A new-born brain injury multi-modal assessment early warning system

Abstract

The present application relates to cerebral vascular state early warning technical field, specifically to a kind of neonatal brain injury multi-modal assessment early warning system.The system is based on neonatal core body temperature adjustment standard pressure conduction delay interval at each monitoring moment, obtains the discrete event in time-constrained mismatched arterial blood pressure signal and local cerebral oxygen saturation signal, obtains event matching pair.All event matching pairs are extracted features, and cerebral vascular fluid resistance index is obtained.Arterial blood pressure and cerebral vascular fluid resistance index at each monitoring moment are used as two-dimensional data points, the distribution trend of two-dimensional data points under continuous monitoring moment is counted, and blood pressure early warning signal is determined.The present application obtains adaptive conduction delay interval as time constraint through temperature data, uses matching method to obtain event matching pair of real functional buffer result between arterial blood pressure and cerebral blood flow, and then carries out vascular functional feature analysis, to determine the reference strong blood pressure early warning signal.

Description

Technical Field

[0001] This invention relates to the field of cerebrovascular condition early warning technology, specifically to a multimodal assessment and early warning system for neonatal brain injury. Background Technology

[0002] Hypothermia is a common clinical treatment for neonatal hypoxic-ischemic encephalopathy (HIE) to reduce the brain's metabolic rate. Maintaining adequate cerebral perfusion pressure is crucial for preventing secondary brain injury during this period. Cerebrovascular autoregulation (CAR), by regulating vascular smooth muscle tone to buffer systemic blood pressure fluctuations, is a core mechanism for maintaining stable cerebral blood flow. However, assessing this function in hypothermia treatment scenarios faces specific physiological interferences. As core body temperature decreases, blood viscosity increases significantly. This rheological change prolongs the transmission time of pressure waves from the aorta to the cerebral microvessels, while increased fluid viscous resistance naturally attenuates the amplitude of pressure waves reaching the extremities.

[0003] Existing assessment methods are usually based on time-domain correlation of a fixed window or simple linear gain calculation. They cannot distinguish between signal attenuation caused by blood thickening and biological buffering caused by good vascular regulation. Furthermore, they are prone to signal misalignment due to transmission delay drift caused by low temperature. As a result, they cannot effectively assess the vascular function of newborns, and consequently, the system cannot provide a reliable early warning signal for medical staff to refer to. Summary of the Invention

[0004] To address the technical problem that existing clinical monitoring systems for neonatal hypoxic-ischemic encephalopathy are affected by low temperatures, making it impossible to effectively assess neonatal vascular function and resulting in the system's inability to provide reliable early warning signals, this invention aims to provide a multimodal assessment and early warning system for neonatal brain injury. The specific technical solution adopted is as follows: This invention proposes a multimodal assessment and early warning system for neonatal brain injury, the system comprising: The multimodal data acquisition module is used to acquire arterial blood pressure signals, local cerebral oxygen saturation signals, and core body temperature at each monitoring moment; The discrete event matching module is used to obtain a viscosity correction coefficient based on the core body temperature and adjust the standard pressure conduction delay interval to obtain a corrected conduction delay interval; taking the bands in the arterial blood pressure signal and the local brain oxygen saturation signal as discrete events, using the corrected conduction delay interval as a time constraint, and using the time delay of the discrete events between the two modes as a matching benchmark, to obtain event matching pairs; The cerebrovascular function feature extraction module is used to analyze the area enclosed by the two discrete events in the two-dimensional local phase plane in each event pair to obtain the viscoelastic dissipation area; compare the peak values ​​between the discrete events in each event pair to obtain the amplitude transmission characteristics; and obtain the cerebrovascular fluid resistance index based on all viscoelastic dissipation areas and the correction results of all amplitude transmission characteristics by the viscosity correction coefficient. The assessment and early warning module is used to use arterial blood pressure and cerebral vascular fluid resistance index as two-dimensional data points at each monitoring time, statistically analyze the distribution trend of the two-dimensional data points at continuous monitoring time, and determine the blood pressure early warning signal.

[0005] Furthermore, the methods for acquiring the arterial blood pressure signal and the local cerebral oxygen saturation signal include: Using the monitoring time as the endpoint, arterial blood pressure data within a preset first time period is selected as the arterial blood pressure signal, and local cerebral oxygen saturation data within a preset second time period is selected as the local cerebral oxygen saturation signal, wherein the second length is greater than the first length.

[0006] Furthermore, the method for obtaining the viscosity correction coefficient includes: The viscosity correction coefficient is obtained by using the difference between standard body temperature and core body temperature as a temperature term in the Arrhenius empirical equation.

[0007] Furthermore, the method for obtaining the corrected conduction delay interval includes: The lower and upper limits of the standard pressure transmission delay are multiplied by the viscosity correction coefficient to obtain the corrected transmission delay range.

[0008] Furthermore, the method for obtaining the discrete events includes: For any one of the signals to be processed, namely arterial blood pressure signal and local cerebral oxygen saturation signal, the local maximum point in the signal to be processed is obtained by the first derivative zero-crossing point detection algorithm. The time corresponding to the local maximum point is taken as the time of occurrence of discrete event. With the local maximum point as the center, the waveband corresponding to the discrete event is formed within the preset waveband range.

[0009] Furthermore, the matching method for the event matching pairs includes: A matching matrix is ​​constructed using the index of each discrete event in the arterial blood pressure signal as the row and the index of each discrete event in the local cerebral oxygen saturation signal as the column. If the time delay of discrete events between two modes falls within the corrected propagation delay interval, then the difference between the time delay and the center point of the corrected propagation delay interval is used as the element value of the corresponding element in the matching matrix; otherwise, the element value of the corresponding element in the matching matrix is ​​set to the preset maximum value. The matching matrix is ​​solved using the Hungarian algorithm to obtain the event matching pairs.

[0010] Furthermore, the method for obtaining the viscoelastic dissipation area includes: For each event pair, the band corresponding to the discrete event is decentered to obtain a decentered band. The directed area between two decentered bands is calculated using the discrete Green's formula to obtain the viscoelastic dissipation area.

[0011] Furthermore, the method for obtaining the amplitude propagation characteristics includes: For each event matching pair, the discrete event peak value of the arterial blood pressure signal is used as the denominator, and the discrete event peak value of the local brain oxygen saturation signal is used as the numerator to obtain the amplitude transmission characteristics.

[0012] Furthermore, the method for obtaining the cerebral vascular fluid resistance index includes: The median of all viscoelastic dissipation areas is taken as the overall viscoelastic dissipation area, and the median of all amplitude transmission characteristics is taken as the overall amplitude transmission characteristic. The overall amplitude transmission characteristic is weighted by the viscosity correction coefficient to obtain the weighted overall amplitude transmission characteristic. The cerebral vascular fluid resistance index is obtained based on the weighted overall amplitude transmission characteristics and the overall viscoelastic dissipation area.

[0013] Furthermore, the method for determining the blood pressure warning signal includes: For each monitoring time, the arterial blood pressure at consecutive monitoring times constitutes an arterial blood pressure set; If the range in the arterial blood pressure set is less than or equal to a preset range threshold, then the two-dimensional data points are linearly fitted. If the fitted line shows a downward trend, then an early warning signal to increase perfusion pressure is fed back; if the fitted line shows an upward trend, then an early warning signal to decrease perfusion pressure is fed back; if the fitted line shows a flat trend, then the average cerebral vascular fluid resistance index is calculated. If the average cerebral vascular fluid resistance index is less than a preset resistance threshold, then an early warning signal to maintain perfusion pressure is fed back; if the average cerebral vascular fluid resistance index is not less than the preset resistance threshold, then an early warning signal to excessive resistance is fed back. If the range in the arterial blood pressure set is greater than a preset range threshold, then a quadratic polynomial fitting is performed on the two-dimensional data points to obtain the fitting coefficient of the quadratic term. If the fitting coefficient is greater than 0, then the blood pressure value corresponding to the minimum point of the fitting curve is taken as the optimal blood pressure value. If the fitting coefficient is less than or equal to 0, or the minimum point of the fitting curve exceeds a preset healthy range, then the optimal blood pressure value at the previous monitoring time is taken as the optimal blood pressure value at the current monitoring time. The optimal blood pressure value is fed back as an early warning signal.

[0014] The present invention has the following beneficial effects: To avoid the difficulty in distinguishing between signal attenuation due to blood thickening and biological buffering due to good vascular regulation in hypothermic treatment scenarios, this invention first obtains an adaptively corrected conduction delay interval for the neonatal patient's core body temperature. Then, using this corrected conduction delay interval as a time constraint, and the time delay of discrete events between arterial blood pressure and local cerebral oxygen saturation modes as a matching benchmark, event matching pairs are obtained. This matching method, through time constraints, can isolate the physical rheological factors caused by hypothermia, resulting in a matching result that represents the true functional buffering between arterial blood pressure and cerebral blood flow. Furthermore, the viscoelastic dissipation area is quantified to quantify the energy dissipation generated during the expansion and contraction of cerebral vascular smooth muscle under pressure. Using a viscosity correction coefficient to correct the amplitude conduction characteristics, the pressure conduction characteristics under the current hypothermic state can be restored to the equivalent characteristics under the standard state, thus obtaining the cerebral vascular fluid resistance index at each monitoring moment. A higher cerebral vascular fluid resistance index indicates that the current neonatal arterial blood pressure is less suitable. Therefore, based on the distribution trend of two-dimensional data points composed of arterial blood pressure and the cerebral vascular fluid resistance index, the arterial blood pressure status at the current monitoring moment can be determined, and a blood pressure warning signal can be fed back. This invention uses body temperature data to obtain an adaptive conduction delay interval as a time constraint, and uses a matching method to obtain event matching pairs of the real functional buffering results between arterial blood pressure and cerebral blood flow. Then, it performs vascular functional characteristic analysis to determine a blood pressure warning signal with strong reference value. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a structural block diagram of a multimodal assessment and early warning system for neonatal brain injury provided in one 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 following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multimodal assessment and early warning system for neonatal brain injury proposed according to the present invention. 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.

[0019] The specific scheme of the multimodal assessment and early warning system for neonatal brain injury provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Please see Figure 1 The diagram shows a block diagram of a multimodal assessment and early warning system for neonatal brain injury provided by an embodiment of the present invention. The system includes a multimodal data acquisition module 101, a signal discrete event matching module 102, a cerebrovascular function feature extraction module 103, and an assessment and early warning module 104.

[0021] The system proposed in this embodiment of the invention can be run in existing clinical medical monitoring equipment for neonatal hypoxic-ischemic encephalopathy. The multimodal data acquisition module 101 can call the multimodal monitoring results collected in real time by the medical monitoring equipment through a physical interface. The signal discrete event matching module 102, the cerebrovascular function feature extraction module 103, and the assessment and early warning module 104 can process the collected data through the data processing chip. At the same time, the assessment and early warning module 104 can use the visualization screen or other early warning display device on the medical monitoring equipment to feed back blood pressure early warning signals.

[0022] In the system proposed in this embodiment of the invention, the multimodal data acquisition module 101 is used to acquire arterial blood pressure signals, local cerebral oxygen saturation signals, and core body temperature at each monitoring moment. The arterial blood pressure, local cerebral oxygen saturation, and core body temperature data can all be acquired using existing clinical medical monitoring equipment. The acquisition frequency of arterial blood pressure and local cerebral oxygen saturation can be set to no less than 100Hz to preserve the waveform morphological characteristics.

[0023] It should be noted that, in order to eliminate the influence of dimensions, after obtaining the two signals, the embodiments of the present invention can perform normalization processing separately. For the signal value of each signal point, the standard signal value of its respective dimension is compared with the standard signal value of that dimension to obtain two dimensionless signals for subsequent calculations. The specific standard signal value can be set according to the implementation scenario. In the embodiments of the present invention, the standard signal value of arterial blood pressure can be set to 50 mmHg, and the standard value of cerebral oxygen saturation can be set to 80%.

[0024] Preferably, in some implementations of the present invention, considering the objective time delay in the physical transmission of blood pressure waves in the vascular network, if the input signal is truncated using a time window of completely uniform length, the cerebrovascular response signal triggered by an arterial blood pressure event at the end of the window may be truncated due to exceeding the window range, resulting in a loss of the causal chain in the time domain. Therefore, for arterial blood pressure signals and local cerebral oxygen saturation signals, they should not be completely corresponding in time. Therefore, the methods for acquiring arterial blood pressure signals and local cerebral oxygen saturation signals include: Using the monitoring time as the endpoint, arterial blood pressure data within a preset first time period is selected as the arterial blood pressure signal, and local cerebral oxygen saturation data within a preset second time period is selected as the local cerebral oxygen saturation signal, where the second length is greater than the first length. Through this non-equal-length truncation method, the system aligns potential input stimuli and output responses in the time domain.

[0025] In one specific implementation of this invention, the first length is set to 60 seconds and the second length is set to 65 seconds. It should be noted that the monitoring time in this embodiment can be selected in 10-second increments. Furthermore, in the initial monitoring phase, because the monitoring duration is less than 65 seconds, no data processing is performed in this initial stage; data collection is required to ensure a solid data foundation.

[0026] In some implementations of this invention, considering low-frequency interference such as respiratory movements, mechanical ventilation, and baseline drift, the acquired electrical signals need to be filtered to remove high-frequency noise from the heart rate, thereby extracting only the extremely low-frequency oscillation components that reflect the spontaneous regulatory activity of vascular smooth muscle. Therefore, detrending processing and bandpass filtering are applied to the two signals respectively. As a specific implementation, a fourth-order Butterworth bandpass filter can be used, with the passband cutoff frequency set to 0.05 Hz to 0.15 Hz. In this specific implementation, considering that subsequent modules need to utilize arterial blood pressure at the monitoring time, and that real-time arterial blood pressure may also be subject to drift interference, leading to inaccurate data, the average value of the filtered arterial blood pressure signal can be used as the arterial blood pressure at the current monitoring time for subsequent early warning analysis. Similarly, the average body temperature within the time period corresponding to the arterial blood pressure signal is used as the core temperature.

[0027] The signal discrete event matching module 102 first obtains a viscosity correction coefficient based on the core body temperature and adjusts the standard pressure transmission delay interval to obtain a corrected transmission delay interval. Because low temperatures significantly increase blood viscosity, this rheological change prolongs the transmission time of pressure waves from the aorta to the cerebral microvessels. This embodiment of the invention aims to eliminate this rheological effect of low temperature. Therefore, a viscosity correction coefficient is first obtained based on the core body temperature. This viscosity correction coefficient represents the relationship between the current core body temperature and blood viscosity. The standard pressure transmission delay interval is then adjusted based on this viscosity correction coefficient. The standard pressure transmission delay interval represents the time interval between the minimum and maximum physical transmission delays of pressure waves from the neonatal aorta to the cerebral cortex microvessels under standard body temperature and normal physiological conditions. Therefore, the corrected transmission delay interval obtained after adjustment based on the viscosity correction coefficient characterizes the time delay conditions of pressure transmission that the current target neonate can exhibit at the current monitoring time.

[0028] In existing fluid dynamics model analysis, the observed pressure amplitude attenuation may originate from either good vasodilation or frictional loss due to blood thickening. Since blood thickening caused by low temperature can lead to conduction delay drift, the influence of low temperature can be eliminated by using a modified conduction delay interval as a time constraint. Therefore, the signal discrete event matching module 102 further treats the bands in the arterial blood pressure signal and the local cerebral oxygen saturation signal as discrete events, uses the modified conduction delay interval as a time constraint, and uses the time delay of the discrete events between the two modes as a matching benchmark. That is, only when the time delay is within the time constraint condition does it conform to the actual physiological changes in vascular function. Furthermore, event matching pairs are obtained through one-to-one matching. That is, the event matching pair contains a discrete event of arterial blood pressure and a discrete event of local cerebral oxygen saturation. The two discrete events are considered as a set of events with a causal relationship that can characterize the vascular functional state.

[0029] Preferably, in this embodiment of the invention, the method for obtaining the viscosity correction coefficient includes: The viscosity correction coefficient is obtained by using the difference between standard body temperature and core body temperature as a temperature term in the Arrhenius empirical equation. The Arrhenius empirical equation demonstrates that liquid viscosity decreases exponentially with increasing temperature and increases exponentially with decreasing temperature, thus reflecting the effect of current temperature on liquid viscosity. It should be noted that the classic Arrhenius empirical equation is in the following form: ;in Let be the viscosity correction factor at the i-th monitoring time. is the pre-exponential factor, R is the gas constant, E is the activation energy, T is the absolute temperature, and exp is the exponential function with the natural constant as the base.

[0030] The equation form used in the specific implementation of this invention is as follows: ;in Let be the core body temperature at the i-th monitoring time, and 37 represent the standard physiological temperature. This is the temperature-sensitive constant for blood rheology. Unlike the classical equation, the form used in this embodiment of the invention is a simplified form of the classical equation, which incorporates elements from the classical equation... Performing a Taylor expansion and substituting the result into the classical equation yields a simplified approximation, namely... In the embodiments of the present invention The final value is 0.05. This simplified form is suitable for scenarios with small temperature variations, such as the neonatal treatment scenario in this embodiment of the invention. It can quickly estimate the final viscosity correction coefficient. That is, when the body temperature is maintained at 37 degrees Celsius, the viscosity correction coefficient is 1, indicating that the viscosity does not need to be corrected and the body temperature does not affect the blood viscosity. As the body temperature decreases, the viscosity correction coefficient increases exponentially. The specific meaning of the equation is a technical means well known to those skilled in the art and will not be elaborated or limited here.

[0031] Preferably, in this embodiment of the invention, the method for obtaining the modified conduction delay interval includes: The corrected transmission delay interval is obtained by multiplying the lower and upper limits of the standard pressure transmission delay by the viscosity correction factor. Since the viscosity correction factor obtained using the Arrhenius empirical equation is a dimensionless value greater than or equal to 1, it can be directly used as a multiplier and multiplied by the corresponding upper and lower limits to complete the correction.

[0032] Specifically, in one implementation of the present invention, under standard body temperature of 37°C and normal physiological conditions, the minimum physical transmission delay of pressure wave transmission from the neonatal aorta to the microvessels of the cerebral cortex is set to be 0.5 seconds and the maximum physical transmission delay is 3 seconds, that is, the lower limit of the standard pressure transmission delay is 0.5 seconds and the upper limit is 3 seconds.

[0033] Preferably, in this embodiment of the invention, the method for obtaining discrete events includes: For any one of the signals to be processed, namely arterial blood pressure signal and local cerebral oxygen saturation signal, a first derivative zero-crossing detection algorithm is used to obtain the local maximum point in the signal to be processed. That is, the local maximum point is the peak point within a band. Therefore, the time corresponding to the local maximum point is taken as the time of occurrence of the discrete event. With the local maximum point as the center, a band corresponding to the discrete event is formed within a preset band range. In this embodiment of the invention, the preset band range can be set to 5 seconds. If the signal time sequence boundary is exceeded within the range, an interpolation algorithm can be used to complete it, and finally multiple discrete events with the same time range are obtained. The specific interpolation algorithm and the first derivative zero-crossing detection algorithm are all technical means well known to those skilled in the art, and will not be described in detail here.

[0034] Preferably, in this embodiment of the invention, the event matching method includes: A matching matrix is ​​constructed using the index of each discrete event in the arterial blood pressure signal as the row and the index of each discrete event in the local cerebral oxygen saturation signal as the column. If the time delay of discrete events between two modes falls within the corrected conduction delay interval, it indicates that the two discrete events are a combination of event responses caused by changes in vascular function. Therefore, it is possible to further analyze whether the time delay between the two is optimal, and the difference between the time delay and the center point of the corrected conduction delay interval is used as the element value of the corresponding element in the matching matrix. Otherwise, it indicates that it may be a combination of non-vascular functional response events affected by low temperature or other situations that clearly do not conform to actual physiological laws. In this case, the element value of the corresponding element in the matching matrix is ​​set to a preset maximum value. In this embodiment of the invention, the preset maximum value can be directly set to infinity.

[0035] The matching matrix is ​​solved using the Hungarian algorithm, and the event matching pairs can be obtained by finding the bipartite graph matching scheme with the minimum total cost. The Hungarian algorithm is used instead of simple nearest neighbor matching because it can utilize global information and maintain the correct alignment of the overall causal sequence even in the presence of artifact peak insertion or missing true peaks, thus avoiding cascading misjudgments. The Hungarian algorithm includes steps such as matrix construction and preprocessing. It constructs the matrix by inserting virtual rows or columns, and the specific solution method is a technique well-known to those skilled in the art, which will not be elaborated or limited here.

[0036] It should be noted that the difference between the time delay and the center point of the corrected conduction delay interval is the square of the difference. If the final number of event matching pairs is empty, or less than the preset threshold, it indicates that the data at this monitoring time is unreliable, the current monitoring time is invalid, and further calculations are not required. The threshold can be set comprehensively based on multiple dimensions such as signal length and accuracy requirements in the specific implementation scenario, which will not be elaborated here.

[0037] As a bio-viscoelastic material, cerebral vascular smooth muscle undergoes stress relaxation during expansion and contraction under pressure. Its pressure-volume trajectory does not coincide during loading and unloading, forming a hysteresis loop. The area of ​​this loop represents energy dissipation. When vascular regulation is impaired (vasoparalysis), smooth muscle tension is lost, and the blood vessel degenerates into a simple elastic conduit, eliminating the hysteresis phenomenon. Therefore, in the cerebral vascular function feature extraction module, the area enclosed by the two discrete events in each event pair within the two-dimensional local phase plane is first analyzed to obtain the viscoelastic dissipation area. That is, the larger the viscoelastic dissipation area, the more normal the neonatal vascular function, the better the stress relaxation, and the better the tension of the vascular smooth muscle.

[0038] Further comparison of peak values ​​between discrete events in each event pair yields amplitude transmission characteristics. Since arterial blood pressure acts as the driving force for expansion and contraction, and local cerebral oxygen saturation is the response to this force, if vascular function is normal, the cerebral blood vessels will exhibit a smaller local cerebral oxygen saturation peak value due to stress relaxation caused by the generation of this force. Conversely, the closer the local cerebral oxygen saturation is to arterial blood pressure, the greater the pressure penetration rate of arterial blood pressure, indicating a failure of cerebral vascular fluid resistance regulation. Therefore, all viscoelastic dissipation areas can be statistically analyzed, and viscosity correction coefficients can be used to correct all amplitude transmission characteristics, restoring the true pressure penetration rate masked by low temperatures, thus obtaining the cerebral vascular fluid resistance index. The smaller the index, the stronger the vascular buffering capacity; the larger the index, the more likely the newborn's blood vessels are in a passively dilated state due to severe hypoxia and ischemia, exhibiting a pure elastomer state with weakened vascular regulation and failed resistance regulation.

[0039] Preferably, in this embodiment of the invention, the method for obtaining the viscoelastic dissipation area includes: For each event matching pair, the band corresponding to the discrete event is decentered to obtain a decentered band. The purpose of decentering is to eliminate the interference of baseline drift on the origin of the phase plane. This can be achieved by directly subtracting the average signal value of the band from each signal point on the band corresponding to the discrete event.

[0040] The directed area between two decentralized bands is calculated using the Discrete Green's formula to obtain the viscoelastic dissipation area. The Discrete Green's formula is widely used in vascular mechanics for calculating the hysteresis loop area, and its formula can be expressed as: ;in Match the viscoelastic dissipation area for the k-th event. To match the signal value of the j-th signal point in the decentralized band of the middle artery blood pressure signal for the k-th event, Let j+1 be the signal value of the (j+1)th signal point in the decentralized band of the local brain oxygen saturation signal in the matching pair of the kth event. To match the signal value of the (j+1)th signal point in the decentralized band of the middle artery blood pressure signal for the k-th event, Let J be the signal value of the j-th signal point in the decentered band of the local brain oxygen saturation signal in the k-th event matching pair, where J is the number of signal points in the decentered band.

[0041] The Discrete Green's Theorem is a well-known technique in the field, and its principles will not be elaborated here.

[0042] Preferably, in this embodiment of the invention, the method for obtaining wave amplitude propagation characteristics includes: For each event pair, the discrete event peak value of the arterial blood pressure signal is used as the denominator, and the discrete event peak value of the local cerebral oxygen saturation signal is used as the numerator to obtain the amplitude conduction characteristic. That is, the larger the final amplitude conduction characteristic, the worse the vascular resistance regulation function.

[0043] Preferably, in this embodiment of the invention, the method for obtaining the cerebral vascular fluid resistance index includes: The median of all viscoelastic dissipation areas is used as the overall viscoelastic dissipation area, and the median of all amplitude transmission characteristics is used as the overall amplitude transmission characteristic. By using median statistics, we can find overall characteristic data that can characterize the overall result among all event matching pairs obtained at the current monitoring time.

[0044] The overall amplitude conduction characteristics are weighted using a viscosity correction coefficient to obtain a weighted overall amplitude conduction characteristic. By weighting with the viscosity correction coefficient, the overall amplitude conduction characteristics are amplified, restoring the pressure conductance under the current low temperature and high viscosity state to the equivalent conductance at standard body temperature (37℃), thus eliminating the physical signal attenuation caused by blood thickening.

[0045] The cerebrovascular resistance index is obtained based on the weighted overall amplitude transmission characteristics and the overall viscoelastic dissipation area. Specifically, a larger weighted overall amplitude transmission characteristic and a smaller overall viscoelastic dissipation area indicate poor cerebrovascular resistance regulation function in the newborn at the current monitoring time, resulting in a higher cerebrovascular resistance index.

[0046] As a specific example, in one implementation of this invention, the formula for obtaining the cerebral vascular fluid resistance index includes: ;in Let be the cerebral vascular fluid resistance index at the i-th monitoring time. The overall amplitude propagation characteristics at the i-th monitoring time are... Let be the viscosity correction factor at the i-th monitoring time. To preset the normalization factor, Let be the total viscoelastic dissipation area at the i-th monitoring time.

[0047] In the above formula, the blood viscosity correction coefficient at low temperatures is greater than 1. The numerator term amplifies the overall amplitude transmission characteristics through multiplication, eliminating the physical signal attenuation caused by blood thickening. The denominator term uses the viscoelastic dissipation area to characterize vascular regulatory activity, and the preset normalization factor is chosen based on the typical physiological state of a healthy, full-term newborn. It can be approximately equal to 1. When vascular regulation is good, the viscoelastic dissipation area is large, and the denominator is significantly greater than 1, resulting in a smaller final cerebral vascular fluid resistance index, which characterizes strong vascular buffering capacity. Conversely, if... When the value approaches 0, the denominator approaches 1. At this point, the cerebrovascular fluid resistance index is completely dominated by the high permeability of the molecule, and the value is significantly larger, indicating that the regulation of vascular resistance has failed.

[0048] It should be noted that if the final cerebral vascular fluid resistance coefficient is significantly abnormal, it indicates that the data at the current monitoring time is greatly affected by noise and is unreliable. In this case, subsequent calculations should be stopped, and the data should be marked as invalid. A suitable physiological range can be set, such as 0 to 5. If the cerebral vascular fluid resistance coefficient exceeds this physiologically suitable range, it should be marked as invalid.

[0049] The assessment and early warning module 104 is used to comprehensively analyze all the feature data obtained from the system. It uses arterial blood pressure and cerebrovascular resistance index at each monitoring time as two-dimensional data points. The horizontal axis of these data points can be set to arterial blood pressure, and the vertical axis to cerebrovascular resistance index. The higher the two-dimensional data point is on the coordinate system, the worse the neonatal's vascular function, requiring blood pressure adjustment intervention. Furthermore, during a longer monitoring and treatment process, neonates will experience natural fluctuations of 5-15 mmHg over continuous monitoring periods due to factors such as changes in body odor, metabolism, and self-regulation. The trend of these blood pressure fluctuations, reflected in the two-dimensional data points, can further determine the appropriate blood pressure adjustment trend for the current neonate. Therefore, the distribution trend of the two-dimensional data points over continuous monitoring times can be statistically analyzed to determine blood pressure early warning signals.

[0050] Preferably, in this embodiment of the invention, considering that the two-dimensional data points presented during continuous monitoring have multiple states of data representation in actual monitoring, in order to obtain more effective early warning signals, the early warning signals are specifically divided into multiple scenarios based on the changes in the neonatal's arterial blood pressure during the process, specifically including: For each monitoring moment, the arterial blood pressure at consecutive monitoring moments constitutes an arterial blood pressure set. It should be noted that, in this embodiment of the invention, the consecutive monitoring moments constituting the arterial blood pressure set can be the arterial blood pressure at all monitoring moments before the current monitoring moment; alternatively, the system can maintain a first-in-first-out queue of fixed length to store valid monitoring moments in the queue. If the data volume in the queue is not full, it can prompt "Data accumulation in progress" without providing a warning signal, and proceed with subsequent steps only after the data volume in the queue is sufficient.

[0051] Over a relatively long blood pressure range, the regulatory capacity of vascular resistance exhibits a U-shaped change as blood pressure increases. Specifically, only within the appropriate blood pressure range does it show a relatively low cerebrovascular resistance index. Within the smaller blood pressure range before the appropriate range, the cerebrovascular resistance index shows a decreasing trend, while within the larger blood pressure range after the appropriate range, it shows a continuously increasing trend. Therefore, if the complete change process can be observed during monitoring, the appropriate blood pressure for the target newborn can be obtained by fitting two-dimensional data points, and intervention can be implemented.

[0052] However, in reality, individual factors may play a role. Newborns may be deeply sedated and unaffected by external blood pressure fluctuations, maintaining stable blood pressure throughout the treatment and monitoring process. This can result in the obtained two-dimensional data points not fully covering the aforementioned "U"-shaped curve range, potentially remaining within a smaller, more suitable, or even a larger blood pressure range. Therefore, these scenarios should be considered separately.

[0053] Therefore, in this embodiment of the invention, the range of the arterial blood pressure set is statistically analyzed. The range represents the magnitude of data change. If the range is small, it indicates that the neonatal blood pressure range is small and cannot fully cover the "U"-shaped curve range. In this case, polynomial fitting should not be performed directly, but linear fitting should be performed, and the current neonatal state is determined based on the linear fitting result. Therefore, if the range of the arterial blood pressure set is less than or equal to the preset range threshold, linear fitting is performed on the two-dimensional data points. If the fitted line shows a downward trend, it indicates that the neonatal is exhibiting the first half of the "U"-shaped curve, and the cerebral vascular fluid resistance will continue to decrease as blood pressure increases. Therefore, an early warning signal for increasing perfusion pressure can be fed back. If the fitted line shows an upward trend, it indicates that the neonatal is exhibiting the second half of the "U"-shaped curve, and the cerebral vascular fluid resistance will continue to increase as blood pressure increases. Therefore, an early warning signal for decreasing perfusion pressure can be fed back. If the fitted straight line shows a flat trend, it cannot reflect a significant change trend. Therefore, it is necessary to calculate the average cerebral vascular fluid resistance index. If the average cerebral vascular fluid resistance index is less than the preset resistance threshold, it indicates that the average cerebral vascular fluid resistance index is small and the current vascular resistance regulation ability is strong, which can maintain the perfusion pressure. If the average cerebral vascular fluid resistance index is not less than the preset resistance threshold, it indicates that the average cerebral vascular fluid resistance index is large and the current blood pressure range is inappropriate. It is necessary to remind medical staff to adjust the blood pressure according to the specific situation. The feedback of the high resistance warning signal is used to remind medical staff.

[0054] If the range of the arterial blood pressure set is greater than a preset range threshold, it indicates that the range of arterial blood pressure is large. Polynomial fitting can be used for analysis. Therefore, a quadratic polynomial fitting is performed on the two-dimensional data points to obtain the fitting coefficient of the quadratic term. If the fitting coefficient is greater than 0, it indicates that the fitting curve conforms to the characteristic of opening upwards, and the blood pressure value corresponding to the minimum point of the fitting curve is taken as the optimal blood pressure value. If the fitting coefficient is less than or equal to 0, or the minimum point of the fitting curve exceeds a preset healthy range, it indicates that the current data does not show typical regulatory characteristics, and the optimal blood pressure value at the previous monitoring time can be taken as the optimal blood pressure value at the current monitoring time. Furthermore, the optimal blood pressure value can be fed back as an early warning signal, reminding medical staff to intervene in neonatal patients based on this optimal blood pressure value to maintain the infant's cerebral blood flow in an optimal regulatory state.

[0055] In this embodiment of the invention, the range threshold can be set according to the typical fluctuation range of the neonatal mean arterial pressure, and can be set between 5 and 8. In this embodiment, it can be specifically set to 5. The method for judging the three trends of decreasing, increasing, and flattening of the fitted straight line obtained by linear fitting can also be judged using the slope threshold. If the slope is less than the negative slope threshold, it indicates a decreasing trend; if the slope is greater than the slope threshold, it indicates an increasing trend; if the absolute value of the slope is less than the slope threshold, it indicates a flattening trend. The slope threshold can be set between 0.01 and 0.05, and in this embodiment, it is set to 0.05. Since the cerebrovascular fluid resistance obtained in a specific implementation of this embodiment is close to 1 in healthy newborns, the resistance threshold can be set between 1.2 and 1.5, and can be specifically set to 1.3. The healthy range can be set between 30 mmHg and 80 mmHg. Both linear fitting and quadratic polynomial fitting can be implemented using the least squares method. The specific fitting method is a well-known technique to those skilled in the art and will not be described in detail here.

[0056] In summary, this invention, at each monitoring time, adjusts the standard pressure conduction delay interval based on neonatal core body temperature to obtain a corrected conduction delay interval as a time constraint to match discrete events in arterial blood pressure signals and local cerebral oxygen saturation signals, resulting in event matching pairs. Feature extraction is performed on all event matching pairs to obtain the cerebral vascular resistance index at each monitoring time. Using arterial blood pressure and the cerebral vascular resistance index at each monitoring time as two-dimensional data points, the distribution trend of these two-dimensional data points across continuous monitoring times is statistically analyzed to determine a blood pressure warning signal. This invention uses body temperature data to obtain an adaptive conduction delay interval as a time constraint, utilizes a matching method to obtain event matching pairs representing the true functional buffering results between arterial blood pressure and cerebral blood flow, and then performs vascular functional feature analysis to determine a highly reliable blood pressure warning signal.

[0057] 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.

[0058] 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 multimodal assessment and early warning system for neonatal brain injury, characterized in that, The system includes: The multimodal data acquisition module is used to acquire arterial blood pressure signals, local cerebral oxygen saturation signals, and core body temperature at each monitoring moment; The discrete event matching module is used to obtain a viscosity correction coefficient based on the core body temperature and adjust the standard pressure conduction delay interval to obtain a corrected conduction delay interval; taking the bands in the arterial blood pressure signal and the local brain oxygen saturation signal as discrete events, using the corrected conduction delay interval as a time constraint, and using the time delay of the discrete events between the two modes as a matching benchmark, to obtain event matching pairs; The cerebrovascular function feature extraction module is used to analyze the area enclosed by the two discrete events in the two-dimensional local phase plane in each event pair to obtain the viscoelastic dissipation area; compare the peak values ​​between the discrete events in each event pair to obtain the amplitude transmission characteristics; and obtain the cerebrovascular fluid resistance index based on all viscoelastic dissipation areas and the correction results of all amplitude transmission characteristics by the viscosity correction coefficient. The assessment and early warning module is used to use arterial blood pressure and cerebral vascular fluid resistance index as two-dimensional data points at each monitoring time, statistically analyze the distribution trend of the two-dimensional data points at continuous monitoring time, and determine the blood pressure early warning signal.

2. The neonatal brain injury multimodal assessment and early warning system according to claim 1, characterized in that, The methods for acquiring the arterial blood pressure signal and the local cerebral oxygen saturation signal include: Using the monitoring time as the endpoint, arterial blood pressure data within a preset first time period is selected as the arterial blood pressure signal, and local cerebral oxygen saturation data within a preset second time period is selected as the local cerebral oxygen saturation signal, wherein the second length is greater than the first length.

3. The neonatal brain injury multimodal assessment and early warning system according to claim 1, characterized in that, The method for obtaining the viscosity correction coefficient includes: The viscosity correction coefficient is obtained by using the difference between standard body temperature and core body temperature as a temperature term in the Arrhenius empirical equation.

4. The neonatal brain injury multimodal assessment and early warning system according to claim 3, characterized in that, The method for obtaining the corrected conduction delay interval includes: The lower and upper limits of the standard pressure transmission delay are multiplied by the viscosity correction coefficient to obtain the corrected transmission delay range.

5. A multimodal assessment and early warning system for neonatal brain injury according to claim 1, characterized in that, The method for obtaining the discrete events includes: For any one of the signals to be processed, namely arterial blood pressure signal and local cerebral oxygen saturation signal, the local maximum point in the signal to be processed is obtained by the first derivative zero-crossing point detection algorithm. The time corresponding to the local maximum point is taken as the time of occurrence of discrete event. With the local maximum point as the center, the waveband corresponding to the discrete event is formed within the preset waveband range.

6. The neonatal brain injury multimodal assessment and early warning system according to claim 1, characterized in that, The matching methods for the event matching pairs include: A matching matrix is ​​constructed using the index of each discrete event in the arterial blood pressure signal as the row and the index of each discrete event in the local cerebral oxygen saturation signal as the column. If the time delay of discrete events between two modes falls within the corrected propagation delay interval, then the difference between the time delay and the center point of the corrected propagation delay interval is used as the element value of the corresponding element in the matching matrix; otherwise, the element value of the corresponding element in the matching matrix is ​​set to the preset maximum value. The matching matrix is ​​solved using the Hungarian algorithm to obtain the event matching pairs.

7. The neonatal brain injury multimodal assessment and early warning system according to claim 1, characterized in that, The method for obtaining the viscoelastic dissipation area includes: For each event pair, the band corresponding to the discrete event is decentered to obtain a decentered band. The directed area between two decentered bands is calculated using the discrete Green's formula to obtain the viscoelastic dissipation area.

8. A multimodal assessment and early warning system for neonatal brain injury according to claim 1, characterized in that, The method for obtaining the amplitude propagation characteristics includes: For each event matching pair, the discrete event peak value of the arterial blood pressure signal is used as the denominator, and the discrete event peak value of the local brain oxygen saturation signal is used as the numerator to obtain the amplitude transmission characteristics.

9. A multimodal assessment and early warning system for neonatal brain injury according to claim 8, characterized in that, The method for obtaining the cerebral vascular fluid resistance index includes: The median of all viscoelastic dissipation areas is taken as the overall viscoelastic dissipation area, and the median of all amplitude transmission characteristics is taken as the overall amplitude transmission characteristic. The overall amplitude transmission characteristic is weighted by the viscosity correction coefficient to obtain the weighted overall amplitude transmission characteristic. The cerebral vascular fluid resistance index is obtained based on the weighted overall amplitude transmission characteristics and the overall viscoelastic dissipation area.

10. A multimodal assessment and early warning system for neonatal brain injury according to claim 1, characterized in that, The method for determining the blood pressure warning signal includes: For each monitoring time, the arterial blood pressure at consecutive monitoring times constitutes an arterial blood pressure set; If the range in the arterial blood pressure set is less than or equal to a preset range threshold, then the two-dimensional data points are linearly fitted. If the fitted line shows a downward trend, then an early warning signal to increase perfusion pressure is fed back; if the fitted line shows an upward trend, then an early warning signal to decrease perfusion pressure is fed back; if the fitted line shows a flat trend, then the average cerebral vascular fluid resistance index is calculated. If the average cerebral vascular fluid resistance index is less than a preset resistance threshold, then an early warning signal to maintain perfusion pressure is fed back; if the average cerebral vascular fluid resistance index is not less than the preset resistance threshold, then an early warning signal to excessive resistance is fed back. If the range in the arterial blood pressure set is greater than a preset range threshold, then a quadratic polynomial fitting is performed on the two-dimensional data points to obtain the fitting coefficient of the quadratic term. If the fitting coefficient is greater than 0, then the blood pressure value corresponding to the minimum point of the fitting curve is taken as the optimal blood pressure value. If the fitting coefficient is less than or equal to 0, or the minimum point of the fitting curve exceeds a preset healthy range, then the optimal blood pressure value at the previous monitoring time is taken as the optimal blood pressure value at the current monitoring time. The optimal blood pressure value is fed back as an early warning signal.

Images