Eicu patient risk dynamic grading early warning system based on multi-parameter fusion

Abstract

This invention relates to the field of personal health risk assessment technology, specifically to a dynamic risk grading and early warning system for EICU patients based on multi-parameter fusion. The system includes a data acquisition module and an early warning module. The data acquisition module acquires ECG, IBP, and PPG signals from EICU patients. The early warning module obtains a vascular stiffness rate index from the time interval between the central pulse initiation time and the peripheral pulse arrival time. It obtains pulsatility transmission efficiency based on the PPG signal between two adjacent peripheral pulse arrival times and the IBP signal between two adjacent central pulse initiation times. This yields stiffness deviation and transmission efficiency deviation, respectively. The system analyzes the complexity deviation of stiffness deviation, transmission efficiency deviation, and PPG signal within a preset reference time period to grade and warn of EICU patient risk, overcoming the limitations of single parameters and improving the accuracy of EICU patient risk warnings.

Description

Technical Field

[0001] This invention relates to the field of personal health risk assessment technology, specifically to an EICU patient risk dynamic grading and early warning system based on multi-parameter fusion. Background Technology

[0002] In emergency intensive care unit (EICU) clinical practice, accurate assessment of a patient's circulatory status is crucial for early identification of shock risk. The human circulatory system possesses a robust compensatory mechanism. In the early stages of shock (the compensatory phase), the body activates the sympathetic nervous system, causing peripheral vasoconstriction and increasing vascular resistance, thereby temporarily maintaining central blood pressure within the normal range despite a decrease in cardiac output. This illusion of "normal blood pressure" often masks the underlying pathological reality of tissue hypoperfusion.

[0003] Current monitoring techniques primarily rely on absolute values ​​of invasive arterial pressure (IBP) or a single linear perfusion index (PI) for assessment. However, this monitoring model faces a challenge: how to distinguish between "pathological compensation" and "physiological stress." Because physiological stress states such as pain, chills, anxiety, or postoperative recovery agitation can also cause strong sympathetic nerve excitation, leading patients to exhibit vasoconstriction (high resistance) characteristics highly similar to those in the early stages of shock.

[0004] In this context of high vascular tension, traditional monitoring methods struggle to decouple neural regulatory activity from vascular mechanical responses. Simple blood pressure monitoring cannot identify hypoperfusion, and a single drop in the perfusion index cannot distinguish between microcirculatory failure caused by shock and transient vasospasm caused by pain. Such monitoring methods can lead to misinterpreting pain as shock and performing unnecessary fluid resuscitation, or misinterpreting occult shock as a stress response, thus affecting the accuracy of risk warnings for EICU patients. Summary of the Invention

[0005] To address the technical problem of low accuracy in existing EICU patient risk warning systems, the present invention aims to provide a dynamic risk grading and early warning system for EICU patients based on multi-parameter fusion. The specific technical solution adopted is as follows: This invention provides a dynamic risk grading and early warning system for EICU patients based on multi-parameter fusion, comprising: The data acquisition module is used to acquire the electrocardiogram signal, invasive arterial pressure signal, and photoplethysmography signal of the EICU patient during the current sampling cycle. The early warning module, electrically connected to the data acquisition module, is configured as follows: The vascular stiffness rate index is obtained from the time interval between the start time of the central pulse and the arrival time of the peripheral pulse in the current sampling period. The pulsational transmission efficiency is obtained by considering the accumulation of photoplethysmography (PPG) signals between the arrival times of two adjacent peripheral pulses and the accumulation of invasive arterial pressure signals relative to a preset invasive arterial pressure baseline between the initiation times of two adjacent central pulses. Determine the degree of stiffness deviation of the vascular stiffness rate index relative to the stiffness baseline of the current sampling period, and the degree of transmission efficiency deviation of the pulsation transmission efficiency relative to the transmission efficiency baseline of the current sampling period; Determine the degree of complexity deviation of the photoplethysmography (PPG) signal within a preset reference time period; By analyzing the deviations in hardness, transmission efficiency, and complexity, a risk classification and early warning system for EICU patients can be implemented.

[0006] In an exemplary embodiment, the process of obtaining the central pulse initiation time and the peripheral pulse arrival time includes: Determine the heartbeat trigger time of the electrocardiogram signal in the current sampling period; The minimum point of the invasive arterial pressure signal is searched within the first time interval, and the time corresponding to the minimum point of the invasive arterial pressure signal is the time when the central pulse begins; the first time interval is a time interval of a first preset duration after the time when the heartbeat is triggered. The minimum point of the photoplethysmography signal is searched within the second time interval, and the time corresponding to the minimum point of the photoplethysmography signal is the time when the peripheral pulse arrives; the second time interval is a time interval of a second preset duration starting from the time when the central pulse begins.

[0007] In an exemplary embodiment, the process of obtaining the pulsational transmission energy efficiency includes: Determine the standardized signal of the photoplethysmography (PPG) signal; The standardized signal between the arrival times of two adjacent peripheral pulses is integrated to obtain the first integral result; Determine the signal difference between the invasive arterial pressure signal and the preset invasive arterial pressure baseline at each time point between the onset of two adjacent central pulses; The signal differences between the start times of two adjacent central pulses are integrated to obtain a second integral result; The pulsational transmission efficiency is obtained from the first integral result and the second integral result; the pulsational transmission efficiency is positively correlated with the first integral result and negatively correlated with the second integral result.

[0008] In an exemplary embodiment, the process of acquiring the standardized signal includes: determining the DC component and AC component of the photoplethysmography (PPG) signal, calculating the ratio of the AC component to the DC component, and obtaining the standardized signal.

[0009] In an exemplary embodiment, the process of obtaining the hardness baseline and transfer energy efficiency baseline for the current sampling period includes: Determine whether the risk of EICU patients in the previous sampling cycle is safe in the current sampling cycle; If the condition is safe, update the hardness baseline of the previous sampling period to obtain the hardness baseline of the current sampling period; update the transfer energy efficiency baseline of the previous sampling period to obtain the transfer energy efficiency baseline of the current sampling period.

[0010] In one exemplary embodiment, the hardness baseline includes a hardness average and a hardness standard deviation, and the energy transfer efficiency baseline includes an energy transfer efficiency average and an energy transfer efficiency standard deviation. The process of updating the hardness baseline of the previous sampling period to obtain the hardness baseline of the current sampling period includes: Determine the weighting coefficients of the average hardness value of the previous sampling period and the vascular hardness rate index of the current sampling period, and perform a weighted summation of the average hardness value of the previous sampling period and the vascular hardness rate index of the current sampling period to obtain the average hardness value of the current sampling period. The difference between the vascular stiffness rate index of the current sampling period and the average stiffness of the current sampling period is determined. The weighting coefficient of the square of the difference and the stiffness variance of the previous sampling period is determined. The square of the difference and the stiffness variance of the previous sampling period are weighted and summed to obtain the stiffness variance of the current sampling period. The stiffness standard deviation of the current sampling period is obtained from the stiffness variance of the current sampling period.

[0011] In an exemplary embodiment, the process of obtaining the hardness deviation includes: The hardness deviation is obtained by taking the ratio of the difference to the hardness standard deviation of the current sampling period. The process of obtaining the transmission performance deviation includes: The difference between the pulsational transmission efficiency of the current sampling period and the average transmission efficiency of the current sampling period is used as the ratio to the standard deviation of the transmission efficiency of the current sampling period to obtain the transmission efficiency deviation.

[0012] In one exemplary embodiment, the complexity is the sample entropy of the photoplethysmography signal within a preset reference time period.

[0013] In one exemplary embodiment, the risk classification and early warning for EICU patients includes: The comprehensive deviation characteristics are determined based on the hardness deviation and the transmission efficiency deviation. Compare the overall deviation feature with the magnitude of the first preset deviation threshold; When the overall deviation feature is greater than the first preset deviation threshold, the risk level of the EICU patient is determined based on the sign of the hardness deviation and the transmission efficiency deviation, as well as the magnitude of the complexity deviation and the second preset deviation threshold.

[0014] In one exemplary embodiment, determining the risk level of an EICU patient includes: If the hardness deviation is positive, the transmission efficiency deviation is negative, and the complexity deviation is greater than or equal to the second preset deviation threshold, or if both the hardness deviation and the transmission efficiency deviation are positive, then the risk level of the EICU patient is Level 1. If the hardness deviation is positive, the transmission efficiency deviation is negative, and the transmission efficiency deviation is greater than the preset lower limit of complexity deviation, and the complexity deviation is less than the second preset deviation threshold and greater than the preset lower limit of complexity deviation, then the risk level of the EICU patient is level two. If the deviation of the delivery efficiency is less than or equal to the preset lower limit of the delivery efficiency deviation, or if the deviation of the complexity is less than or equal to the preset lower limit of the complexity deviation, then the risk level of the EICU patient is level three; the risk level of level one, level two and level three gradually increases.

[0015] This invention offers the following advantages: It collects electrocardiogram (ECG) signals, invasive arterial pressure signals, and photoplethysmography (PPG) pulse wave signals, covering the complete physiological process from ECG triggering to central arterial pulsation to peripheral pulse conduction and perfusion, laying a data foundation for multi-dimensional parameter fusion. By measuring the time difference between the initiation of the central pulse and the arrival of the peripheral pulse, it determines the vascular stiffness rate index, essentially measuring pulse wave conduction time; the stiffer the blood vessel, the faster the conduction. This transforms the traditional, relatively static assessment of vascular stiffness into a continuous, dynamically calculable rate index, capable of reflecting real-time trends in vascular tension. In determining pulsation transmission efficiency, it creatively links cardiac afterload with peripheral perfusion, sensitively reflecting the overall efficiency of the heart in overcoming resistance and effectively perfusing the extremities, making it a core indicator for assessing circulatory efficiency. The calculation of stiffness deviation and transmission efficiency deviation can focus on the patient's own condition. The relative changes and the trend of deviation are closely related to early risks, thus solving the problem of hysteresis. This invention also analyzes the complexity deviation of photoplethysmography (PPG) signals within a preset reference time period as a third-dimensional identification feature. For example, physiological stress (such as pain) is usually accompanied by active neural regulatory variations (high complexity), while pathological shock is characterized by rigidity and regulatory failure of microcirculation mechanisms (low complexity). Involving these features in risk warning can greatly enhance the sensitivity and accuracy of the warning. Finally, by comprehensively analyzing the deviation of the three dimensions, multi-dimensional information fusion and risk quantification output are achieved, overcoming the one-sidedness of a single parameter. Comprehensive judgment is made from three levels: vascular properties, macroscopic efficacy, and microscopic regulation, improving the accuracy of risk warning for EICU patients. At the same time, it can output graded warning results, transforming complex physiological information into intuitive warning results. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the module composition of an EICU patient risk dynamic grading and early warning system based on multi-parameter fusion provided in an embodiment of the present invention; Figure 2 This is a data processing flowchart of an early warning module provided in one embodiment of the present invention; Figure 3 This is a flowchart illustrating the acquisition of pulsational transmission energy efficiency according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All data and information collected in this application have been obtained with full consent.

[0019] This embodiment provides a multi-parameter fusion-based dynamic risk grading and early warning system for EICU patients (hereinafter referred to as the system), used for risk grading and early warning of EICU patients (hereinafter referred to as patients). Figure 1 As shown, the system includes a data acquisition module and an early warning module. The data acquisition module and the early warning module are electrically connected. The connection between the data acquisition module and the early warning module can be wired or wireless.

[0020] This embodiment acquires the patient's electrocardiogram (ECG) signal, invasive arterial pressure (IBP) signal, and photoplethysmography (PPG) signal through a data acquisition module. ECG signals are typically acquired using ECG electrodes (usually wet or dry electrodes attached to the skin surface) that detect minute potential differences (typically in the millivolt range) on the body surface caused by the heart's electrical activity. IBP signals are typically acquired using a pressure sensor, and PPG signals are acquired using a photoelectric sensor (usually composed of a light-emitting diode and a photodetector). It should be understood that acquiring these three signals using corresponding sensors is a conventional technique and will not be described in detail further.

[0021] In this embodiment, the ECG, IBP, and PPG signals are sampled at the same frequency and are sampled synchronously. The sampling frequency is set according to actual monitoring needs; in this embodiment, a minimum of 125Hz is used. Additionally, this embodiment can include a first-in-first-out (FIFO) circular buffer, the length of which is set according to actual needs, such as 10 seconds, to align these three signals.

[0022] The processing flow in this embodiment uses a single heartbeat as the basic unit, ensuring that all calculated features are strictly subordinate to the same cardiac ejection event in terms of temporal causality. The sampling period is determined based on a single heartbeat; for example, the duration of a single heartbeat can be considered as a sampling period. This embodiment uses the current sampling period as an example. The data acquisition module acquires the ECG, IBP, and PPG signals of the EICU patient during the current sampling period.

[0023] The early warning module receives ECG, IBP, and PPG signals from the data acquisition module and then processes the received data signals. The early warning module can be a conventional data processing chip, such as a CPU (Central Processing Unit), or a System on Chip (SoC). The SoC can be configured in a medical monitoring platform or in the smart handheld terminal of an EICU patient.

[0024] Figure 2 The implementation steps of the data processing process for the early warning module are as follows: Figure 2 As shown, the data processing steps of the early warning module include: Step S1: Obtain the vascular stiffness rate index from the time interval between the start time of the central pulse and the arrival time of the peripheral pulse in the current sampling period; Step S2: Based on the accumulation of photoplethysmography (PPG) wave signals between the arrival times of two adjacent peripheral pulses and the accumulation of invasive arterial pressure signals relative to a preset invasive arterial pressure baseline between the initiation times of two adjacent central pulses, the pulsation transmission efficiency is obtained. Step S3: Determine the stiffness deviation of the vascular stiffness rate index relative to the stiffness baseline of the current sampling period, and the pulsatility transmission efficiency deviation relative to the transmission efficiency baseline of the current sampling period. Step S4: Determine the complexity deviation of the photoplethysmography signal within a preset reference time period; Step S5: Analyze the deviation of stiffness, the deviation of transmission efficiency, and the deviation of complexity to classify and warn of the risk of EICU patients.

[0025] The following is a detailed explanation of each step.

[0026] Step S1: Obtain the vascular stiffness rate index from the time interval between the start time of the central pulse and the arrival time of the peripheral pulse in the current sampling period.

[0027] Hemodynamic waveforms require a certain transmission time to travel from the heart to the extremities. Directly capturing data at the same absolute moment would lead to a mismatch between the "central drive of the current heartbeat" and the "peripheral response of the previous heartbeat." To resolve this timing alignment issue, this embodiment uses the physiological sequence of blood flow as a trigger source, cascading and locking the response times of the central arteries and peripheral microcirculation—that is, the initiation time of the central pulse and the arrival time of the peripheral pulse. For ease of explanation, this embodiment sets the current sampling period to the kth heartbeat, i.e., the kth cardiac cycle.

[0028] First, the heartbeat trigger time of the ECG signal in the current sampling period is determined. In an exemplary embodiment, this embodiment applies an R-wave peak detection algorithm to scan the ECG signal in the k-th heartbeat, identifies the R-wave peak of the k-th heartbeat, and marks the R-wave peak of the k-th heartbeat as the heartbeat trigger time of the k-th heartbeat, denoted as . Then, this moment is defined as the starting reference of the kth heartbeat, i.e., the "starting gun" moment.

[0029] Considering that mechanical ejection lags behind electrical excitation (electromechanical delay), this embodiment uses the cardiac triggering time. Starting from a certain point in time, a time interval is defined as the first time interval, with a first preset duration as the starting point. In an exemplary embodiment, based on physiological common sense, the starting point of the first time interval is the moment the heartbeat is triggered. After 10ms, the first preset duration is 100ms, therefore the first time interval is... In this embodiment of the invention, ms represents milliseconds.

[0030] Based on the IBP signal, from The minimum point (i.e., wave foot) of the IBP signal is searched within the time interval. The wave foot is the turning point in the IBP signal where the end of end-diastole and the beginning of systole begin. The pressure value at this point is generally considered to be closest to the diastolic blood pressure (although a more precise diastolic blood pressure may be after the dicrotic notch). It is the primary landmark for waveform identification in hemodynamic analysis. The time corresponding to this wave foot is taken as the central pulse initiation time, denoted as _____. This represents the moment when the aortic valve opens and blood is ejected into the central artery.

[0031] This embodiment uses the starting time of the central pulse. Starting from a second preset duration, a time interval is defined as the second time interval. In an exemplary embodiment, the duration of the second time interval is 600ms, i.e., the second time interval is... Furthermore, this embodiment also considers the delay in pulse wave transmission to the fingertip. To avoid false detection of the waveform of the next cardiac cycle (i.e., to prevent "futurism" errors), this embodiment sets the latest cutoff time of the second time interval to the moment of the R-wave peak of the (k+1)th heartbeat. Therefore, the upper limit of the cutoff time of the second time interval is limited to the moment of the R-wave peak of the (k+1)th heartbeat. Therefore, the second time interval is .in, This represents the function that takes the minimum value, i.e., takes... and Earlier in time.

[0032] Based on PPG signals, from The minimum point (i.e., wave foot) of the PPG signal is searched within the time interval. The time corresponding to the minimum point (i.e., the wave foot) of the PPG signal is taken as the time of arrival of the peripheral pulse, denoted as . This represents the moment when pulse energy is transmitted to the microcirculation of the fingertips. Therefore, the trigger moment of the k-th heartbeat... The start time of the current sampling period is the time when the peripheral pulse of the (k+1)th heartbeat arrives.

[0033] It should be understood that if the minimum value point satisfying the timing constraints is not detected in the k-th heartbeat (e.g., due to signal dropout or severe artifacts), this embodiment marks the k-th heartbeat as invalid and does not perform subsequent calculations to avoid erroneous data affecting the accuracy of data processing.

[0034] Vascular stiffness rate is a key physical quantity reflecting the tension state of the vascular wall. Under stress or compensatory shock, sympathetic nerve excitation leads to vascular smooth muscle contraction, wall stiffening, increased pulse wave velocity, and shortened pulse arrival time (PAT). Therefore, determining the central pulse initiation time is crucial. With the arrival time of the peripheral pulse The time interval between them is the pulse conduction time. .

[0035] Based on pulse conduction time The vascular stiffness rate index was obtained, and the vascular stiffness rate index was correlated with pulse conduction time. Inverse correlation. In an exemplary embodiment, pulse conduction time is used. The reciprocal of is used as an indicator of the rate of vascular stiffness.

[0036] In this embodiment, to prevent false detection of wave feet due to signal noise (e.g., misidentifying artifacts near the R wave as PPG wave feet), which could lead to the denominator approaching zero or generating non-physiologically abnormally high stiffness values, this embodiment introduces a minimum physiological conduction time threshold. (For example, set to 50ms). Then, the formula for calculating the vascular stiffness rate index of the k-th heartbeat is as follows: ; in, The denominator represents the vascular stiffness rate index of the k-th heartbeat, and `max` represents the maximum clamping function. The denominator uses the maximum clamping function, which is applied when the measured pulse conduction time is... Less than At that time, the system forces the use of Perform calculations to ensure the numerical stability of the results. A higher value indicates a faster pulse wave transmission, reflecting a higher tension or stiffness of the patient's blood vessel walls during the current sampling period. This vascular stiffness rate index reflects the relative rate of vascular stiffness. Using the same calculation method, the vascular stiffness rate index for each heartbeat before the k-th heartbeat was obtained.

[0037] Step S2: Based on the accumulation of photoplethysmography (PPG) wave signals between the arrival times of two adjacent peripheral pulses and the accumulation of invasive arterial pressure signals relative to a preset invasive arterial pressure baseline between the initiation times of two adjacent central pulses, the pulsation transmission efficiency is obtained.

[0038] The amplitude of a single peripheral pulse wave is easily affected by various factors and cannot truly reflect the perfusion efficiency of the circulatory system. To objectively evaluate the "effective pulsatile volume" generated by "unit central driving energy" at the periphery, this embodiment analyzes the conversion efficiency between central arterial pressure work and peripheral volume changes during the same ejection event. Specifically, it determines the accumulation of photoplethysmography (PPG) pulse wave signals between the arrival times of two adjacent peripheral pulses, and the accumulation of invasive arterial pressure signals relative to a preset invasive arterial pressure baseline between the onset times of two adjacent central pulses. Based on these two accumulations, the pulsatile transmission efficiency is obtained.

[0039] In one exemplary embodiment, such as Figure 3 As shown, the following is one process for obtaining pulsational transmission energy efficiency: Step S21: Determine the standardized signal of the photoplethysmography (PPG) signal.

[0040] First, PPG signals are susceptible to fluctuations in amplitude due to sensor contact pressure, ambient light, and device gain, resulting in non-standardized relative values ​​that lack direct comparability between different patients or at different times. To eliminate this gain difference, this embodiment standardizes the PPG signal to obtain a standardized PPG signal.

[0041] For any moment within the k-th heartbeat, defined as the t-th moment, the PPG signal at the t-th moment is denoted as... Get The signal contains both DC and AC components, where the DC component is typically the average or envelope floor of the PPG signal amplitude during the k-th heartbeat. The ratio of the AC to DC components is calculated to obtain the standardized signal, using the following formula: ; in, express Standardized signals, express The amount of communication, express The DC component. In this embodiment, to ensure data stability, in DC component When it is less than the preset lower limit of the DC component, The calculation result is limited to the preset lower limit of the DC component to avoid the denominator being too small and causing overflow.

[0042] The above The standardized signal obtained through standardization is essentially the rate of change of photoplethysmography, which can eliminate interference from non-physiological factors such as contact pressure, providing a physically consistent input source for subsequent calculations of pulsatility transfer efficiency. Simultaneously, the system can be configured with a high-pass filter with a cutoff frequency of 0.5Hz to further eliminate low-frequency baseline drift caused by respiratory motion.

[0043] Step S22: Integrate the standardized signal between the arrival times of two adjacent peripheral pulses to obtain the first integral result.

[0044] Step S1 is used to obtain the arrival time of the peripheral pulse in each heartbeat, thereby obtaining the arrival times of two adjacent peripheral pulses. In this embodiment, the arrival times of two adjacent peripheral pulses are specifically as follows: and , The time when the peripheral pulse arrives at the (k+1)th heartbeat.

[0045] Get and The standardized PPG signals at each time point within the time interval are obtained, and then integrated to obtain the first integral result. To prevent negative area caused by waveform polarity reversal or baseline fluctuations, this embodiment first takes the absolute value of the standardized signal before integration. This ensures that regardless of the signal morphology, this term always quantifies the total volumetric pulsatility of the peripheral vascular bed. The calculation method for the first integral result is as follows: ; in, This represents the first integral result corresponding to the k-th heartbeat.

[0046] Step S23: Determine the signal difference between the invasive arterial pressure signal and the preset invasive arterial pressure baseline at each time between the start times of two adjacent central pulses.

[0047] Step S1 is used to obtain the starting time of the central pulse in each heartbeat, thereby obtaining the starting time of two adjacent central pulses. In this embodiment, the starting time of two adjacent central pulses is specifically as follows: and , This is the starting time of the central pulse of the (k+1)th heartbeat.

[0048] This embodiment predetermines an invasive arterial pressure baseline, which is specifically the moment the central pulse begins. The amplitude of the IBP signal at point is denoted as .

[0049] calculate and IBP signal at each time point and The signal difference. In an exemplary embodiment, the process of obtaining the signal difference is as follows: obtaining... and The average amplitude of the IBP signal at each time point is calculated, and then the IBP signal at each time point is compared with... The absolute value of the amplitude difference is calculated, and the ratio of this absolute value of the amplitude difference to the average amplitude is used to achieve dimensionless representation of the absolute value of the amplitude difference, making the signal difference a dimensionless rate of change of signal. This embodiment aims to quantify only the work done by the effective pulse pressure generated by cardiac contraction; therefore, the absolute value of the difference is calculated.

[0050] Step S24: Integrate the signal differences between the start times of two adjacent central pulses to obtain the second integral result.

[0051] right and IBP signal at each time point and The signal difference is integrated to obtain the second integral result, and the setting is... and The IBP signal at time t is The calculation method for the results in the second section is as follows: ; in, This represents the second integral result corresponding to the k-th heartbeat. express and The average amplitude of the IBP signal at each time point between.

[0052] It should be understood that and The average amplitude of the IBP signal at each time point cannot be 0. If there is a case where it is 0, it indicates that there is a serious abnormality in the IBP signal of the k-th heartbeat. In this embodiment, the k-th heartbeat is marked as invalid and subsequent calculations are not performed to avoid erroneous data affecting the accuracy of data processing.

[0053] Step S25: Obtain the pulsational transmission efficiency from the first integral result and the second integral result.

[0054] The first integral result corresponding to the k-th heartbeat represents the peripheral response, and the second integral result represents the central drive. The pulsatility transfer efficiency is obtained from the first and second integral results, where the pulsatility transfer efficiency is positively correlated with the first integral result and inversely correlated with the second integral result. In an exemplary embodiment, the ratio of the first and second integral results (i.e., using the first integral result as the numerator and the second integral result as the denominator) is taken, and the result is the pulsatility transfer efficiency corresponding to the k-th heartbeat, also known as the pulsatility energy transfer ratio. It should be understood that, to prevent the denominator from approaching 0 and causing a division-by-zero error, this embodiment adds a small positive number to the denominator (e.g., ...). The added small positive number has the same dimension as the denominator.

[0055] The lower the value of the pulsatile transmission efficiency corresponding to the kth heartbeat, the less effective pulsatile blood flow is obtained at the fingertip, even if there is pressure drive in the central artery. This reflects excessive microcirculatory resistance or energy dissipation during transmission.

[0056] Step S3: Determine the deviation of the vascular stiffness rate index from the stiffness baseline of the current sampling period, and the deviation of the pulsatility transmission efficiency from the transmission efficiency baseline of the current sampling period.

[0057] Based on the vascular stiffness rate index and pulsational transmission efficiency corresponding to the kth heartbeat, and combined with the stiffness baseline and transmission efficiency baseline corresponding to the kth heartbeat, the stiffness deviation and transmission efficiency deviation corresponding to the kth heartbeat are obtained, so as to map the vascular stiffness rate index and pulsational transmission efficiency onto the same statistical scale.

[0058] This embodiment addresses the baseline drift problem caused by individual physiological heterogeneity and disease progression in intensive care settings when determining the stiffness baseline and transmission efficiency baseline. Since there are significant differences in the baseline physiological parameters of different patients, and critically ill patients may already be in an abnormal state upon admission to monitoring, traditional fixed threshold methods are difficult to adapt. Therefore, this embodiment introduces a recursive statistical framework based on population priors, achieving a smooth transition from a "general population model" to a "specific individual model." More importantly, this embodiment introduces a state-gated baseline freezing strategy to prevent the system from erroneously updating the pathological state to a normal baseline when the patient's condition continues to deteriorate. In this embodiment, the stiffness baseline includes two parts: the mean stiffness and the standard deviation of stiffness. The transmission efficiency baseline includes the mean transmission efficiency and the standard deviation of transmission efficiency. The mean stiffness represents the average value of the vascular stiffness rate index, and the standard deviation of stiffness represents the standard deviation of the vascular stiffness rate index. The mean transmission efficiency represents the average value of pulsatility transmission efficiency, and the standard deviation of transmission efficiency represents the standard deviation of pulsatility transmission efficiency.

[0059] To provide effective relative assessment during the initial patient admission phase (cold start phase), this embodiment first constructs a population prior model using large-scale historical clinical data. This embodiment pre-defines a set of population prior parameters, including the baseline prior mean of stiffness. Prior standard deviation of hardness baseline , transfer energy efficiency baseline prior mean and the prior standard deviation of the energy efficiency baseline These prior parameters can be statistically derived from publicly available intensive care databases such as MIMIC-IV. For example, data from adult patients with stable vital signs (e.g., mean arterial pressure (MAP) > 65 mmHg and serum lactate < 2 mmol / L) within the first hour of ICU admission can be selected to calculate the statistical distribution of their vascular stiffness rate index and pulsatility transmission efficiency. For example, setting... The rate value corresponding to a pulse conduction time of 250 ms These correspond to the standard deviation of the population data. These prior values ​​serve only as initial anchors for cold starts and will be gradually corrected by individual data as monitoring progresses.

[0060] For any k-th heartbeat, in (i.e., at the start of the monitoring period), the individual baseline is initialized using the aforementioned population prior parameters: ; ; ; ; During the continuous monitoring phase (i.e.) (Greater than 1), this embodiment requires updating the individual baseline based on newly acquired feature values. However, if recursive updates are performed unconditionally, when the patient is in a state of shock for a long time, abnormal vascular stiffness rate indicators and pulsatility transmission efficiency will be gradually "learned" into the baseline, causing the deviation indicators to become invalid. Therefore, taking the k-th heartbeat as an example, this embodiment implements the following state-gated update logic: Determine whether the patient in the previous sampling period (i.e., the index of the previous heartbeat after the k-th heartbeat, i.e., the (k-1)-th heartbeat) is in a safe risk state, i.e., obtain the cyclic risk level of the (k-1)-th heartbeat output. If it is a safe state, that is... If the level is 0 (i.e., steady state), then the k-th heartbeat is deemed suitable for baseline correction. The stiffness baseline of the (k-1)-th heartbeat is updated to obtain the stiffness baseline of the k-th heartbeat. The transmission efficiency baseline of the (k-1)-th heartbeat is updated to obtain the transmission efficiency baseline of the k-th heartbeat.

[0061] The process of correcting the hardness baseline: This embodiment utilizes an exponentially weighted moving average algorithm for updates. Specifically: The weighting coefficients of the average hardness value from the previous sampling period and the vascular hardness rate index from the current sampling period are determined. The average hardness value from the previous sampling period and the vascular hardness rate index from the current sampling period are then weighted and summed to obtain the average hardness value for the current sampling period. The difference between the vascular hardness rate index from the current sampling period and the average hardness value from the current sampling period is determined. The weighting coefficients of the square of this difference and the hardness variance from the previous sampling period are then determined. The square of this difference and the hardness variance from the previous sampling period are then weighted and summed to obtain the hardness variance for the current sampling period. The standard deviation of the hardness value for the current sampling period is then obtained from the hardness variance. The details are as follows: ; ; in, This represents the average hardness value in the baseline hardness range during the k-th heartbeat. This indicates a preset forgetting factor (e.g., a value of...). ), used to adjust the baseline's sensitivity to new data, This represents the average hardness value in the baseline hardness range of the (k-1)th heartbeat. This represents the rate of vascular stiffness during the k-th heartbeat. As a weighting coefficient for the rate of vascular stiffness during the k-th heartbeat, The weighting factor for the average hardness value in the hardness baseline of the (k-1)th heartbeat.

[0062] This represents the variance of the stiffness of the k-th heartbeat. This represents the standard deviation of stiffness in the baseline of the k-th heartbeat. This represents the variance of the stiffness of the (k-1)th heartbeat. This represents the standard deviation of hardness in the baseline of the (k-1)th heartbeat. As the kth heartbeat The weighting coefficient of the square, The weighting coefficient is used as the variance of the stiffness of the (k-1)th heartbeat.

[0063] If the patient's risk is not in a safe state during the (k-1)th heartbeat, i.e. For risk states other than level 0, such as level I, level II, or level III as mentioned later, the k-th heartbeat is determined to be in a pathological risk state. To protect the reference value of the baseline, baseline updates are paused (frozen), and the baseline parameters of the (k-1)-th heartbeat are directly inherited. ; .

[0064] Using the above-described process for updating the stiffness baseline, the transmission efficiency baseline is updated to obtain the transmission efficiency baseline for the k-th heartbeat, as follows: If the patient's risk is not in a safe state during the (k-1)th heartbeat, then the kth heartbeat is deemed suitable for baseline correction, and the transmission efficiency baseline of the (k-1)th heartbeat is updated to obtain the transmission efficiency baseline of the kth heartbeat, as follows: The weighting coefficients for the average transfer efficiency of the previous sampling period and the pulsatile transfer efficiency of the current sampling period are determined. A weighted sum of the average transfer efficiency of the previous sampling period and the pulsatile transfer efficiency of the current sampling period is then obtained to obtain the average transfer efficiency of the current sampling period. The difference between the pulsatile transfer efficiency of the current sampling period and the average transfer efficiency of the current sampling period is determined. The weighting coefficients for the square of this difference and the variance of the transfer efficiency of the previous sampling period are then determined. A weighted sum of the square of this difference and the variance of the transfer efficiency of the previous sampling period is then obtained to obtain the variance of the transfer efficiency of the current sampling period. The standard deviation of the transfer efficiency of the current sampling period is then derived from the variance of the transfer efficiency of the current sampling period. The details are as follows: ; ; in, This represents the average transmission efficiency in the baseline of transmission efficiency for the k-th heartbeat. This represents the average transmission efficiency in the baseline of the (k-1)th heartbeat. This represents the pulsational efficiency of the k-th heartbeat. As a weighting coefficient for the pulsational energy transmission efficiency of the k-th heartbeat, The weighting coefficient for the average value of the transmission efficiency in the baseline of the transmission efficiency of the (k-1)th heartbeat.

[0065] This represents the variance of the transmission efficiency of the k-th heartbeat. This represents the standard deviation of transmission efficiency in the baseline of transmission efficiency for the k-th heartbeat. This represents the variance of the transmission efficiency of the (k-1)th heartbeat. This represents the standard deviation of the transmission efficiency in the baseline of the (k-1)th heartbeat. As the kth heartbeat The weighting coefficient of the square, The weighting coefficients are used as the variance of the transmission energy efficiency of the (k-1)th heartbeat.

[0066] If the patient's risk is not in a safe state during the (k-1)th heartbeat, i.e. For risk states other than level 0, such as level I, level II, or level III as mentioned later, the k-th heartbeat is determined to be in a pathological risk state. To protect the reference value of the baseline, baseline updates are paused (frozen), and the transmission efficiency parameters of the (k-1)-th heartbeat are directly inherited. ; .

[0067] Through the above-mentioned correction and update process, it can both adapt to the physiological drift of individual patients and ensure that the baseline is always anchored at the "healthy / homeostatic" level before the onset of the disease when the condition worsens, thereby ensuring the continuous effectiveness of risk warning.

[0068] Then, based on the individual baseline parameters maintained by the above logic, the dimensionless deviation (i.e., Z-score) of the current heartbeat is calculated. This step maps the completely different physical dimensions of "rate" and "ratio" onto the same statistical scale. The formula for calculating stiffness deviation is as follows: ; in, This indicates the degree of stiffness deviation of the k-th heartbeat. Indicates the lower limit of the preset standard deviation of hardness (e.g.) As a minimum clamping parameter, it is used to prevent the standard deviation from approaching zero due to extremely flat pulsation transmission efficiency or artifacts, which would then cause division by zero overflow.

[0069] The formula for calculating the deviation of transmission effectiveness is as follows: ; in, This indicates the deviation in transmission efficiency of the k-th heartbeat. This indicates the lower limit of the preset standard deviation of transmission performance (e.g.) As a minimum clamping parameter, it is used to prevent the standard deviation from approaching zero due to extremely flat vascular stiffness rate indicators or artifacts, which would then cause divide-by-zero overflow.

[0070] Through the above calculations, it is possible to achieve the following: A value greater than 0 indicates that the blood vessel stiffness is higher than the patient's average level. A value less than 0 indicates that the perfusion efficiency is lower than the patient's average level, providing standardized coordinate input for subsequent geometric plane analysis.

[0071] Step S4: Determine the complexity deviation of the photoplethysmography signal within a preset reference time period.

[0072] This step introduces an information dimension feature reflecting the complexity of microcirculation regulation, which is then fused with the stiffness deviation and transmission efficiency deviation features obtained in the previous steps to construct a complete hemodynamic state description. Addressing the issue of high overlap in vasoconstriction intensity between physiological stress (such as pain and chills) and pathological shock, this step specifically identifies pathological rigidity of the microcirculation at the information dimension by extracting nonlinear microcirculation fluctuation regularity features.

[0073] During physiological stress, neural regulation is active, and vasomotor activity exhibits high-frequency variations. However, microcirculatory failure caused by shock manifests as decoupling of regulatory mechanisms, with rigid, monotonous contractions of the vessel walls and a significant decrease in signal complexity. To quantify this characteristic, this embodiment uses the complexity of the PPG signal within a preset reference time period as an indicator. Further, complexity is represented by sample entropy.

[0074] The specific implementation method of the preset reference time period is as follows: the arrival time of the peripheral pulse during the k-th heartbeat process. The preset reference time period serves as the cutoff point. The duration of this preset reference time period is set according to actual conditions; in this embodiment, it is 30 seconds. Therefore, the preset reference time period is a 30-second historical backtracking window. It should be understood that the neural regulatory state of microcirculation has a certain physiological inertia and is a slowly changing process. Although this step is based on data from the past 30 seconds, in subsequent beat-by-beat determinations, the sample entropy is considered as an indicator of the "microcirculation regulatory background" of the current heartbeat, used to assess the current neural regulatory environment, rather than the instantaneous pulse morphology.

[0075] Get preset reference time period PPG signal segment within .in, This indicates the duration of the preset reference time period. To reduce computational load and focus on low-frequency physiological regulatory components, this embodiment can also use the PPG signal segment... Downsample to a preset frequency, such as 20Hz.

[0076] The sample entropy algorithm is used to calculate the preset reference time period. Sample entropy of the PPG signal segment within The algorithm parameters are set as follows: embedding dimension. Similarity tolerance , This represents the standard deviation of the signal within the PPG signal segment. The lower the value, the higher the repeatability (high regularity) of microcirculatory fluctuations, suggesting that vascular regulation tends to be rigid; the higher the value, the more complex the fluctuations, suggesting that neural regulation is active or there is interference.

[0077] To incorporate sample entropy into a unified evaluation system, this embodiment also employs the recursive statistics and state gating logic from step S3 to maintain the baseline parameters of the entropy value. and .in, This represents the average sample entropy. It represents the standard deviation of sample entropy.

[0078] The update logic for the average hardness and standard deviation of hardness in step S3, as well as the average energy transfer efficiency and standard deviation of energy transfer efficiency, is exactly the same, as shown below: First, pre-determine the prior parameters for the mean sample entropy and the standard deviation of sample entropy.

[0079] Taking the k-th heartbeat as an example, if the patient's risk is not in a safe state during the (k-1)-th heartbeat, then the k-th heartbeat is deemed suitable for baseline correction. The sample entropy mean and sample entropy standard deviation of the (k-1)-th heartbeat are updated to obtain the sample entropy mean and sample entropy standard deviation of the k-th heartbeat, as detailed below: ; ; in, This represents the average sample entropy of the k-th heartbeat. This represents the average sample entropy of the (k-1)th heartbeat. Let represent the sample entropy of the k-th heartbeat. As a weighting coefficient for the sample entropy of the k-th heartbeat, The weighting coefficient is used as the average sample entropy of the (k-1)th heartbeat.

[0080] This represents the sample entropy variance of the k-th heartbeat. This represents the standard deviation of the sample entropy of the k-th heartbeat. This represents the sample entropy variance of the (k-1)th heartbeat. This represents the standard deviation of the sample entropy of the (k-1)th heartbeat. As the kth heartbeat The weighting coefficient of the square, The weighting coefficients are used as the sample entropy variance of the (k-1)th heartbeat.

[0081] If the patient's risk is not in a safe state during the (k-1)th heartbeat, i.e. For risk states other than level 0, such as level I, level II, or level III as mentioned later, the k-th heartbeat is determined to be in a pathological risk state. To protect the reference value of the baseline, baseline updates are paused (frozen), and the sample entropy parameter of the (k-1)-th heartbeat is directly inherited. ; .

[0082] Calculate the sample entropy deviation based on baseline parameters: ; in, This represents the deviation of the sample entropy for the k-th heartbeat. This represents the lower limit of the preset sample entropy standard deviation (e.g.) This parameter, known as the minimum clamping parameter, is used to prevent the standard deviation from approaching zero due to extremely flat sample entropy or artifacts, thus avoiding division-by-zero overflow. Standardization makes... This intuitively reflects whether the current microcirculatory complexity is "more active" (positive value) or "more rigid" (negative value) relative to the patient's baseline. The entropy deviation of this sample is the complexity deviation.

[0083] Step S5: Analyze the deviation of stiffness, the deviation of transmission efficiency, and the deviation of complexity to classify and warn of the risk of EICU patients.

[0084] Combining the above steps, construct the first... Three-dimensional hemodynamic eigenvectors of the second heartbeat : ; Where T represents matrix transpose.

[0085] Three-dimensional hemodynamic eigenvectors Orthogonal integration incorporates three types of key information: Mechanical input characterizing vascular tension (the higher the input, the tighter the tension); Mechanical output characterizing peripheral irrigation (the higher the better); The informational attributes characterizing microcirculatory regulation (lower values ​​indicate greater rigidity) are represented. This complete feature vector fully describes the current hydrodynamic state of the heartbeat, providing deterministic input data for subsequent risk grading and early warning of patients.

[0086] Based on three-dimensional hemodynamic feature vectors By using geometric position analysis to map abstract statistical deviations back to specific pathological states, the core logic lies in using vector directionality under a unified zero-point benchmark to identify the nonlinear decoupling (scissor difference) between "tension input" and "perfusion output," and using microcirculation fluctuation regularity as a specific filter to achieve early detection of hidden risks.

[0087] This embodiment first examines the three-dimensional hemodynamic eigenvectors. The projection onto the energy efficiency plane. This energy efficiency plane is defined by the horizontal axis. (Representing vascular tension input) and longitudinal axis (Representing injection output efficiency) is composed primarily of hardness deviation. and deviation of transmission efficiency Calculate the three-dimensional hemodynamic eigenvectors Projected modulus on the energy efficiency plane: ; in, Indicates hardness deviation and deviation of transmission efficiency The projection modulus on the energy efficiency plane, then... Essentially, it is due to the deviation in hardness. and deviation of transmission efficiency The obtained comprehensive deviation characteristics.

[0088] To avoid the system being in a steady state near the baseline (i.e., hardness deviation) and deviation of transmission efficiency When all values ​​are close to 0, minute signal noise can cause drastic fluctuations in the calculation results (singularity problem). This embodiment introduces a dead zone determination mechanism. Specifically, this embodiment presets a first preset deviation threshold, which is used to compare with the comprehensive deviation feature. When the deviation is less than or equal to the first preset deviation threshold, it indicates that the current state deviates from the baseline insignificantly, and subsequent processing is skipped directly, determining the patient's risk level as a safe state, i.e., level 0 in the following text. The first preset deviation threshold represents the significance threshold, and the specific value is set according to the actual judgment needs. This embodiment uses 1.5 (i.e., 1.5 times the deviation of the corresponding standard deviation) as an example.

[0089] When the overall deviation characteristic exceeds the first preset deviation threshold, it indicates a significant deviation in the current state, requiring further assessment to determine the patient's risk level, as detailed below: When the hardness deviates from the degree It is a positive value, and the transmission efficiency deviation is... When the value is positive, the physiological meaning is that the body increases vascular tone input and simultaneously achieves higher perfusion output; the two are positively correlated, usually corresponding to a physiologically hyperdynamic state (such as the recovery period after exercise), indicating good circulatory reserve function. When the stiffness deviates from the desired value... It is a positive value, and the transmission efficiency deviation is... When the value is negative, the physiological meaning is: "efficiency scissors" state, in which the body significantly increases vascular tension input (increased stiffness), but the perfusion output is lower than the baseline (decreased efficiency). This suggests that the peripheral blood vessels are extremely constricted to maintain central blood pressure, but the microcirculatory resistance is too high, resulting in impaired effective perfusion, which is a typical mechanical feature of compensatory shock.

[0090] Therefore, based on the hardness deviation and deviation of transmission efficiency The sign of the value, and the deviation of the sample entropy. The patient's risk level is determined by the magnitude of the deviation from a second preset threshold. The second preset threshold is used to assess the sample entropy deviation. Whether it is too high, the second preset deviation threshold is a negative value. The specific value is set according to the actual judgment needs. In this embodiment, -2 is used as an example, that is, twice the opposite of the corresponding standard deviation is used as the threshold. Moreover, this embodiment also presets a lower limit value for transmission efficiency deviation and a lower limit value for sample entropy deviation. The lower limit value for transmission efficiency deviation represents the minimum allowable value for transmission efficiency deviation. When the transmission efficiency deviation is less than the lower limit value, it indicates that the transmission efficiency deviation is very serious and abnormal. The lower limit value for sample entropy deviation represents the minimum allowable value for sample entropy deviation. When the sample entropy deviation is less than the lower limit value, it indicates that the sample entropy deviation is very serious and abnormal. In an exemplary embodiment, both the lower limit value for transmission efficiency deviation and the lower limit value for sample entropy deviation are taken as -3, that is, three times the opposite of the corresponding standard deviation is used as the threshold.

[0091] This embodiment predetermines three danger levels: Level 1 (Level I), Level 2 (Level II), and Level 3 (Level III), with the corresponding danger levels gradually increasing. That is, Level I is less dangerous than Level II, and Level II is less dangerous than Level III. Level I is a high-dynamic stress indicator, Level II is a latent compensatory warning, and Level III is a microcirculatory failure alarm.

[0092] If ①: Hardness deviation Positive number, deviation of transmission efficiency Negative and sample entropy deviation Greater than or equal to the second preset deviation threshold, or, ②: hardness deviation and deviation of transmission efficiency If all values ​​are positive, the patient's risk level is determined as follows: Level I, and a Level I warning signal is output. Clinical significance: Although there is vascular hypertonia or dysfunction, the microcirculatory regulation is active (sample entropy deviation is normal or high), suggesting a physiological stress response caused by pain, chills, or anxiety, rather than microcirculatory failure. Systemic action: Displays a "hyperdynamic stress state" indicator, recommending that healthcare professionals pay attention to analgesia or sedation management.

[0093] If the hardness deviates from the degree Positive number, deviation of transmission efficiency It is a negative number, and the transmission efficiency deviation is... The deviation is greater than the preset lower limit for transmission efficiency, and the sample entropy deviation is... If the deviation is less than the second preset deviation threshold but greater than the preset lower limit of sample entropy deviation, the patient's risk level is classified as Level II, and a Level II warning signal is output. Clinical significance: This indicates a significant efficacy gap, accompanied by pathological rigidity of the microcirculation regulation mechanism. This is a typical case of occult compensatory shock with "normal blood pressure but impaired perfusion." System action: Triggers a warning prompt, displaying "occult low perfusion risk."

[0094] If the transmission efficiency deviates Less than or equal to the preset lower limit of delivery efficiency deviation (indicating extremely low output), or sample entropy deviation If the deviation is less than or equal to the preset lower limit of sample entropy (indicating extreme rigidity of regulation), the patient's risk level is Level III, and a Level III warning signal is output. Clinical significance: Regardless of the direction of compensation, an extreme abnormality exceeding 3 times the corresponding standard deviation in perfusion efficiency or microcirculation regulation function indicates that the patient has entered the decompensated stage or microcirculation failure stage. System action: Triggers a high-level audible and visual alarm and outputs a "microcirculation failure" indicator.

[0095] It should be understood that this embodiment refers to: transmission performance deviation. The judgment of deviation from the preset lower limit of transmission efficiency, and the deviation of sample entropy. The judgment of deviation from the preset sample entropy lower limit is used as the highest priority criterion. These two judgment processes are performed first, that is, it is first judged whether it meets the level III. If it does not meet the level III, then the subsequent judgments of level I and level II are performed.

[0096] Furthermore, if the above three levels of assessment are not met, the patient's risk level is determined to be level 0, meaning they are in a safe state. Clinical significance: The system is in a stable state within the individual's baseline range, or the deviation is within physiologically permissible limits. System action: Displays a green safety indicator.

[0097] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0098] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A dynamic risk grading and early warning system for EICU patients based on multi-parameter fusion, characterized in that, include: The data acquisition module is used to acquire the electrocardiogram signal, invasive arterial pressure signal, and photoplethysmography signal of the EICU patient during the current sampling cycle. The early warning module, electrically connected to the data acquisition module, is configured as follows: The vascular stiffness rate index is obtained from the time interval between the start time of the central pulse and the arrival time of the peripheral pulse in the current sampling period. The pulsational transmission efficiency is obtained by considering the accumulation of photoplethysmography (PPG) signals between the arrival times of two adjacent peripheral pulses and the accumulation of invasive arterial pressure signals relative to a preset invasive arterial pressure baseline between the initiation times of two adjacent central pulses. Determine the degree of stiffness deviation of the vascular stiffness rate index relative to the stiffness baseline of the current sampling period, and the degree of transmission efficiency deviation of the pulsation transmission efficiency relative to the transmission efficiency baseline of the current sampling period; Determine the degree of complexity deviation of the photoplethysmography (PPG) signal within a preset reference time period; By analyzing the deviations in hardness, transmission efficiency, and complexity, a risk classification and early warning system for EICU patients can be implemented.

2. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 1, characterized in that, The process of obtaining the central pulse initiation time and the peripheral pulse arrival time includes: Determine the heartbeat trigger time of the electrocardiogram signal in the current sampling period; The minimum point of the invasive arterial pressure signal is searched within the first time interval, and the time corresponding to the minimum point of the invasive arterial pressure signal is the time when the central pulse begins; the first time interval is a time interval of a first preset duration after the time when the heartbeat is triggered. The minimum point of the photoplethysmography signal is searched within the second time interval, and the time corresponding to the minimum point of the photoplethysmography signal is the time when the peripheral pulse arrives; the second time interval is a time interval of a second preset duration starting from the time when the central pulse begins.

3. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 1, characterized in that, The process of obtaining the pulsational transmission energy efficiency includes: Determine the standardized signal of the photoplethysmography (PPG) signal; The standardized signal between the arrival times of two adjacent peripheral pulses is integrated to obtain the first integral result; Determine the signal difference between the invasive arterial pressure signal and the preset invasive arterial pressure baseline at each time point between the onset of two adjacent central pulses; The signal differences between the start times of two adjacent central pulses are integrated to obtain a second integral result; The pulsational transmission efficiency is obtained from the first integral result and the second integral result; the pulsational transmission efficiency is positively correlated with the first integral result and negatively correlated with the second integral result.

4. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 3, characterized in that, The process of acquiring the standardized signal includes: determining the DC component and AC component of the photoplethysmography (PPG) signal, calculating the ratio of the AC component to the DC component, and obtaining the standardized signal.

5. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 1, characterized in that, The process of acquiring the hardness baseline and transfer energy efficiency baseline for the current sampling period includes: Determine whether the risk of EICU patients in the previous sampling cycle is safe in the current sampling cycle; If the condition is safe, update the hardness baseline of the previous sampling period to obtain the hardness baseline of the current sampling period; update the transfer energy efficiency baseline of the previous sampling period to obtain the transfer energy efficiency baseline of the current sampling period.

6. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 5, characterized in that, The hardness baseline includes the average hardness and the standard deviation of hardness, and the energy transfer efficiency baseline includes the average energy transfer efficiency and the standard deviation of energy transfer efficiency. The process of updating the hardness baseline of the previous sampling period to obtain the hardness baseline of the current sampling period includes: Determine the weighting coefficients of the average hardness value of the previous sampling period and the vascular hardness rate index of the current sampling period, and perform a weighted summation of the average hardness value of the previous sampling period and the vascular hardness rate index of the current sampling period to obtain the average hardness value of the current sampling period. The difference between the vascular stiffness rate index of the current sampling period and the average stiffness of the current sampling period is determined. The weighting coefficient of the square of the difference and the stiffness variance of the previous sampling period is determined. The square of the difference and the stiffness variance of the previous sampling period are weighted and summed to obtain the stiffness variance of the current sampling period. The stiffness standard deviation of the current sampling period is obtained from the stiffness variance of the current sampling period.

7. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 6, characterized in that, The process of obtaining the hardness deviation includes: The hardness deviation is obtained by taking the ratio of the difference to the hardness standard deviation of the current sampling period. The process of obtaining the transmission performance deviation includes: The difference between the pulsational transmission efficiency of the current sampling period and the average transmission efficiency of the current sampling period is used as the ratio to the standard deviation of the transmission efficiency of the current sampling period to obtain the transmission efficiency deviation.

8. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 1, characterized in that, The complexity is the sample entropy of the photoplethysmography (PPG) signal within a preset reference time period.

9. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 1, characterized in that, The risk classification and early warning system for EICU patients includes: The comprehensive deviation characteristics are determined based on the hardness deviation and the transmission efficiency deviation. Compare the overall deviation feature with the magnitude of the first preset deviation threshold; When the overall deviation feature is greater than the first preset deviation threshold, the risk level of the EICU patient is determined based on the sign of the hardness deviation and the transmission efficiency deviation, as well as the magnitude of the complexity deviation and the second preset deviation threshold.

10. The EICU patient risk dynamic grading and early warning system based on multi-parameter fusion as described in claim 9, characterized in that, The determination of the risk level of EICU patients includes: If the hardness deviation is positive, the transmission efficiency deviation is negative, and the complexity deviation is greater than or equal to the second preset deviation threshold, or if both the hardness deviation and the transmission efficiency deviation are positive, then the risk level of the EICU patient is Level 1. If the hardness deviation is positive, the transmission efficiency deviation is negative, and the transmission efficiency deviation is greater than the preset lower limit of complexity deviation, and the complexity deviation is less than the second preset deviation threshold and greater than the preset lower limit of complexity deviation, then the risk level of the EICU patient is level two. If the deviation of the delivery efficiency is less than or equal to the preset lower limit of the delivery efficiency deviation, or if the deviation of the complexity is less than or equal to the preset lower limit of the complexity deviation, then the risk level of the EICU patient is level three; the risk level of level one, level two and level three gradually increases.

Images