] The description above is merely an example of applying the principles of the present invention, and other configurations may be further included without departing from the scope of the present invention. For example, at least some of the various embodiments of the present invention described above may be combined.

Claims

1. A sensor unit including a PPG sensor that detects the photoplethysmography (PPG) signal of a subject; A data collection unit that collects a PPG signal from the sensor unit and stores it together with time information; A feature extraction unit that extracts one or more biosignal features among Systolic Timespan (ST), Systolic Timespan based on Cross Tangent Change (STVIT), and Systolic Timespan based on Maximum Slope Change (STVMD) derived from a specific point of the above PPG signal; A blood pressure estimation unit that estimates systolic blood pressure (SBP) and diastolic blood pressure (DBP) using biosignal features extracted from the above feature extraction unit; A non-invasive continuous blood pressure estimation device comprising 2. In Paragraph 1, The above data collection unit is, Collecting the above PPG signals at the same sampling frequency and aligning the collected PPG signals based on the time axis, Non-invasive continuous blood pressure estimation device.

3. In Paragraph 1, The above feature extraction unit is, Extracting one or more of the following from the above PPG signal: Peak (P), Valley (V), Intersecting Tangent (IT), Maximal Derivative (MD), PPG Intensity (PI), Branch Width (BW), Systolic Branch Width (SBW), and Diastolic Branch Width (DBW). Non-invasive continuous blood pressure estimation device.

4. In Paragraph 1, The above blood pressure estimation unit is, Estimating the systolic blood pressure and the diastolic blood pressure through a blood pressure estimation model including a Temporal Convolutional Network (TCN) using the extracted biosignal feature values ​​as input, Non-invasive continuous blood pressure estimation device.

5. In Paragraph 4, The above blood pressure estimation model is, It includes a plurality of temporal blocks, each temporal block includes two convolutional layers, and the dilation factor increases exponentially with the depth of the convolutional layers, Non-invasive continuous blood pressure estimation device.

6. In Paragraph 4, The above blood pressure estimation model is, The input sequence consists of 16 consecutive PPG signal segments, includes 3 hidden layers, and applies dropout to prevent overfitting, Non-invasive continuous blood pressure estimation device.

7. In Paragraph 4, The above blood pressure estimation model is, Learned by using Mean Squared Error (MSE) as the loss function and applying the Rectified Adam (RAdam) algorithm as the optimizer, Non-invasive continuous blood pressure estimation device.

8. In Paragraph 4, The above blood pressure estimation unit is, Calculating Mean Arterial Pressure (MAP) using systolic and diastolic blood pressure values ​​derived through a learned blood pressure estimation model, Non-invasive continuous blood pressure estimation device.

9. In Paragraph 8, The above blood pressure estimation model is, Training by splitting the training dataset into training, validation, and test sets in a 2:1:1 ratio, and saving the model at the point where the validation loss is minimized, Non-invasive continuous blood pressure estimation device.

10. In Paragraph 1, The above blood pressure estimation unit is, The above ST VIT and the above ST VMD Learning a PPG signal-based feature set including, Non-invasive continuous blood pressure estimation device.

11. A step of detecting the photoplethysmography (PPG) signal of the subject; A step of collecting the above PPG signal and storing it together with time information; Systolic Timespan (ST) derived from a specific point of the above PPG signal, and Systolic Timespan variation based on cross-tangential variation (ST VIT ) and systolic interval variation based on maximum slope point variation (ST VMD A step of extracting one or more biosignal features from ); and A method comprising the step of estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) using the extracted biosignal features. Non-invasive continuous blood pressure estimation method.

12. In Paragraph 11, The step of collecting the above PPG signal is, A method comprising the step of collecting the above PPG signals at the same sampling frequency and aligning the collected above PPG signals based on the time axis to perform data synchronization. Non-invasive continuous blood pressure estimation method.

13. In Paragraph 11, The step of extracting the above biosignal features is, A method comprising the step of extracting one or more of a peak (P), valley (V), intersecting tangent (IT), maximal derivative (MD), photoplethysmography intensity (PI), branch width (BW), systolic branch width (SBW), and diastolic branch width (DBW) from the above PPG signal. Non-invasive continuous blood pressure estimation method.

14. In Paragraph 11, The step of estimating the above blood pressure is, A step comprising estimating the systolic blood pressure and the diastolic blood pressure through a blood pressure estimation model including a Temporal Convolutional Network (TCN) using the extracted biosignal feature values ​​as input. Non-invasive continuous blood pressure estimation method.

15. In Paragraph 14, The above blood pressure estimation model is, It includes a plurality of temporal blocks, each temporal block includes two convolutional layers, and the dilation factor increases exponentially with the depth of the convolutional layers, Non-invasive continuous blood pressure estimation method.

16. In Paragraph 14, The above blood pressure estimation model is, A method comprising the step of configuring the input sequence into 16 consecutive PPG signal segments, including three hidden layers, and applying dropout to prevent overfitting, Non-invasive continuous blood pressure estimation method.

17. In Paragraph 14, The step of training the above blood pressure estimation model is, A step including a learning step that uses Mean Squared Error (MSE) as the loss function and applies the Rectified Adam (RAdam) algorithm as the optimizer, Non-invasive continuous blood pressure estimation method.

18. In Paragraph 14, The step of estimating the above blood pressure is, A method comprising the step of calculating the Mean Arterial Pressure (MAP) using systolic and diastolic blood pressures calculated through a learned blood pressure estimation model, Non-invasive continuous blood pressure estimation method.

19. In Paragraph 17, The step of training the above blood pressure estimation model is, A method comprising the step of splitting a training dataset into training, validation, and test sets in a 2:1:1 ratio for training, and saving the model at the point where the validation loss is minimized. Non-invasive continuous blood pressure estimation method.

20. In Paragraph 11, The above blood pressure estimation unit is, The above ST VIT and ST VMD A step comprising improving the prediction accuracy of systolic and diastolic blood pressure by learning a PPG signal-based feature set including Non-invasive continuous blood pressure estimation method.

Images