System and method for monitoring patient risk by using bio-signal obtained in non-contact type
A non-contact biosignal monitoring system using radar and infrared sensors with machine learning models addresses errors and discomfort in vital sign monitoring, enabling continuous and accurate patient risk assessment and early emergency detection.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SARAMEUL IHAEHADA CO LTD
- Filing Date
- 2024-12-16
- Publication Date
- 2026-06-04
AI Technical Summary
Existing methods for monitoring vital signs in hospitalized patients are prone to errors due to nurse proficiency and equipment discomfort, and are limited by cost and suitability for general wards, necessitating a non-contact solution for continuous and accurate patient risk assessment.
A non-contact patient risk monitoring system using a biosignal acquisition device that measures heart rate, respiratory rate, body temperature, and estimates blood pressure and electrocardiogram through radar and infrared sensors, combined with machine learning models to predict severe disease risk.
Enables continuous, comfortable monitoring of patients, reducing errors and discomfort, and early detection of emergencies by transmitting alarm signals to external terminals.
Smart Images
Figure KR2024096999_04062026_PF_FP_ABST
Abstract
Description
Patient risk monitoring system and method using biosignals acquired by a non-contact method
[0001] The present invention relates to a patient risk monitoring system and method using biosignals acquired in a non-contact manner. More specifically, the invention relates to a patient risk monitoring system and method using biosignals acquired in a non-contact manner that acquires a patient's biosignals using a non-contact sensor and predicts the risk of severe disease in the patient based on the acquired biosignals.
[0002] For hospitalized patients, nurses perform rounds once every 0.5 to 4 hours and check vital signs (heart rate, respiratory rate, body temperature, blood pressure) and input them into the Electronic Medical Records (EMR).
[0003] To measure the vital signs of an hospitalized patient, nurses sequentially perform hand disinfection (washing hands or disinfecting with disinfectant), equipment preparation, patient identification (verifying patient registration number, name, and date of birth), temperature measurement, heart rate measurement, respiratory rate measurement, and blood pressure measurement, as well as completing the record sheet and entering the data into the electronic medical record. If at least one of the temperature, heart rate, respiratory rate, or blood pressure falls outside the normal range, the nurse must immediately report it to a physician or the person in charge. However, when a nurse directly measures heart rate and respiratory rate, errors may occur depending on the nurse's proficiency, and there is a possibility of typographical errors or omissions when entering the measured vital signs into the record sheet or electronic medical record.
[0004] In addition, depending on the patient's condition, an emergency such as cardiac arrest, heart failure, respiratory failure, or anaphylaxis may occur while a nurse is measuring the vital signs of the next patient, potentially leading to the patient's death; therefore, technology is required to automatically collect and monitor the patient's vital signs at least every minute.
[0005] However, medical devices that automatically collect and monitor patients' vital signs have limitations, such as causing significant discomfort to the patient because they are attached to specific parts of the body (e.g., fingertips, chest, etc.) and being difficult to use in general wards rather than intensive care units due to their high cost.
[0006] Therefore, there is a need for technology to collect and monitor patients' vital signs non-contactually.
[0007] The technology forming the background of the present invention is disclosed in Korean Registered Patent No. 10-1927705 (published Dec. 11, 2018).
[0008] As such, according to the present invention, a patient risk monitoring system and method using biosignals obtained in a non-contact manner are provided, which acquire a patient's biosignals using a non-contact sensor and predict the risk of severe disease in the patient based on the acquired biosignals.
[0009] According to an embodiment of the present invention for achieving such technical challenges, a method for monitoring patient risk using biosignals acquired in a non-contact manner comprises: a biosignal automatic collection device comprising the steps of: periodically irradiating a radar toward a patient; measuring the patient's heart rate and respiratory rate using radar signals reflected from the patient; measuring the patient's body temperature using an infrared body temperature sensor; estimating the patient's blood pressure using the heart rate; estimating the patient's electrocardiogram (ECG) and oxygen saturation (SpO₂) using the heart rate and blood pressure values; and predicting the risk of severe disease in the patient using the heart rate, respiratory rate, blood pressure, body temperature, ECG, and oxygen saturation. The method may further include the step of transmitting an alarm signal to an external terminal if the predicted risk is greater than a reference value.
[0010] The above-mentioned automatic biosignal collection device is installed on the ceiling or a part of the bed and can project a radar toward a patient on the bed.
[0011] The step of estimating the blood pressure of the patient may include the step of generating heart rate variability (HRV) data using the heart rate, and the step of inputting the heart rate and heart rate variability data into a first learning model to estimate the blood pressure of the patient.
[0012] The step of estimating the electrocardiogram and oxygen saturation of the patient may include the step of estimating the patient's electrocardiogram by inputting the heart rate and blood pressure values into a previously trained second learning model, and the step of estimating the patient's oxygen saturation by inputting the heart rate and blood pressure values into a previously trained third learning model.
[0013] The step of predicting the risk of heart disease in the patient can predict the risk of heart disease in the patient by inputting the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, oxygen saturation, and the patient's medical record information into a previously trained fourth learning model.
[0014] The medical record information of the above patient may include at least one of the patient's age, gender, medical history, changes in vital condition over time, surgical history, and surgical information within a pre-set first period.
[0015] The method may further include a step of determining the reference value by inputting the above medical record information into a previously trained fifth learning model.
[0016] The above fifth learning model can be trained so that the reference value is lowered as the age difference obtained by subtracting the preset reference age from the patient's age increases, or as vital signs deteriorate during the preset second period, or as the history of surgery exceeds the preset number of times, or as surgery is performed within the preset first period.
[0017] The above fifth learning model can be trained so that the reference value decreases as the distance between the biosignal automatic collection device and the patient increases beyond the reference distance.
[0018] The method may further include the step of estimating the blood pressure, electrocardiogram, and oxygen saturation when at least one of the measured patient's heart rate, respiratory rate, and body temperature falls outside the reference range.
[0019] A patient risk monitoring system using biosignals acquired in a non-contact manner according to another embodiment of the present invention comprises: a non-contact automatic biosignal acquisition device that periodically irradiates a radar toward a patient to measure the patient's heart rate and respiratory rate using the reflected radar signal, measures the patient's body temperature using an infrared body temperature sensor, estimates the patient's blood pressure using the heart rate, and estimates the patient's electrocardiogram (ECG) and oxygen saturation (SpO2) using the heart rate and blood pressure values; and a risk prediction server that receives the heart rate, respiratory rate, blood pressure, body temperature, ECG, and oxygen saturation from the automatic biosignal acquisition device and predicts the risk of severe disease in the patient using the heart rate, respiratory rate, blood pressure, body temperature, ECG, and oxygen saturation.
[0020] As such, according to the present invention, since the patient's vital signs are acquired in a non-contact manner, the patient's discomfort is relieved, and by continuously monitoring the patient's condition, an emergency situation can be detected early and an alarm signal can be transmitted to the outside.
[0021] FIG. 1 is a configuration diagram of a patient risk monitoring system using biosignals acquired in a non-contact manner according to one embodiment of the present invention.
[0022] FIG. 2 is a flowchart of a method for monitoring patient risk using biosignals obtained in a non-contact manner according to another embodiment of the present invention.
[0023] FIG. 3 is a diagram schematically illustrating a method for monitoring patient risk using biosignals obtained in a non-contact manner according to another embodiment of the present invention.
[0024] FIG. 4 is a diagram illustrating an example of predicting an early warning score and risk level according to another embodiment of the present invention.
[0025] Preferred embodiments according to the present invention will be described in detail below with reference to the attached drawings. In this process, the thickness of lines or the size of components shown in the drawings may be exaggerated for clarity and convenience of explanation.
[0026] Throughout the specification, when a part is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.
[0027] Furthermore, the terms described below are defined in consideration of their functions within the present invention, and these may vary depending on the intent or practice of the user or operator. Therefore, the definitions of these terms should be based on the content throughout this specification.
[0028] In the embodiments described below, a patient risk monitoring system using biosignals acquired by a non-contact method is explained with a specific example of monitoring the risk of heart disease in an inpatient hospitalized in a hospital. However, the present invention is not limited thereto and can monitor the risk of heart disease, cerebrovascular disease, or respiratory disease in inpatient hospitalized in a hospital or in patients with chronic diseases recuperating at home.
[0029] FIG. 1 is a configuration diagram of a patient risk monitoring system using biosignals acquired in a non-contact manner according to one embodiment of the present invention.
[0030] As illustrated in FIG. 1, a patient risk monitoring system using biosignals acquired in a non-contact manner may include an automatic biosignal collection device (100) and a risk prediction server (200).
[0031] First, the biosignal automatic collection device (100) can periodically irradiate a radar toward a patient in a non-contact manner to measure the patient's heart rate (bpm) and respiration rate (breaths / min) using the reflected radar signal, measure the patient's body temperature using an infrared body temperature sensor, estimate the patient's blood pressure using the heart rate, and estimate the patient's electric cardiographic (ECG) and oxygen saturation (SpO₂(%)) using the heart rate and blood pressure values. At this time, the biosignal automatic collection device (100) can be installed on the ceiling of a hospital room or a part of the bed to irradiate the radar toward the patient on the bed.
[0032] Additionally, the biosignal collection device (100) may include a radar sensor unit (110), a heart rate and respiratory rate measuring unit (120), a body temperature measuring unit (130), a blood pressure estimating unit (140), an electrocardiogram estimating unit (150), an oxygen saturation estimating unit (160), and a communication unit (170).
[0033] Specifically, the radar sensor unit (110) can periodically irradiate the radar toward the patient (e.g., every minute) and receive the radar signal reflected from the patient.
[0034] Additionally, the heart rate and respiratory rate measuring unit (120) can measure the patient's heart rate and respiratory rate using the received radar signal. At this time, since the technique for measuring the patient's heart rate and respiratory rate based on the radar signal reflected from the patient is the same as that previously disclosed, a detailed description is omitted.
[0035] Additionally, the body temperature measuring unit (130) is implemented in the form of an infrared body temperature sensor, and the patient's body temperature can be measured through the infrared body temperature sensor. At this time, since the technology for measuring the patient's body temperature through the infrared body temperature sensor is the same as that previously disclosed, a detailed description is omitted.
[0036] Additionally, the blood pressure estimation unit (140) can generate heart rate variability (HRV) data using the measured heart rate and input the heart rate and heart rate variability data into a first learning model that has been trained to estimate the patient's blood pressure. At this time, heart rate variability is a calculated value calculated using three beats as the time difference between consecutive heart beat intervals, and the first learning model is a model trained to estimate a blood pressure value when it receives a heart rate and the corresponding heart rate variability data, using multiple heart rates, heart rate variability data, and blood pressure values (or systolic blood pressure values) as training data.
[0037] To elaborate, the blood pressure estimation unit (140) generates heart rate variability data representing changes in the heart rate time series using the interval between heart beats based on the measured heart rate, and inputs the generated heart rate and heart rate variability data into a first learned model to estimate the patient's systolic blood pressure (Systolic BP(mmHg)).
[0038] Additionally, the electrocardiogram estimation unit (150) can estimate the patient's electrocardiogram by inputting heart rate and blood pressure values into a second learning model that has been previously learned. Here, the second learning model is a model trained to estimate and derive an electrocardiogram graph when the patient's heart rate and blood pressure values are input, using multiple heart rates, corresponding blood pressure values, and electrocardiogram graphs as learning data.
[0039] Additionally, the oxygen saturation estimation unit (160) can estimate the patient's oxygen saturation by inputting heart rate and blood pressure values into a previously trained third learning model. Here, the third learning model is a model trained to derive oxygen saturation when the patient's heart rate and blood pressure values are input, using multiple heart rates and corresponding blood pressure and oxygen saturation values as learning data.
[0040] At this time, the biosignal automatic collection device (100) may not estimate blood pressure, electrocardiogram, and oxygen saturation when the measured heart rate, respiratory rate, and body temperature are within the normal range.
[0041] As a result, the biosignal automatic collection device (100) can reduce the power consumption used for monitoring the patient's risk level.
[0042] In other words, the biosignal automatic collection device (100) can estimate blood pressure, electrocardiogram, and oxygen saturation when at least one of the measured patient's heart rate, respiratory rate, and body temperature is outside the reference range.
[0043] Additionally, the communication unit (170) can transmit heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation to the risk prediction server (200).
[0044] Next, the risk prediction server (200) receives heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation from the biosignal automatic collection device (100), and can predict the risk of severe disease in the patient using the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation.
[0045] Specifically, the risk prediction server (200) can predict the risk of heart disease in a patient by inputting heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram and oxygen saturation and the patient's medical record information into a previously trained fourth learning model. Here, the patient's medical record information includes at least one of the patient's age, gender, medical history, changes in vital signs over time (e.g., patient's vital sign data from the last time the patient was admitted until the present), the patient's surgical history, and surgical information within a pre-set first period (e.g., the last 3 years) of the patient (e.g., surgical site, time of surgery, post-operative prescription history, specific details, etc.) stored on a server (e.g., a Hospital Information System (HIS) or a server of the National Health Insurance Service). The fourth learning model is a model trained to predict the patient's risk of heart disease based on the patient's medical record information, including multiple heart rates, respiratory rates, blood pressure values, body temperature, electrocardiogram, oxygen saturation, and corresponding patient medical record information, using the risk of heart disease as learning data.
[0046] In addition, the risk prediction server (200) can transmit an alarm signal to an external terminal (e.g., guardian terminal, medical staff (nurse, doctor, nursing assistant, etc.) terminal, hospital terminal, rescue facility terminal, etc.) when the predicted risk level is greater than the reference value.
[0047] Here, the risk prediction server (200) can determine a reference value by inputting the patient's medical record information into a previously trained fifth learning model.
[0048] At this time, the fifth learning model is a model trained by setting the patient's medical record information, including at least one of the patient's age, gender, medical history, changes in vital signs over time (e.g., the patient's vital sign data from the last time the patient was admitted until now), the patient's surgical history, and surgical information within the patient's pre-set first period (e.g., the last 3 years) (e.g., surgical site, time of surgery, post-operative prescription history, specific details, etc.), as input data and setting the risk level as output data. The fifth learning model may have a lower reference value if the age difference obtained by subtracting the pre-set reference age (e.g., 60 years) from the patient's age is greater, or if changes in vital signs are worsening during the pre-set second period (e.g., from the time of admission until now), or if the patient's surgical history is greater than the pre-set number of times (e.g., 2 times), or if the patient has undergone surgery at least once within the pre-set first period (e.g., within one month, last 3 years).
[0049] In other words, the fifth learning model can be trained to lower the threshold value as the patient ages, as changes in vital signs over time worsen, as the patient has a history of surgeries, or as they have recently undergone surgery.
[0050] In addition, the fifth learning model can be trained so that the reference value decreases as the distance between the biosignal automatic collection device (100) and the patient increases beyond the reference distance.
[0051] The first to fifth learning models are learning models based on supervised learning and may include machine learning, neural networks, deep learning, LSTM (Long Short-term Memory) models, and DNN (Deep Neural Network) models.
[0052] Below, a method for monitoring patient risk using biosignals acquired in a non-contact manner is explained in more detail using FIGS. 2 to 4.
[0053] FIG. 2 is a flowchart of a method for monitoring patient risk using biosignals obtained in a non-contact manner according to another embodiment of the present invention, and FIG. 3 is a diagram schematically illustrating a method for monitoring patient risk using biosignals obtained in a non-contact manner according to another embodiment of the present invention.
[0054] As illustrated in FIGS. 2 and 3, the radar sensor unit (110) can periodically irradiate the radar toward the patient (e.g., every minute) (S210).
[0055] Specifically, the radar sensor unit (110) can periodically irradiate the radar toward the patient and receive the radar signal reflected from the patient.
[0056] Next, the heart rate and respiratory rate measuring unit (120) can measure the patient's heart rate and respiratory rate using a radar signal reflected from the patient (S220). At this time, since the technique for measuring the patient's heart rate and respiratory rate based on the radar signal reflected from the patient is the same as that previously disclosed, a detailed explanation is omitted.
[0057] Next, the body temperature measuring unit (130) can measure the patient's body temperature through an infrared body temperature sensor (S230). At this time, since the technique for measuring the patient's body temperature through an infrared body temperature sensor is the same as that previously disclosed, a detailed explanation is omitted.
[0058] Next, the blood pressure estimation unit (140) estimates the patient's blood pressure using the measured heart rate (S240).
[0059] That is, by using the radar sensor unit (110) to measure heart rate and respiratory rate, blood pressure can be estimated in real time according to heart rate and respiratory rate.
[0060] To elaborate, the blood pressure estimation unit (140) can generate heart rate variability data using the measured heart rate (S241). Here, the blood pressure estimation unit (140) can generate heart rate variability data representing changes in the heart rate time series using the interval between heart beats based on the measured heart rate. At this time, the heart rate variability is a calculated value calculated using the number of beats as the time difference of consecutive heart beat intervals.
[0061] Additionally, the blood pressure estimation unit (140) can estimate the patient's blood pressure by inputting the generated heart rate and heart rate variability data into a first learning model that has been trained (S242). Here, the first learning model is a model trained to estimate a blood pressure value when it receives a heart rate and corresponding heart rate variability data, using multiple heart rates, heart rate variability data, and blood pressure values (systolic blood pressure values and diastolic blood pressure values) as training data.
[0062] Here, Heart Rate Variability (HRV) is an indicator representing the change in the interval between heartbeats and reflects the activity of the autonomic nervous system.
[0063] Next, the biosignal automatic collection device (100) can estimate the patient's electrocardiogram and oxygen saturation using heart rate and blood pressure values (S250).
[0064] Specifically, the electrocardiogram estimation unit (150) can estimate the patient's electrocardiogram by inputting heart rate and blood pressure values into a second learning model that has been previously learned (S251). Here, the second learning model is a model trained to estimate and derive an electrocardiogram graph when the patient's heart rate and blood pressure values are input, using multiple heart rates, corresponding blood pressure values, and an electrocardiogram graph as learning data.
[0065] Additionally, the oxygen saturation estimation unit (160) can estimate the patient's oxygen saturation by inputting the heart rate and blood pressure values into a previously trained third learning model (S252). Here, the third learning model is a model trained to derive oxygen saturation when the patient's heart rate and blood pressure values are input, using multiple heart rates and corresponding blood pressure values and oxygen saturation as learning data.
[0066] At this time, the biosignal automatic collection device (100) may not estimate blood pressure, electrocardiogram, and oxygen saturation when the measured heart rate, respiratory rate, and body temperature are within the normal range.
[0067] As a result, the biosignal automatic collection device (100) can reduce the power consumption used for monitoring the patient's risk level and reduce the amount of communication between the biosignal automatic collection device (100) and the risk prediction server (200).
[0068] In other words, the biosignal automatic collection device (100) can estimate blood pressure, electrocardiogram, and oxygen saturation when at least one of the measured patient's heart rate, respiratory rate, and body temperature is outside the reference range.
[0069] Next, the risk prediction server (200) receives heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation from the biosignal automatic collection device (100), and predicts the risk of severe disease in the patient using heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation (S260).
[0070] Specifically, the risk prediction server (200) can predict the risk of heart disease in a patient by inputting heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram and oxygen saturation and the patient's medical record information into a previously trained fourth learning model. Here, the patient's medical record information includes at least one of the patient's age, gender, medical history, changes in vital signs over time (e.g., patient's vital sign data from the last time the patient was admitted until the present), surgical history, and recent surgical information (e.g., surgical site, time of surgery, post-operative prescription details, specific details, etc.) stored in a server (e.g., a hospital information system or a server of the National Health Insurance Service).
[0071] The fourth learning model is a model trained to use multiple heart rates, respiratory rates, blood pressure values, body temperature, electrocardiogram, and oxygen saturation, along with corresponding patient medical record information and risk levels for heart disease, as training data to calculate an early warning score based on the heart rate, respiratory rate, blood pressure values, body temperature, electrocardiogram, oxygen saturation, and corresponding patient medical record information, and to predict the patient's risk of heart disease according to the calculated early warning score.
[0072] FIG. 4 is a diagram illustrating an example of predicting an early warning score and risk level according to another embodiment of the present invention.
[0073] As shown in Fig. 4, the fourth learning model can calculate an early warning score by assigning scores to each of the respiratory rate, oxygen saturation, body temperature, systolic blood pressure, heart rate, and the patient's level of consciousness (e.g., AVPU(Alert, response to verbal order, response to pain, unresponse)) and summing them.
[0074] According to one embodiment of the present invention, in the case of respiratory rate, the fourth learning model can be trained to assign 3 points if the respiratory rate exceeds 35 breaths per minute or is less than 7 breaths per minute, assign 2 points if the respiratory rate is 31 breaths per minute or more and 35 breaths per minute or less, assign 1 point if the respiratory rate is 21 breaths per minute or more and 30 breaths per minute or less, and assign 0 points if the respiratory rate is 9 breaths per minute or more and 20 breaths per minute or less.
[0075] According to one embodiment of the present invention, in the case of oxygen saturation, the fourth learning model may assign 3 points if the oxygen saturation is less than 85%, assign 2 points if the oxygen saturation is 85% or more and 89% or less, assign 1 point if the oxygen saturation is 90% or more and 92% or less, and assign 0 points if the oxygen saturation is greater than 92%.
[0076] According to one embodiment of the present invention, in the case of body temperature, the fourth learning model can be trained to assign 3 points if the body temperature is less than 34℃, assign 2 points if the body temperature is greater than 38.9℃ or is 34℃ or higher and 34.9℃ or lower, assign 1 point if the body temperature is 38℃ or higher and 38.9℃ or lower or 35℃ or higher and 35.9℃ or lower, and assign 0 points if the body temperature is 36℃ or higher and 37.9℃ or lower.
[0077] According to one embodiment of the present invention, in the case of a systolic blood pressure value, the fourth learning model can be trained to assign 3 points if the systolic blood pressure value is less than 70 mmHg, assign 2 points if the systolic blood pressure value is greater than 199 mmHg or is 70 mmHg or higher and 79 mmHg or lower, assign 1 point if the systolic blood pressure value is 80 mmHg or higher and 99 mmHg or lower, and assign 0 points if the systolic blood pressure value is 100 mmHg or higher and 199 mmHg or lower.
[0078] According to one embodiment of the present invention, in the case of heart rate, the fourth learning model may assign 3 points if the heart rate exceeds 129 beats per minute or is less than 30 beats per minute, assign 2 points if the heart rate is 110 beats per minute or more and 129 beats per minute or 30 beats per minute or more and 39 beats per minute or less, assign 1 point if the heart rate is 100 beats per minute or more and 109 beats per minute or 40 beats per minute or more and 49 beats per minute or less, and assign 0 points if the heart rate is 50 beats per minute or more and 99 beats per minute or less.
[0079] According to one embodiment of the present invention, regarding the patient's level of consciousness, the fourth learning model may assign 0 points when the patient's level of consciousness is alert, 1 point when the patient's level of consciousness responds to verbal instructions, 2 points when the patient's level of consciousness responds to pain stimuli, and 3 points when the patient's level of consciousness is unresponsive to pain stimuli.
[0080] In addition, the fourth learning model may be a model trained to calculate the risk level based on the calculated early warning score.
[0081] According to one embodiment of the present invention, the fourth learning model can predict the risk level as 80 when the calculated early warning score is 5.
[0082] Next, the risk prediction server (200) can determine a reference value by inputting the patient's medical record information into a previously trained fifth learning model (S270).
[0083] At this time, the fifth learning model is a model trained by setting the patient's medical record information, including at least one of the patient's age, gender, medical history, changes in vital signs over time (e.g., patient's vital sign data from the last time the patient was admitted until now), the patient's surgical history, and surgical information within the patient's pre-set first period (e.g., the last 3 years) (e.g., surgical site, time of surgery, post-operative prescription history, specific details, etc.), as input data and setting the risk level as output data. The fifth learning model may have a lower reference value if the age difference obtained by subtracting the pre-set reference age (e.g., 60 years) from the patient's age is greater, if changes in vital signs are worsening during the pre-set second period (e.g., from the time of admission until now), if the patient's surgical history is greater than the pre-set number of times (e.g., 2 times), or if the patient has undergone surgery at least once within the pre-set first period (e.g., the last 3 years).
[0084] According to one embodiment of the present invention, when the patient's age is 78 years, the risk prediction server (200) can input the patient's medical record information into a previously trained fifth learning model to determine a reference value of 71 points.
[0085] According to one embodiment of the present invention, if the patient is 25 years old and has had surgery within the last 3 months, the risk prediction server (200) can input the patient's medical record information into a previously trained fifth learning model to determine a reference value of 77 points.
[0086] According to one embodiment of the present invention, if the patient is 30 years old, has good changes in vital condition, and has no history of surgery, the risk prediction server (200) can input the patient's medical record information into a previously trained fifth learning model to determine a reference value of 85 points.
[0087] In other words, the fifth learning model can be trained to lower the threshold value as the patient ages, as changes in vital signs over time worsen, as the patient has a history of surgeries, or as they have recently undergone surgery.
[0088] As such, according to an embodiment of the present invention, the standard value for severity may not have a consistent judgment standard value, but may change by considering various variables such as age and symptoms.
[0089] In addition, the fifth learning model can be trained so that the reference value decreases as the distance between the biosignal automatic collection device (100) and the patient increases beyond the reference distance.
[0090] That is, if the distance between the biosignal automatic collection device (100) attached to the ceiling or the top of the bed and the patient is greater than the standard distance (e.g., 1.4m), the accuracy of the measurement is bound to decrease. Therefore, to prepare for dangerous situations, the fifth learning model can be trained so that the standard value decreases as the distance between the biosignal automatic collection device (100) and the patient increases.
[0091] Here, the distance between the biosignal automatic collection device (100) and the patient can be measured through the radar sensor unit (110).
[0092] Therefore, in the case of a hospital room with a high ceiling, the installation height of the biosignal automatic collection device (100) attached to the ceiling is high, so the distance from the patient is inevitably far, making it impossible to accurately measure the patient's heart rate and respiratory rate. Accordingly, to prevent the alarm signal from not being generated despite the high risk, the fifth learning model is trained so that the reference value decreases as the distance from the biosignal automatic collection device (100) to the patient increases beyond the reference distance.
[0093] Next, the risk prediction server (200) can transmit an alarm signal to an external terminal (e.g., guardian terminal, medical staff (nurse, doctor, nursing assistant, etc.) terminal, hospital terminal, rescue facility terminal, etc.) when the predicted risk level is greater than the reference value (S280).
[0094] According to one embodiment of the present invention, a risk prediction server (200) can transmit an alarm signal to an external terminal (e.g., guardian terminal, medical staff (nurse, doctor, nursing assistant, etc.) terminal, hospital terminal, rescue facility terminal, etc.) when the predicted risk level is greater than a reference value (e.g., 80 points). At this time, the risk prediction server (200) can request the medical staff (e.g., doctor, nurse, nursing assistant, etc.) to increase the frequency of observation of the patient compared to before.
[0095] Additionally, the risk prediction server (200) can shorten the radar irradiation cycle of the radar sensor unit (110) through the communication unit (170) when the predicted risk level is greater than a reference value (e.g., 80 points).
[0096] Accordingly, the present invention can assist in the prompt detection of emergency situations by adjusting the frequency of patient observation differently according to the patient's condition.
[0097] According to an embodiment of the present invention, since the patient's vital signs are acquired in a non-contact manner, the patient's discomfort is relieved, and by continuously monitoring the patient's condition, an emergency situation can be detected early and an alarm signal can be transmitted to the outside.
[0098] The present invention has been described with reference to the embodiments illustrated in the drawings, but this is merely illustrative, and those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be determined by the technical spirit of the following claims.
[0099] [Explanation of the symbol]
[0100] 100: Automatic biosignal acquisition device
[0101] 110: Radar sensor unit
[0102] 120: Heart rate and respiratory rate measurement unit
[0103] 130: Body temperature measuring unit
[0104] 140: Blood pressure estimation section
[0105] 150: ECG Estimation Section
[0106] 160: Oxygen saturation estimation section
[0107] 170: Communications Department
[0108] 200: Risk Prediction Server
Claims
1. A method for monitoring patient risk using biosignals acquired by a non-contact method, The biosignal automatic acquisition device includes the step of periodically irradiating the patient with radar, A step of measuring the heart rate and respiratory rate of the patient using a radar signal reflected from the patient, A step of measuring the body temperature of the patient using an infrared body temperature measuring sensor, A step of estimating the patient's blood pressure using the heart rate, A step of estimating the patient's electrocardiogram (ECG) and oxygen saturation (SpO₂) using the heart rate and blood pressure values, and A patient risk monitoring method comprising the step of predicting the risk of severe disease in the patient using the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation.
2. In Paragraph 1, A patient risk monitoring method comprising the step of transmitting an alarm signal to an external terminal if the predicted risk level is greater than a reference value.
3. In Paragraph 1, The above biosignal automatic collection device is installed on the ceiling or part of the bed, and is a patient risk monitoring method that irradiates radar toward a patient on the bed.
4. In Paragraph 1, The step of estimating the blood pressure of the above patient is, A step of generating heart rate variability (HRV) data using the above heart rate, and A patient risk monitoring method comprising the step of estimating the patient's blood pressure by inputting the heart rate and heart rate variability data into a first learning model.
5. In Paragraph 1, The step of estimating the electrocardiogram and oxygen saturation of the above patient is, A step of estimating the patient's electrocardiogram by inputting the heart rate and blood pressure values into a previously trained second learning model, and A patient risk monitoring method comprising the step of estimating the patient's oxygen saturation by inputting the heart rate and blood pressure values into a previously trained third learning model.
6. In Paragraph 2, The step of predicting the risk of heart disease for the above patient is, A patient risk monitoring method for predicting the risk of heart disease in a patient by inputting the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation and the patient's medical record information into a previously trained fourth learning model.
7. In Paragraph 6, The medical record information of the above patient is, A patient risk monitoring method comprising at least one of the patient's age, gender, medical history, changes in vital status over time, surgical history, and surgical information within a pre-set first period.
8. In Paragraph 7, A patient risk monitoring method further comprising the step of determining the reference value by inputting the above medical record information into a previously trained fifth learning model.
9. In Paragraph 8, The above-mentioned fifth learning model is a patient risk monitoring method in which the reference value is learned to decrease as the age difference obtained by subtracting a preset reference age from the patient's age increases, or as vital signs deteriorate during a preset second period, or as the history of surgery exceeds a preset number, or as surgery is performed within a preset first period.
10. In Paragraph 8, The above-mentioned fifth learning model is, A patient risk monitoring method in which the above biosignal automatic collection device learns to lower the reference value as the distance from the patient increases beyond the reference distance.
11. In Paragraph 1, A patient risk monitoring method further comprising the step of estimating blood pressure, electrocardiogram, and oxygen saturation when at least one of the measured patient's heart rate, respiratory rate, and body temperature is outside the reference range.
12. In a patient risk monitoring system using biosignals acquired by a non-contact method, A non-contact automatic biosignal acquisition device that periodically irradiates a radar toward a patient to measure the patient's heart rate and respiratory rate using the reflected radar signal, measures the patient's body temperature using an infrared body temperature sensor, estimates the patient's blood pressure using the heart rate, and estimates the patient's electrocardiogram (ECG) and oxygen saturation (SpO2) using the heart rate and blood pressure values; A patient risk monitoring system comprising a risk prediction server that receives heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation from an automatic biosignal collection device, and predicts the risk of severe disease in the patient using the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation.
13. In Paragraph 12, The above-mentioned risk prediction server is, A patient risk monitoring system that transmits an alarm signal to an external terminal if the predicted risk level is greater than a threshold value.
14. In Paragraph 12, The above-mentioned automatic biosignal collection device is installed on the ceiling or part of the bed and is a patient risk monitoring system that irradiates radar toward a patient on the bed.
15. In Paragraph 12, The above biosignal automatic collection device is, A radar sensor unit that periodically irradiates radar toward the patient and receives radar signals reflected from the patient, A heart rate and respiratory rate measuring unit that measures the heart rate and respiratory rate of the patient using the above radar signal, A temperature measuring unit that is implemented in the form of a radar body temperature sensor and measures the body temperature of the patient, A blood pressure estimation unit that generates heart rate variability (HRV) data using the heart rate and inputs the heart rate and heart rate variability data into a pre-trained first learning model to estimate the patient's blood pressure, An electrocardiogram estimation unit that estimates the patient's electrocardiogram by inputting the heart rate and blood pressure values into a previously trained second learning model, An oxygen saturation estimation unit that estimates the oxygen saturation of the patient by inputting the above heart rate and blood pressure values into a previously trained third learning model, and A patient risk monitoring system comprising a communication unit that transmits the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation to the risk prediction server.
16. In Paragraph 13, The above-mentioned risk prediction server is, A patient risk monitoring system that predicts the risk of heart disease in the patient by inputting the heart rate, respiratory rate, blood pressure, body temperature, electrocardiogram, and oxygen saturation and the patient's medical record information into a previously trained fourth learning model.
17. In Paragraph 16, The medical record information of the above patient is, A patient risk monitoring system comprising at least one of the patient's age, gender, medical history, changes in vital status over time, surgical history, and surgical information within a pre-set first period.
18. In Paragraph 17, The above-mentioned risk prediction server is, A patient risk monitoring system that determines the reference value by inputting the above medical record information into a previously trained fifth learning model.
19. In Paragraph 18, The above-mentioned fifth learning model is a patient risk monitoring system that learns to lower the reference value as the age difference obtained by subtracting a preset reference age from the patient's age increases, or as vital signs deteriorate during a preset second period, or as the history of surgery exceeds a preset number, or as surgery is performed within a preset first period.
20. In Paragraph 18, The above-mentioned fifth learning model is, A patient risk monitoring system in which the above biosignal automatic collection device learns to lower the reference value as the distance from the patient increases beyond the reference distance.
21. In Paragraph 12, The above biosignal automatic collection device is, A patient risk monitoring system that estimates blood pressure, electrocardiogram, and oxygen saturation when at least one of the measured patient's heart rate, respiratory rate, and body temperature is outside the reference range.