KR102360697B1 - Contactless vital-sign measuring system - Google Patents

Abstract

The present invention relates to a pre-diagnosis method through non-contact pulse signal measurement and analysis, and more particularly, to obtain a bio-signal from a pulse wave obtained by photographing a subject with an imaging device, the correlation between the pulse wave and the bio-signal is preliminarily analyzed by machine learning. Learning, but obtaining emergency and disease situation data, which is difficult to obtain data during supervised learning, through computational fluid analysis (CFD), and assisting with data It relates to a non-contact biosignal measurement system that can be quickly estimated.

Description

Contactless vital-sign measuring system

The present invention relates to a pre-diagnosis method through non-contact pulse signal measurement and analysis, and more particularly, to a non-contact bio-signal measurement system capable of acquiring a bio-signal from a pulse wave obtained by photographing a subject with an imaging device.

The body's vital signs are called vital-signs. Biosignals can be classified into several types according to the classification method. Typically, four biosignals consisting of body temperature, heart rate (pulse), respiration, and blood pressure are used. Based on this, it is possible to roughly check the condition of the body and select the corresponding treatment. Therefore, various methods are used to measure biosignals. Photoplethsmography (PPG) detects changes in blood volume (pulse waves) in the microvascular layer of the body, and uses a transmission type, a reflection type, and the like. Among them, transmissive PPG equipment including a light source and a light receiver is transmitted when light is irradiated to the skin from a light source such as a light emitting diode (LED), and absorption of light occurs in blood, bone, and tissue, and some light is emitted from the opposite side. reach the receiver. The degree of light absorption is proportional to the amount of blood in the skin and tissue in the path, and since the components do not change except for the change in blood flow (pulse wave) caused by heartbeat, the amount of light absorbed corresponds to the amount of change in blood flow. Therefore, since the transmitted light detected by the light receiving unit is received by subtracting the amount of light absorbed by the contacted portion, the change in the amount of transmitted light may also change the blood flow. Accordingly, it is possible to detect a change in blood flow corresponding to a heartbeat through the amount of light measured by the optical receiver. Therefore, when acquiring the patient's bio-signals through PPG, it is common to attach the equipment to the areas where the irradiated light is easy to transmit, such as the patient's fingers, wrists, and earlobes. In the case of various side effects such as skin necrosis and secondary damage through contact with a patient, the use is limited. In order to solve this problem, International Patent Publication No. 2015-055709 ("DEVICE AND METHOD FOR OBTAINING A VITAL SIGN OF A SUBJECT" 2015.04.23.) discloses a method of acquiring a patient's biosignal in a non-contact manner through an imaging device. However, when measuring a patient's bio-signals through imaging equipment, there may be a problem of a decrease in reliability due to interference from atmospheric temperature, humidity, and external factors. Furthermore, in the case of a large number of seriously injured patients at the scene of an accident, it is necessary to obtain biosignals to determine priorities for each emergency patient, and rescuers and medical staff must provide appropriate treatment to the patient in a short time to increase the survival rate of the patient. In the case of obtaining the patient's biosignals by the above method, the measurement is difficult, that is, in the case of contact PPG, the condition of the patient can be further aggravated by secondary injury while contacting the affected part of the patient, and in the case of non-contact PPG equipment, the equipment installation Due to limitations or interference from external factors, accuracy is lowered and the patient's survival rate is also lowered.

1. International Patent Publication No. 2015-055709 ("DEVICE AND METHOD FOR OBTAINING A VITAL SIGN OF A SUBJECT" 2015.04.23.)

1. Kim YW et al., Journal of Mechanical Science and Technology, 2016. 2. Matsui Y et al., Hypertension Research, 2004. 3. Wilkinson IB et al., Journal of the American College of Cardiology, 2002. 4. Manisty CH et al., Hypertension, 2009.

The present invention has been devised to solve the above problems, and the non-contact bio-signal measurement system of the present invention is a system for deriving bio-signals by obtaining the pulse wave of a subject in a non-contact manner through portable photographing equipment such as smartphones, drones, and wearable devices. Through pre-supervised learning, the correlation between the pulse wave and the bio-signal from the data set consisting of the pulse wave and the bio-signal measured in advance is derived after machine learning. The purpose of this is to improve the accuracy and reliability of the algorithm for deriving bio-signals from pulse waves with the result values derived from the computational fluid analysis of a data set composed of pulse waves and bio-signals, assuming emergency situations that cannot be measured in advance.

In order to solve the above problems, a non-contact biosignal measuring system of the present invention includes: a pulse wave acquisition means for acquiring pulse wave data of a subject; a biosignal derivation unit for deriving a correlation between the pulse wave data obtained from the pulse wave obtaining means and the biosignal with an algorithm using a machine learning technique, and applying the pulse wave data to the algorithm to derive the biosignal of the subject; and an output terminal for receiving and outputting the subject's biosignal derived from the biosignal derivation unit.

Here, the bio-signal derivation unit includes: a learning unit for receiving a data set composed of pre-stored pulse waves and bio-signals and learning the correlation between the pulse waves and bio-signals through machine learning; an algorithm derivation unit for deriving an algorithm based on the correlation between the pulse wave and the biosignal learned by the learning unit; and an algorithm application unit that receives the pulse wave data from the pulse wave acquisition means and derives a biosignal of a subject by applying the algorithm.

Here, the learning unit includes a computational fluid analysis unit that performs computational fluid analysis (CFD) by applying the dataset composed of the pulse wave and biosignals to the virtual computer circulatory system model as a condition, and the computational fluid analysis (CFD) is performed with the computational fluid analysis unit. It is characterized in that the correlation between the pulse wave learned by the learning unit and the biosignal is supplemented with data.

In this case, the learning unit includes: a main factor extracting unit for pre-processing data to extract the main factors from the pulse wave; and a first learning model unit for learning the correlation between the main factor and the biosignal, and a second learning model unit for classifying the area of the biosignal estimated from the main factor.

At this time, the first learning unit is characterized in that the machine learning using a Gaussian regression analysis model (Gaussian Processes).

Furthermore, the second learning unit is characterized in that machine learning using a support vector machine model (Support Vector Machine).

In addition, the algorithm application unit transmits a warning message to the output terminal when the derived bio-signal of the examinee is out of the range of the pre-stored bio-signal, and the output terminal outputs the received warning message.

In this case, the warning message includes any one or more of the subject's severity, rescue order, time limit, and target hospital.

In addition, the pulse wave acquisition means is characterized in that the terminal includes a photographing device capable of photographing the subject.

In addition, the terminal is characterized in that any one of a drone, a smartphone, and a wearable device.

The non-contact biosignal measurement system of the present invention according to the above configuration can derive the correlation between the pulse wave and the biosignal by an algorithm after machine learning through supervised learning, and it is necessary for the accuracy of the algorithm, but it is difficult to obtain emergency situations and diseases Computational fluid analysis (CFD) assists in situations involving elements to increase the reliability and accuracy of the algorithm. Ultimately, the pulse wave of the examinee obtained with the imaging equipment is quickly derived as a biosignal, and based on the outputted biosignal, the examinee It has the effect of providing information such as severity, rescue order, time limit, and target hospital to photographers and rescue personnel.

1 is a conceptual diagram illustrating a non-contact biosignal measurement system 1000 of the present invention.
2 is a block diagram illustrating a biosignal derivation unit of the non-contact biosignal measurement system 1000 of the present invention.
3 is a block diagram illustrating a learning unit of the non-contact biosignal measurement system 1000 .
4 is a graph showing characteristics of a pulse wave;
5 is a timing diagram for acquiring a biosignal from a pulse wave of a subject using the non-contact biosignal measuring system 1000 of the present invention.
6 is a timing diagram illustrating learning a correlation between a pulse wave and a biosignal using the non-contact biosignal measurement system 1000 of the present invention.
7 is a conceptual diagram illustrating an embodiment using the non-contact biosignal measurement system 1000 of the present invention.

Hereinafter, the technical idea of the present invention will be described in more detail with reference to the accompanying drawings.

Since the accompanying drawings are merely examples shown in order to explain the technical idea of the present invention in more detail, the technical idea of the present invention is not limited to the form of the accompanying drawings.

Before describing the non-contact biosignal measuring system 1000 of the present invention, the pulse wave obtained by measuring the subject 10 in the pulse wave obtaining means 100 is called pulse wave data, and the pulse wave described later is used and measured by the subject. It is a notice in advance that interpretations may differ depending on the For example, the pulse wave used in the learning unit 200 is a data set composed of pre-stored pulse wave and bio-signals, that is, clinical data accumulated by medical staff for many years, that is, big-data of a dataset composed of pulse waves and bio-signals. to be. Furthermore, the data set composed of pre-stored pulse waves and bio-signals used in the computational fluid analysis unit 240 included in the learning unit 200 may use clinical data accumulated over many years from patients by medical staff, but some of them is the main purpose of the computational fluid analysis unit 240, that is, an example of a situation in which it is difficult to obtain clinical data separately from the dataset composed of pulse waves and biological signals that can be easily acquired and accumulated as clinical data by the learning unit 200 For example, it is desirable to use the hypothesized pulse wave and biosignal data set in order to solve the problem that it is not easy to obtain clinical data during an emergency situation in which a patient fights for life. At this time, the assumed pulse wave and biosignal data set can be used in various research papers, academic journals and medical journals, and furthermore, computational fluid analysis (CFD) performed by professional medical staff, but arbitrarily performed according to medical theory and principles It can be assumed from the pulse wave and biosignals according to Therefore, the non-contact bio-signal measurement system 1000 of the present invention learns data of various situations as a learning model and obtains a more reliable bio-signal from the pulse wave data of the subject to be measured in the future, and through high reliability and quick response, the subject's It can help with rescue and treatment.

As shown in FIG. 1 , the non-contact biosignal measuring system 1000 of the present invention includes a pulse wave acquisition means 100 , a biosignal derivation unit 200 , and an output terminal 300 . Each component is described in detail as follows.

The pulse wave acquisition means 100 is a means for acquiring pulse wave data from the subject. Specifically, a non-contact measurement of a region from which the pulse wave data can be obtained is transmitted to the biosignal derivation unit 200 to transmit the measured pulse wave data of the subject. Preferably, the pulse wave acquisition means 100 includes a separate transceiver or communication unit for transmitting the measured pulse wave data to the biosignal determination unit 200 .

The biosignal derivation unit 200 serves to convert the subject's pulse wave data received from the pulse wave obtaining means 100 into a biosignal, and has a pulse wave and a biometric signal before the time of receiving the subject's pulse wave data measured by the pulse wave obtaining means 100 . A data set composed of signals is received, the correlation between the pulse wave and the biosignal is learned through machine learning, and an algorithm is derived from the correlation, and the subject pulse wave data received from the pulse wave acquisition means 100 is converted into a subject biosignal. and serves to output information on the condition of the subject to the output terminal 300 based on the derived bio-signals. The biosignal derivation unit 220 is composed of a high-performance computer system capable of calculating machine learning, computational flow analysis, etc., and a separate storage device capable of storing the calculated information and a communication unit or transmitting/receiving unit for transmitting and receiving information. it is preferable

The output terminal 300 serves to receive and receive the subject's bio-signals output from the bio-signal derivation unit 200 and the subject's status information based on the bio-signals. The output terminal 300 may transmit/receive various information including a separate transceiver or communication unit, and of course includes a display device for displaying it.

Here, the pulse wave acquisition means 100 is formed of a terminal including a photographing device capable of photographing a subject, and the terminal may have several embodiments capable of photographing the subject with any one of a drone, a smartphone, and a wearable device. have. Furthermore, the output terminal 300 and the pulse wave acquisition means 100 may have an embodiment capable of acquiring a pulse wave and output information in one terminal in an integrated manner.

As shown in FIG. 2 , the biosignal derivation unit 200 of the non-contact biosignal measurement system 1000 of the present invention includes a learning unit 210 , an algorithm derivation unit 220 , an algorithm application unit 230 , and computational fluid analysis. part 240 . Each part is described in detail as follows.

The learning unit 210 receives a data set composed of a pre-stored pulse wave and bio-signal, and serves to learn the correlation between the pulse wave and the bio-signal through machine learning. Machine learning learned here is a method of machine learning to infer a function from training data as supervised learning. At this time, the training data is a data set composed of pulse waves and biosignals, and one function is a vital signal. As the data set composed of the pulse wave and the biosignal, it is preferable to use the pulse wave and the biosignal measured in an actual clinical test. Furthermore, it goes without saying that the pulse wave data measured by the pulse wave acquisition means 100 and the converted bio-signal can also be used as learning data.

The algorithm derivation unit 220 serves to derive an algorithm for outputting a biosignal by receiving a pulse wave from the correlation between the pulse wave and the biosignal learned by the learning unit 210 .

The algorithm application unit 230 stores the algorithm derived from the algorithm derivation unit 220 , receives pulse wave data of the subject measured by the pulse wave acquisition means 100 , and converts it into a biosignal.

The computational fluid analysis unit 240 is a human body circulatory system model assuming emergency and disease situations that cannot be obtained from a clinical test or a data set consisting of pulse waves and biosignals that are actually measured in the learning data provided to the learning unit 210 . is implemented by a computer to measure the flow circulating inside the circulatory system model, and serves to prepare a pulse wave and biosignal data set to provide to the learning unit 210 . In the case of an actual emergency, it is not easy to accumulate clinical data because it is difficult to measure various data for patients in a situation where life is dealt with. Accordingly, the computational fluid analysis unit 240 acquires pulse waves and biosignals assumed to be blood flowing in the circulatory system through computer simulation. The data set composed of the pulse wave and the bio-signal prepared through the computational fluid analysis unit 240 includes the correlation between the pulse wave and the bio-signal derived from the learning unit 210 and the accuracy of the algorithm for deriving the bio-signal from the pulse wave. It serves to improve reliability.

In addition, it is desirable to increase the reliability of the flow characteristics of fluids circulating in the circulatory system by providing the circulatory system model with actual patient personal information such as the patient's condition, age, and weight as a boundary condition to the circulatory system model. At this time, it is preferable to complement the correlation between the pulse wave and the bio-signal learned in the learning unit with the computational fluid analysis (CFD) data by the computational fluid analysis unit 240 . Here, mutual complementation means that patients other than the pulse wave and biosignal datasets that can be easily acquired and accumulated as clinical data among the big-data of the clinical data accumulated by medical staff for many years, that is, the dataset consisting of pulse waves and biosignals, save lives. In order to solve the problem that it is not easy to obtain clinical data during an emergency situation, the pulse wave learned by the learning unit 200 is applied to the computational fluid analysis unit 240 by applying the assumed pulse wave and biosignal data set. A low-reliability learning result through the learning unit 200 from a pulse wave and biosignal dataset that can be easily acquired and accumulated with actual clinical data and complementary to the correlation between and biosignals and improving the reliability of the algorithm derived therefrom This means that the analyzed results can be complemented by applying the data group to the computational fluid analysis unit 240 for the data group showing .

As shown in FIG. 3 , the learning unit 210 of the non-contact biosignal measuring system 1000 of the present invention includes a main factor extracting unit 211 , a first learning model unit 212 and a second learning model unit 213 . Each part is described in detail as follows.

The major factor extraction unit 211 extracts major factors related to biosignals from pulse waves and pre-processes data. The main factor extractor 211 and the major factors related to the biosignals in the pulse wave will be dealt with in detail in the description of FIG. 4 .

The first learning model unit 212 learns the correlation between the main factor extracted from the pulse wave by the main factor extractor 211 and the biosignal. The first learning model unit 212 performs machine learning (learning) to infer a biosignal from a pulse wave using a Gaussian regression analysis model (Gaussian Processes). In the first learning model unit 212 of the present invention through the Gaussian regression analysis model (Gaussian Processes), learning is performed while satisfying the following equation.

Furthermore, the second learning model unit 213 learns to classify the biosignal region estimated from the main factor extracted from the pulse wave by the main factor extracting unit 211 . The second learning model unit 212 performs machine learning (learning) to classify which biosignal region the machine learning pulse wave corresponds to by using a support vector machine model. At this time, the basic concept of the support vector machine model is to generate a multidimensional surface called a hyperplane, and perform prediction based on the surface through the data rearranged through the kernel formula. The basic formula is as follows. The second learning model unit 213 of the present invention through the support vector machine model (Support Vector Machine) learning is performed while satisfying the following equation.

As shown in FIG. 4 , the characteristics of a pulse wave for obtaining a biosignal may be defined as follows. The content shown in FIG. 4 and the content to be described later are related to general pulse waves, which helps to understand the detailed description of the non-contact biosignal measurement system 1000 of the present invention, and the main factor extractor 211 according to the embodiment of the present invention. The purpose of this study is to limit the major factors in the process of extracting major factors related to biosignals from pulse waves.

Basically, the peak value (high point) of the pulse wave appears twice. The first peak value (Primary Pick) is the peak value that appears at the time of maximum blood pressure, and the second peak value (Secondary peak) is caused by the backward wave that occurs when blood is reflected from the entire blood vessel and rises upside down from the capillary. is the peak value. This second peak value is the same principle as ultrasonic waves hitting an object and returning. Here, the main factors of the pulse wave are variables inferred from the correlation between the first peak value and the second peak value, and it is important to analyze the backward wave because it changes according to the state of the blood vessel. However, since it is difficult to know the absolute value of the magnitude or time of the reflux wave alone, it is analyzed in comparison with the forward wave. That is, the main factors extracted from the pulse wave are values defined to calculate the magnitude or time of the reflected wave versus the traveling wave. There are a total of three main factors extracted from the pulse wave to implement the non-contact biosignal measurement system 1000 , which are as follows. First, the stiffness factor (Stiffness index (SI)), the second expansion factor (Augmentation index (AIx)), and finally the refractory factor (Reflectory index (RI)). All of the above three factors have been widely used in analyzing the state of the circulatory system. (Kim YW et al., Journal of Mechanical Science and Technology, 2016) In the case of SI, it is a factor used to predict the absolute value of blood pressure based on the pulse wave or to analyze whether there is a disease such as stenosis. (Matsui Y et al., Hypertension Research, 2004) In the case of AIx, it is an index mainly used to judge diseases related to patient stiffness. (Wilkinson IB et al., Journal of the American College of Cardiology, 2002) In the case of RI, it is a factor mainly used when analyzing drug administration effects and symptoms. (Manisty CH et al., Hypertension, 2009) Therefore, the non-contact biosignal measurement system 1000 of the present invention uses the main factor extraction unit 211 from the pulse wave input for learning as disclosed in the above-mentioned non-patent literature. The process of extracting the above three factors is performed. Thereafter, the correlation between the main factors and the biosignal of the first learning model 212 and the second learning model 213 is learned. Furthermore, the pulse wave data obtained from the subject is also extracted from the three factors and then applied to the algorithm derived through the correlation learned in the first learning model 212 and the second learning model 213 to derive biosignals. However, in case of emergency such as a disaster, there is a possibility that the subject's golden hour may be missed due to the response time delay of the non-contact biosignal measurement system 1000, leading to death. A response speed improvement mode in which biosignals are derived without using the main factor extractor 211 or a reliability improvement mode in which a biosignal is derived by more precisely diagnosing the condition of the subject using the main factor extracting unit 211, etc. Of course, it is possible to have an embodiment including

As shown in FIG. 5 , a process of acquiring a biosignal from a pulse wave of a subject using the non-contact biosignal measuring system 1000 of the present invention is as follows.

First, pulse wave data is acquired from the subject by using the pulse wave acquiring means 100 . Thereafter, the acquired pulse wave data is transmitted to the biosignal derivation unit 200 using an information communication network. Thereafter, the subject bio-signal information is output through an algorithm pre-stored in the bio-signal derivation unit 200 , and the output information is transmitted to the output terminal 300 . In this case, it is desirable to use an information communication network such as 5G with good responsiveness for information transmission. Furthermore, when the biosignal derivation unit 200 detects that the subject's estimated biosignal is out of the range of the prestored biosignal, a warning message is transmitted to the output terminal, and the warning message is related to the subject's severity, structure order, time limit, and It is preferable to include any one or more of the target hospitals.

As shown in FIG. 6 , the process of learning the correlation between the pulse wave and the biosignal using the non-contact biosignal measuring system 1000 of the present invention is as follows.

First, a data set composed of pre-stored pulse waves and bio-signals is provided to the learning unit 210 , to be precise, the main factor extracting unit 211 . In this case, it is preferable to use data measured in clinical trials or experiments as the pre-stored pulse waves and biosignals. Thereafter, the main factor extracting unit 211 extracts the main factors related to the biosignal from the pulse wave and then transmits the extracted main factors to the first learning model unit 212 and the second learning model unit 213 . At the same time, the computational fluid analysis unit 240 forms a model of the human circulatory system by computer simulation assuming an emergency situation that is difficult to measure or obtain in reality, and then presents a dataset consisting of pulse waves and biosignals that are computerized flow analysis (CFD). It is transmitted to the first learning model unit 212 and the second learning model unit 213 . Here, the first learning model unit and the second learning model unit perform machine learning to learn the correlation between the main factor extracted from the pulse wave and the biosignal and the correlation deriving the biosignal region estimated from the main factor. The first learning model unit performs machine learning by using a Gaussian regression analysis model (Gaussian Processes) and the second learning model unit using a support vector machine model (Support Vector Machine). Then, based on the derived correlation, the algorithm derivation unit 220 forms an algorithm for deriving a biosignal from the pulse wave, and transmits it to the algorithm application unit 230 . Ultimately, although not shown in FIG. 6 , the pulse wave data of the examinee obtained by the pulse wave acquisition means 100 is transferred to the main factor extractor 211 , and the pulse wave data is pre-processed, and then the algorithm application unit 230 derives it as a biosignal and outputs it This can be confirmed in the terminal 300 . Additionally, when a subject bio-signal out of the previously stored bio-signal range is found, a warning message is generated by the algorithm application unit 230 and transmitted to the output terminal 300 .

As shown in FIG. 7 , an embodiment of use of the non-contact biosignal measuring system 1000 of the present invention is as follows. When a subject (hereinafter referred to as a patient) occurs in a disaster or emergency situation, it is most important to measure the patient's condition. At this time, in order to diagnose the patient's condition, the patient's survival rate increases only when he or she quickly measures vital signs (body temperature, pulse respiration, blood pressure) and then takes appropriate emergency measures. However, in an emergency situation, the patient has several lesions on the body, and even contacting a measuring tool to measure biosignals can damage the patient's lesions and lead to tragic results. Accordingly, as shown in FIG. 7 , the pulse wave data is acquired from the patient by the pulse wave acquisition means 100 including an imaging device and transmitted to the biosignal derivation unit 200 . After that, the biosignal derivation unit 200 learns the correlation between the pulse wave and the biosignal through machine learning and outputs the patient's biosignal to the nearest emergency medical personnel and rescue personnel or the patient occurrence area through an algorithm derived from the output terminal 300, 7) can be delivered to those who have them. In addition, if the biosignal is out of the range of the stored biosignal based on the biosignal of the patient determined by the biosignal derivation unit 200, a warning message is transmitted to the output terminal 300 (not shown in FIG. 7), and the warning message is determined by the severity of the patient. , rescue ranking, limited time to life-threatening, and any one or more of target hospitals are output. Therefore, when a large-scale natural disaster occurs and the number of patients is large, there may be an embodiment in which the priority of the rescue of patients photographed by the pulse wave acquisition means 100 of the non-contact biosignal measurement system 1000 of the present invention is selected.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and detailed description will be given. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be

Unless defined otherwise, all terms used herein, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention as claimed in the claims.

1000: non-contact biosignal measurement system
100: pulse wave acquisition means
200: biosignal derivation unit
210: study department
211: main factor extraction unit
212: first learning model unit
213: second learning model unit
220: algorithm derivation unit
230: algorithm application unit
240: computational fluid analysis unit
300: output terminal
10: subject

Claims (10)

a pulse wave acquisition means for acquiring pulse wave data of the subject;
a biosignal derivation unit for deriving a correlation between the pulse wave data obtained from the pulse wave acquisition means and the biosignal with an algorithm using a machine learning technique, and applying the pulse wave data to the algorithm to derive the biosignal of the subject; and
and an output terminal for receiving and outputting the subject's biosignal derived from the biosignal derivation unit;
The biosignal deriving unit may include: a learning unit configured to receive a data set composed of pre-stored pulse waves and biosignals and learn a correlation between the pulse waves and biosignals through machine learning; and
and a computational fluid analysis unit that performs computational fluid analysis (CFD) by applying the data set consisting of the pulse wave and the biosignal to the virtual computer circulatory system model as a condition,
A non-contact biosignal measurement system, characterized in that the correlation between the pulse wave and the biosignal learned in the learning unit is complemented by the computational fluid analysis (CFD) data by the computational fluid analysis unit.
According to claim 1, wherein the biosignal derivation unit
an algorithm derivation unit for deriving an algorithm based on the correlation between the pulse wave and the biosignal learned by the learning unit; and
an algorithm application unit that receives the pulse wave data from the pulse wave acquisition means and derives a biosignal of a subject by applying the algorithm;
A non-contact biosignal measurement system comprising a.
delete The method of claim 2, wherein the learning unit
a major factor extraction unit pre-processing data to extract major factors from the pulse wave;
A first learning model unit for learning the correlation between the main factors and biosignals; and
A second learning model unit for classifying the area of the biosignal estimated from the main factor
A non-contact biosignal measurement system comprising a.
5. The method of claim 4, wherein the first learning model unit
Non-contact biosignal measurement system, characterized in that machine learning using Gaussian regression analysis model (Gaussian Processes).
5. The method of claim 4, wherein the second learning model unit
A non-contact biosignal measurement system characterized by machine learning using a support vector machine model.
The method of claim 2, wherein the algorithm application unit
When the derived bio-signal of the examinee is out of the range of the pre-stored bio-signal, a warning message is transmitted to the output terminal,
and the output terminal outputs the received warning message.
The method of claim 7, wherein the warning message is
Non-contact biosignal measurement system, characterized in that it includes any one or more of the subject's severity, rescue order, time limit, and target hospital.
The method according to claim 1, wherein the pulse wave acquiring means comprises:
A non-contact biosignal measurement system, characterized in that the terminal includes a photographing device capable of photographing the subject.
10. The method of claim 9, wherein the terminal
A non-contact biosignal measurement system, characterized in that it is any one of a drone, a smartphone, and a wearable device.

Images