Intracranial pressure dynamic multi-parameter fusion monitoring method and system for patient

Abstract

This invention provides a method and system for dynamic multi-parameter fusion monitoring of intracranial pressure in patients, belonging to the field of medical monitoring technology. The method includes: acquiring raw multi-parameter monitoring data of a target patient during the postoperative monitoring period; performing time-series alignment processing on the raw multi-parameter monitoring data to unify the timestamps of each parameter data, obtaining time-synchronized multi-parameter time-series data; constructing a multi-dimensional feature space to characterize the correlation between each monitoring parameter based on the time-synchronized multi-parameter time-series data, and generating a continuous parameter mapping surface within the multi-dimensional feature space; and obtaining multiple feature sub-regions with differentiated data distribution characteristics by adaptively dividing the parameter mapping surface. This invention enables timely identification and accurate graded early warning of abnormal intracranial pressure states.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring technology, and in particular to a method and system for dynamic multi-parameter fusion monitoring of intracranial pressure in patients. Background Technology

[0002] Intracranial pressure monitoring is a crucial part of postoperative care for traumatic brain injury, cerebrovascular disease, etc., and is directly related to patient prognosis. However, existing monitoring technologies mostly rely on a single intracranial pressure parameter or simple superposition of multiple parameters, which has not solved the problem of temporal misalignment caused by the difference in the original sampling frequency of different physiological parameters. Furthermore, the fixed fusion weights and static threshold judgments make it difficult to adapt to the individual pathological characteristics of patients and the risk changes at different time windows after surgery.

[0003] For example, in patients with severe traumatic brain injury after surgery, existing systems simultaneously collect data such as intracranial pressure, cerebral oxygen saturation, and mean arterial pressure. However, due to the lack of temporal alignment and direct fusion, and the absence of dynamic weight adjustment based on the patient's preoperative medical history, when synergistic abnormalities such as a slow decline in cerebral oxygen saturation and a critical increase in intracranial pressure occur during postoperative recovery, these abnormalities are not identified in time due to parameter temporal separation and fixed weights. The warning is only triggered when intracranial pressure rises, delaying early intervention for related complications. This case exposes the core defects of existing technologies, such as insufficient temporal synchronization of multiple parameters, lack of adaptation to individual pathological characteristics, and a lack of dynamic weight fusion mechanisms. This results in one-sided monitoring results and delayed warnings, failing to meet the needs of refined clinical monitoring. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for dynamic multi-parameter fusion monitoring of intracranial pressure in patients, so as to realize timely identification and accurate classification and early warning of abnormal intracranial pressure.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients, the method comprising:

[0007] Acquire raw multi-parameter monitoring data of the target patient during the postoperative monitoring period;

[0008] Based on the original monitoring data of multiple parameters, time-series alignment processing is performed to unify the timestamps of each parameter data, so as to obtain time-synchronized multi-parameter time-series data.

[0009] Based on time-synchronized multi-parameter time-series data, a multi-dimensional feature space is constructed to characterize the correlation between various monitoring parameters, and a continuous parameter mapping surface is generated within the multi-dimensional feature space. By adaptively dividing the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained.

[0010] After mapping the real-time acquired time-series data points to the corresponding feature sub-regions, dynamic weighted fusion calculation is performed by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status.

[0011] Based on the temporal fusion monitoring values, the preset threshold interval of the corresponding postoperative time window is first constructed as a multidimensional geometric feature region. The temporal fusion monitoring values ​​at each time point are converted into multidimensional geometric feature points, and it is determined whether the geometric feature points are within the preset multidimensional geometric feature region. The fusion monitoring values ​​are then compared with the preset threshold interval of the corresponding time window to obtain the comparison results.

[0012] Based on the comparison results, when the fused monitoring value exceeds the preset threshold range, a graded early warning information corresponding to the degree of exceedance is generated and output.

[0013] Furthermore, based on the original multi-parameter monitoring data, time-series alignment processing is performed to unify the timestamps of each parameter, resulting in time-synchronized multi-parameter time-series data, including:

[0014] Analyze the raw monitoring data of multiple parameters to identify the original sampling frequency of the raw time series data stream corresponding to different physiological parameters;

[0015] Based on the original sampling frequency, and using a unified target frequency dynamically set according to clinical monitoring needs as a benchmark, the original time series data streams of each parameter are resampled using an adaptive interpolation algorithm to align the sampling times of all parameters to the same time series.

[0016] Based on a unified time series, a time-synchronized multi-parameter time series data is constructed, which includes synchronized data columns for each parameter.

[0017] Furthermore, based on time-synchronized multi-parameter time-series data, a multi-dimensional feature space is constructed to characterize the correlation between various monitoring parameters, and a continuous parameter mapping surface is generated within the multi-dimensional feature space. By adaptively partitioning the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained, including:

[0018] Receive time-synchronized multi-parameter time-series data, and construct a multi-dimensional feature space based on the physiological significance and clinical relevance of each monitoring parameter. Each dimension corresponds to a monitoring parameter, so that each monitoring parameter forms a spatial distribution point set in the multi-dimensional feature space.

[0019] Based on the spatial distribution point set formed in the multidimensional feature space, a nonlinear mapping algorithm is used to perform surface fitting on the spatial distribution point set to obtain a continuous parametric mapping surface.

[0020] Based on the parametric mapping surface, the geometric features and data distribution characteristics of the parametric mapping surface are analyzed. According to the surface curvature change gradient, data density distribution gradient and clinical risk distribution characteristics, the parametric mapping surface is adaptively divided into regions to obtain multiple feature sub-regions with differentiated data distribution characteristics.

[0021] Furthermore, after mapping the real-time acquired time-series data points to corresponding feature sub-regions, and combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights, a dynamic weighted fusion calculation is performed to finally obtain the time-series fusion monitoring values ​​used to characterize the overall intracranial pressure status, including:

[0022] The system receives the current time data point from the time-synchronized multi-parameter time series data in real time. Based on multiple feature sub-regions with differentiated data distribution characteristics, it maps the current time data point to the corresponding feature sub-region to obtain the data point mapping result.

[0023] Based on the data point mapping results, the statistical characteristics of the feature sub-regions that have been calculated are extracted, including the data distribution center, dispersion and boundary morphology features. At the same time, combined with the individual pathological characteristics of the target patient entered before the operation and the parameter weight configuration preset by the clinical experts, a basic parameter set for dynamic weighted fusion calculation is constructed.

[0024] Based on the basic parameter set, the fusion weight coefficients of each monitoring parameter are dynamically adjusted according to the relative position of the data points in the feature sub-region, the degree of matching of individual pathological features, and the priority of clinical parameter weights. Weighted fusion calculation is then performed to obtain the fusion value of intracranial pressure status at the current moment.

[0025] Based on the current intracranial pressure status fusion value, combined with the fusion value sequence of historical time points, temporal smoothing and trend analysis are performed, and finally, a temporal fusion monitoring value is output to characterize the overall intracranial pressure status.

[0026] Furthermore, based on the temporal fusion monitoring values, a preset threshold interval corresponding to the postoperative time window is first constructed as a multidimensional geometric feature region. The temporal fusion monitoring values ​​at each time point are then converted into multidimensional geometric feature points, and it is determined whether the geometric feature points are within the preset multidimensional geometric feature region. This process completes the comparison between the fusion monitoring values ​​and the preset threshold interval of the corresponding time window, yielding the comparison results, including:

[0027] Receive time-series fusion monitoring values, identify the postoperative time window to which the current moment belongs according to the postoperative time window division rules, and retrieve the corresponding preset threshold interval parameters based on the time window;

[0028] Based on preset threshold range parameters, a multidimensional geometric feature region is constructed, where each dimension of the multidimensional geometric feature region corresponds to a clinical risk assessment dimension.

[0029] The time-series fusion monitoring values ​​are decomposed and mapped according to the clinical risk assessment dimensions to generate corresponding multi-dimensional geometric feature points. The coordinate values ​​of the multi-dimensional geometric feature points in each dimension are determined by the components of the time-series fusion monitoring values ​​in the corresponding risk dimensions.

[0030] Based on multidimensional geometric feature regions and multidimensional geometric feature points, an algorithm for detecting whether a point is inside a polygon is used to determine the spatial positional relationship of the multidimensional geometric feature points relative to the multidimensional geometric feature regions, so as to obtain the judgment result, and generate the comparison result based on the judgment result.

[0031] Furthermore, based on multidimensional geometric feature regions and multidimensional geometric feature points, a point-within-a-polygon detection algorithm is used to determine the spatial positional relationship of the multidimensional geometric feature points relative to the multidimensional geometric feature regions, thereby obtaining a judgment result. A comparison result is then generated based on the judgment result, including:

[0032] Receive boundary coordinate data of multidimensional geometric feature regions and coordinate data of multidimensional geometric feature points, perform geometric normalization processing on the boundary coordinate data of multidimensional geometric feature regions, and obtain a normalized polygon boundary vertex sequence;

[0033] Based on the standardized polygon boundary vertex sequence and coordinate data of multidimensional geometric feature points, a ray intersection counting algorithm is used to calculate the number of intersections between the test ray emanating from the multidimensional geometric feature point and the polygon boundary edge, and the result of the intersection count is obtained.

[0034] Based on the crossover count, the parity of the crossover count determines whether the multidimensional geometric feature point is located within the multidimensional geometric feature region. At the same time, the minimum Euclidean distance from the multidimensional geometric feature point to the polygon boundary is calculated to obtain the spatial location judgment result.

[0035] Based on the spatial location judgment results, and by comparing the minimum Euclidean distance with the preset critical distance threshold, the spatial location judgment results are mapped to clinically interpretable state identifiers, resulting in comparison results that include normal state, critical state, and abnormal state identifiers.

[0036] Furthermore, based on the comparison results, when the fused monitoring value exceeds a preset threshold range, a graded early warning message corresponding to the degree of exceedance is generated and output, including:

[0037] Receive the comparison results, analyze the clinical status identifiers contained in the comparison results, and identify whether the current intracranial pressure status is abnormal.

[0038] Based on the identified abnormal states, retrieve the coordinate data of multidimensional geometric feature points and the boundary data of multidimensional geometric feature regions, and calculate the distance and direction of the multidimensional geometric feature points exceeding the boundary of the multidimensional geometric feature regions;

[0039] Based on the distance and direction of the excess, combined with the individual pathological characteristics of the target patient and the clinical risk level of the postoperative time window, the clinical severity of intracranial pressure abnormality is assessed to obtain a quantitative index of the degree of excess.

[0040] Based on the quantitative indicators of the degree of exceedance, and according to the preset warning classification rule base, the quantitative indicators of the degree of exceedance are mapped to the corresponding warning levels, generating classified warning information that includes the warning level, risk description and suggested handling measures.

[0041] Based on the graded early warning information, the graded early warning information is transmitted to the clinical monitoring terminal in real time through the medical data communication interface, and the timestamp of the early warning event, the early warning level and the patient identification information are recorded.

[0042] Secondly, the patient's intracranial pressure dynamic multi-parameter fusion monitoring system includes:

[0043] The acquisition module is used to acquire raw monitoring data of multiple parameters of the target patient during the postoperative monitoring period;

[0044] The synchronization module is used to perform time-series alignment processing on the original monitoring data of multiple parameters to unify the timestamps of each parameter data and obtain time-synchronized multi-parameter time-series data.

[0045] The mapping module is used to construct a multi-dimensional feature space based on time-synchronized multi-parameter time-series data to characterize the correlation between various monitoring parameters, and to generate a continuous parameter mapping surface within the multi-dimensional feature space; by adaptively dividing the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained.

[0046] The fusion module is used to map real-time acquired time-series data points to corresponding feature sub-regions, and then perform dynamic weighted fusion calculations by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status.

[0047] The comparison module is used to construct a multi-dimensional geometric feature region based on the temporal fusion monitoring value, first construct the preset threshold interval of the corresponding postoperative time window as a multi-dimensional geometric feature region, convert the temporal fusion monitoring value at each time into multi-dimensional geometric feature points, and determine whether the geometric feature points are within the preset multi-dimensional geometric feature region, and complete the comparison between the fusion monitoring value and the preset threshold interval of the corresponding time window to obtain the comparison result.

[0048] The processing module is used to generate and output graded early warning information corresponding to the degree of exceedance when the fused monitoring value exceeds the preset threshold range based on the comparison results.

[0049] Thirdly, a computing device, comprising:

[0050] One or more processors;

[0051] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0052] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0053] The above-described solution of the present invention has at least the following beneficial effects:

[0054] This study employs various techniques, including multi-parameter temporal alignment processing, multi-dimensional feature space construction and adaptive sub-region division, dynamic weighted fusion calculation combining individual pathological features, multi-dimensional geometric feature region comparison across time windows, and a graded early warning mechanism based on the degree of exceedance. These techniques overcome the technical problems of insufficient multi-parameter temporal synchronization, lack of implicit correlation mining between parameters, lack of adaptation to individual pathological features, static threshold judgment, and low early warning accuracy in existing technologies. As a result, the study improves the temporal consistency, comprehensiveness, and accuracy of intracranial pressure monitoring, enabling timely identification and accurate graded early warning of abnormal intracranial pressure states, and reducing the risk of missed and false early warnings. Attached Figure Description

[0055] Figure 1 This is a schematic flowchart of a dynamic multi-parameter fusion monitoring method for intracranial pressure in patients provided by an embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram of a dynamic multi-parameter fusion monitoring system for intracranial pressure provided in an embodiment of the present invention.

[0057] Figure 3 This is a schematic diagram of a computing device. Detailed Implementation

[0058] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0059] like Figure 1 As shown, embodiments of the present invention propose a method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients, the method comprising the following steps:

[0060] Step 1: Obtain raw multi-parameter monitoring data of the target patient during the postoperative monitoring period;

[0061] Step 2: Based on the original monitoring data of multiple parameters, perform time-series alignment processing to unify the timestamps of each parameter data and obtain time-synchronized multi-parameter time-series data;

[0062] Step 3: Based on time-synchronized multi-parameter time-series data, construct a multi-dimensional feature space to characterize the correlation between various monitoring parameters, and generate a continuous parameter mapping surface within the multi-dimensional feature space; by adaptively dividing the parameter mapping surface, obtain multiple feature sub-regions with differentiated data distribution characteristics;

[0063] Step 4: After mapping the real-time acquired time-series data points to the corresponding feature sub-regions, dynamic weighted fusion calculation is performed by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status.

[0064] Step 5: Based on the time-series fusion monitoring values, first construct the preset threshold interval of the corresponding postoperative time window as a multi-dimensional geometric feature region, convert the time-series fusion monitoring values ​​at each time point into multi-dimensional geometric feature points, and determine whether the geometric feature points are within the preset multi-dimensional geometric feature region. Complete the comparison between the fusion monitoring values ​​and the preset threshold interval of the corresponding time window to obtain the comparison results.

[0065] Step 6: Based on the comparison results, when the fused monitoring value exceeds the preset threshold range, generate and output graded early warning information corresponding to the degree of exceedance.

[0066] In this embodiment of the invention, by employing technical means such as acquiring multi-parameter raw monitoring data, performing time-series alignment processing, constructing a multi-dimensional feature space and adaptively dividing feature sub-regions, combining the statistical characteristics of feature sub-regions with individual pathological characteristics to perform dynamic weighted fusion calculation, constructing multi-dimensional geometric feature regions for comparison by time windows, and generating graded warnings based on the degree of exceedance, the technical problems of asynchronous multi-parameter time sequences, difficulty in mining the correlation between parameters, lack of individual adaptability, static threshold judgment, and insufficient warning accuracy in existing intracranial pressure monitoring technologies are overcome. This achieves comprehensive and accurate monitoring of the overall intracranial pressure status, timely capture of abnormal states and provision of targeted graded warnings, improved temporal consistency, correlation, and clinical adaptability of monitoring results, reduced risk of missed warnings, and improved patient prognosis.

[0067] In a preferred embodiment of the present invention, step 1 above may include:

[0068] Step 1.1: Receive raw time-series data streams of at least two types of physiological parameters continuously uploaded by the bedside monitoring device from the target patient; perform integrity verification and filter out abnormal instantaneous values ​​for each received raw time-series data stream to obtain multi-parameter raw monitoring data. Specifically, this includes: First, establishing a stable data communication link with the bedside multi-parameter monitoring device. The communication link is based on a medical-specific data transmission protocol to achieve real-time data transmission. Continuously receive raw time-series data streams of physiological parameters uploaded by the bedside monitoring device from the target patient. The received data streams contain at least two different types of physiological parameters. Common parameter types include intracranial pressure-related data streams, cerebral oxygen saturation-related data streams, mean arterial pressure-related data streams, cerebral perfusion pressure-related data streams, heart rate-related data streams, and central venous pressure-related data streams. All uploaded raw time-series data streams carry the patient's individual identification and initial acquisition time identifier to ensure accurate association between the data and the target patient.

[0069] Secondly, integrity checks are performed on the received raw time-series data streams of various types of physiological parameters. During the check, each data stream is divided into continuous data regions according to preset time segmentation rules. Each region is checked for issues such as missing data points, interrupted data transmission, or incorrect data identification. If a small number of missing data points are identified in a data region, reasonable supplementation is performed based on the continuous data change trend of the parameters before and after the missing points. If a large area of ​​data loss or incorrect identification is identified in a data region, the invalid data unit is marked and removed. At the same time, a data retransmission command is sent to the bedside monitoring device to obtain valid data until all types of raw time-series data streams form a continuous and complete data sequence without critical missing data.

[0070] Finally, an abnormal instantaneous value filtering operation is performed on each raw time-series data stream that has completed integrity verification. The clinical routine physiological fluctuation range and reasonable rate of change threshold of various physiological parameters are pre-loaded. For each data point in the data stream, the value is first compared to see if it is within the routine physiological fluctuation range of the corresponding parameter. Then, it is analyzed whether the change amplitude of the value of the data point and the adjacent data points exceeds the reasonable rate of change threshold. For instantaneous abnormal data with values ​​far exceeding the routine physiological range or sudden rises and falls without physiological logic in a short period of time, they are identified as abnormal instantaneous values ​​and removed. For the data gaps after removing abnormal values, they are smoothly filled in by combining the historical stable change trend of the parameter. Finally, multi-parameter raw monitoring data with complete physiological parameters and no abnormal interference are obtained.

[0071] In this embodiment of the invention, by employing the technical means of receiving at least two types of raw time-series data streams of physiological parameters continuously uploaded by bedside monitoring equipment and performing integrity verification and filtering out abnormal instantaneous values ​​for each data stream, the problems of possible integrity loss and instantaneous abnormal interference in the raw physiological parameter data are overcome, thereby ensuring the reliability and effectiveness of multi-parameter raw monitoring data.

[0072] In a preferred embodiment of the present invention, step 2 above may include:

[0073] Step 2.1 involves analyzing the multi-parameter raw monitoring data to identify the native sampling frequency of the raw time-series data streams corresponding to different physiological parameters. Specifically, after acquiring the processed multi-parameter raw monitoring data, a time-series frequency analysis process is initiated to identify the native sampling frequency of each parameter. First, the multi-parameter raw monitoring data is categorized according to the physiological parameter type, forming multiple data subsets such as the intracranial pressure raw time-series data stream, cerebral oxygen saturation raw time-series data stream, mean arterial pressure raw time-series data stream, and cerebral perfusion pressure raw time-series data stream. Subsequently, time-dimensional sequence analysis is performed on each data subset to extract each... The time interval between two adjacent data acquisition points in the data stream is used to determine the acquisition period of the original time-series data stream for that parameter by averaging multiple consecutive time intervals. Then, the native sampling frequency of the corresponding physiological parameter is calculated based on the reciprocal of the acquisition period. For example, for the intracranial pressure data stream, if the average time interval between adjacent data points is five seconds, then its native sampling frequency is once every five seconds; for the cerebral oxygen saturation data stream, if the average time interval is ten seconds, then its native sampling frequency is once every ten seconds; for the mean arterial pressure data stream, if the average time interval is one second, then its native sampling frequency is once every second.

[0074] Step 2.2: Based on the original sampling frequency and a unified target frequency dynamically set according to clinical monitoring needs, the original time-series data streams of each parameter are resampled using an adaptive interpolation algorithm to align the sampling times of all parameters to the same time series. Specifically, this includes: determining a unified target frequency based on the identified original sampling frequencies of each parameter and clinical monitoring needs. These clinical monitoring needs are dynamically set according to the patient's postoperative recovery stage. For example, in the early postoperative period when intracranial hemorrhage is prevalent, the unified target frequency is set to once every five seconds to ensure monitoring accuracy; during the postoperative recovery period, to balance monitoring accuracy and medical monitoring efficiency, the unified target frequency is set to once every ten seconds. After determining the unified target frequency, adaptive interpolation resampling is performed on the original time-series data streams of each parameter. For data streams with original sampling frequencies higher than the unified target frequency (e.g., the original sampling frequency for mean arterial pressure is once per second while the target frequency is once every five seconds), a sliding window is used. The mean interpolation algorithm calculates the mean of five consecutive one-second interval mean arterial pressure data points to generate a fused data point with a five-second interval, achieving downsampling processing and ensuring that the data sampling time is aligned to a unified time series. For data streams with a native sampling frequency lower than the unified target frequency, such as brain oxygen saturation with a native sampling frequency of once every ten seconds and a target frequency of once every five seconds, the system adopts a physiological trend adaptive interpolation algorithm. It first analyzes the historical trend of brain oxygen saturation data, and then supplements interpolated data points that conform to the physiological change pattern at the five-second time node between two adjacent original data points, achieving upsampling processing and ensuring that the sampling time of this parameter data is also aligned to a unified time series. For data streams with a native sampling frequency consistent with the unified target frequency, such as intracranial pressure with both native and target frequencies of once every five seconds, the timestamp of the original data is directly retained without interpolation processing, ultimately achieving complete alignment of the sampling times of the original time series data streams of all physiological parameters.

[0075] Step 2.3: Based on the unified time series, construct a time-synchronized multi-parameter time series data containing synchronized data columns for each parameter. Specifically, after aligning the sampling times of all parameter data streams, begin constructing the time-synchronized multi-parameter time series data. First, generate a continuous and equally spaced timestamp sequence based on the unified time series. The time interval of the timestamp sequence is consistent with the acquisition period corresponding to the unified target frequency, and each timestamp corresponds to a fixed monitoring time node. Then, integrate the data streams of each physiological parameter after resampling according to the timestamp sequence, matching all physiological parameter monitoring values ​​at each time node. For example, at a time node with a five-second interval, simultaneously integrate all parameter data such as intracranial pressure, cerebral oxygen saturation, mean arterial pressure, and cerebral perfusion pressure at that time. Finally, using the timestamp sequence as the row index and each physiological parameter type as the column index, construct a two-dimensional data matrix of the time-synchronized multi-parameter time series data.

[0076] In this embodiment of the invention, by employing the technical means of identifying the original sampling frequency corresponding to different physiological parameters, performing adaptive interpolation resampling processing on the original time-series data stream of each parameter based on a unified target frequency dynamically set according to clinical monitoring needs, and constructing time-synchronized multi-parameter time-series data containing synchronized data columns of each parameter based on a unified time series, the technical problems of time sequence misalignment and data inability to be synchronized and correlated caused by the difference in the original sampling frequency of different physiological parameters are overcome, thereby achieving the unification of timestamps of each monitoring parameter data and ensuring the time consistency of multi-parameter data.

[0077] In a preferred embodiment of the present invention, step 3 above may include:

[0078] Step 3.1: Receive time-synchronized multi-parameter time-series data. Based on the physiological significance and clinical relevance of each monitoring parameter, construct a multi-dimensional feature space, where each dimension corresponds to a monitoring parameter, forming a spatial distribution set of monitoring parameters within the multi-dimensional feature space. Specifically, this includes: first, receiving the processed time-synchronized multi-parameter time-series data, and then initiating the construction process of the multi-dimensional feature space; in the early stages of construction, the physiological significance and clinical relevance of each monitoring parameter will be thoroughly analyzed and calibrated to clarify the clinical value of each parameter and the logical relationships between parameters: intracranial pressure directly represents the pressure state of the intracranial cavities, and its numerical changes are directly related to complications such as intracranial hemorrhage and cerebral edema; cerebral perfusion pressure reflects the blood perfusion level of brain tissue and has a direct physiological correlation with intracranial pressure and mean arterial pressure; cerebral oxygen saturation reflects the oxygen metabolism balance of brain tissue and can help determine cerebral perfusion. The system monitors whether oxygen supply requirements are met; mean arterial pressure (MAP), as a core indicator of systemic circulation, directly affects the stability of cerebral perfusion pressure (CPP); heart rate indirectly reflects the patient's overall circulatory stress state and shows a synergistic trend with abnormal increases in intracranial pressure (ICP). Based on physiological significance and clinical relevance, each monitoring parameter is assigned an independent dimension to construct a multidimensional feature space. Intracranial pressure corresponds to the first dimension, CPP to the second, CPP to the third, mean arterial pressure to the fourth, and heart rate to the fifth. The numerical ranges of each dimension are standardized based on routine clinical physiological intervals and postoperative abnormal warning intervals. After dimension setting, each time node data in the time-synchronized multi-parameter time-series data is transformed into a spatial distribution point in the multidimensional feature space. The coordinates of this spatial point are determined by the values ​​of the monitoring parameters at the corresponding time node. For example, the IPP value at a certain moment is the first dimension coordinate, the CPP value is the second dimension coordinate, the CPP value is the third dimension coordinate, and so on. As time nodes accumulate, the spatial points corresponding to all time nodes will form a complete set of spatial distribution points covering the postoperative monitoring period within the multidimensional feature space.

[0079] Step 3.2: Based on the spatial distribution point set formed in the multidimensional feature space, a nonlinear mapping algorithm is used to perform surface fitting on the spatial distribution point set to obtain a continuous parametric mapping surface. Specifically, this includes: after obtaining the spatial distribution point set in the multidimensional feature space, starting the parametric mapping surface fitting process, and using a nonlinear mapping algorithm to perform continuous surface fitting on the discrete spatial distribution point set; in the early stage of fitting, the system will first preprocess the spatial distribution point set to remove isolated outliers caused by small fluctuations in the sensor, ensuring that the overall distribution of the point set conforms to the physiological change logic; subsequently, based on the nonlinear physiological correlation between each parameter, the fitting constraints of the mapping algorithm are set. The constraints will take into account the physiological dependence between parameters, such as the decrease in cerebral perfusion pressure as intracranial pressure increases. The system employs inverse correlations and positive correlations, such as the increase in cerebral oxygen saturation with increasing cerebral perfusion pressure, to avoid deviations in fitting results that contradict physiological logic. During the fitting process, a continuous parameter mapping surface is generated based on a spatially distributed point set through local fitting and global stitching of each region. The surface completely covers the distribution range of the spatially distributed point set, and its height and slope are dynamically adjusted according to changes in the combined state of multiple parameters: for example, in regions with low intracranial pressure and high cerebral perfusion pressure, the surface exhibits a gentle, low-slope state; in regions with high intracranial pressure and low cerebral oxygen saturation, the surface exhibits a steep, high-slope state. The final parameter mapping surface can accurately characterize the nonlinear correlation between various monitored parameters, transforming discrete multi-parameter data into a holistic spatial surface with continuous correlation characteristics.

[0080] Step 3.3: Based on the parametrically mapped surface, analyze its geometric features and data distribution characteristics. According to the surface curvature gradient, data density distribution gradient, and clinical risk distribution characteristics, perform adaptive region segmentation on the parametrically mapped surface to obtain multiple feature sub-regions with differentiated data distribution characteristics. Specifically, after obtaining a continuous parametrically mapped surface, perform adaptive region segmentation. The core basis for segmentation is the surface curvature gradient, data density distribution gradient, and clinical risk distribution characteristics. First, comprehensively analyze the geometric features of the parametrically mapped surface, calculate the curvature gradient of each region. Regions with larger curvature gradients represent abrupt changes in multi-parameter combination states, often corresponding to critical states of clinical risk. Simultaneously, statistically analyze the spatial distribution point density of each region to form a data density distribution gradient. Regions with high data density represent stable multi-parameter states that occur more frequently in postoperative monitoring, while regions with low data density represent... The system identifies rare abnormal parameter combinations. Secondly, it supplements the gradient data with clinical risk distribution characteristics, which are set based on the high incidence of complications in different postoperative time windows: In the early postoperative monitoring period, the focus is on identifying parameter combinations related to intracranial hemorrhage, i.e., the surface region where intracranial pressure suddenly rises accompanied by a sudden drop in cerebral perfusion pressure; in the mid-postoperative monitoring period, the focus is on identifying parameter combinations related to cerebral edema, i.e., the surface region where intracranial pressure continues to rise accompanied by a slow decline in cerebral oxygen saturation; in the postoperative recovery period, the focus is on identifying parameter combinations related to circulatory stability, i.e., the surface region where all parameters are within the physiologically normal range. Finally, based on the three types of gradients and features, adaptive region division is performed on the parameter mapping surface. During the division process, the abrupt change line of the curvature gradient and the boundary point of data density are used as the basic boundaries, combined with the calibration range of the clinical risk region, to divide the entire parameter mapping surface into multiple feature sub-regions with differentiated data distribution characteristics. Common sub-regions include hemorrhage risk feature sub-regions, edema risk feature sub-regions, circulatory stability feature sub-regions, and critical transition feature sub-regions.

[0081] In this embodiment of the invention, by employing techniques such as constructing a multidimensional feature space based on the physiological significance and clinical relevance of each monitoring parameter, obtaining a continuous parameter mapping surface by fitting the spatial distribution point set with a nonlinear mapping algorithm, and adaptively dividing the parameter mapping surface into regions by combining the surface curvature change gradient, data density distribution gradient, and clinical risk distribution characteristics, the technical problems of existing technologies, such as difficulty in mining implicit correlations between multiple parameters and inability to distinguish parameter distribution characteristics corresponding to different clinical risks, are overcome. This achieves the goal of characterizing the correlation between each monitoring parameter, realizing accurate stratification of different clinical risk characteristics, and improving the targeted technical effect of comprehensive intracranial pressure status assessment.

[0082] In a preferred embodiment of the present invention, step 4 above may include:

[0083] Step 4.1: Receive the current time data point from the time-synchronized multi-parameter time-series data in real time. Based on multiple feature sub-regions with differentiated data distribution characteristics, map the current time data point to the corresponding feature sub-region to obtain the data point mapping result. Specifically, this includes: maintaining real-time data communication synchronized with the time series, continuously receiving time-synchronized multi-parameter time-series data, accurately capturing the complete data point at the current time. The data point includes synchronized values ​​of all monitored parameters such as intracranial pressure, cerebral perfusion pressure, cerebral oxygen saturation, mean arterial pressure, and heart rate, ensuring data consistency in the time dimension. Subsequently, invoke the boundary determination rules of the pre-divided feature sub-regions. Each feature sub-region has preset values ​​for each monitored parameter dimension. The numerical range of degrees and the associated logical thresholds are used to compare and match the parameter values ​​of each dimension of the data point at the current moment with the boundary conditions of all feature sub-regions one by one: if the intracranial pressure value of the data point increases and the cerebral perfusion pressure value decreases in the same step, which is consistent with the parameter combination pattern of the hemorrhage risk feature sub-region, then the data point is determined to be mapped to the hemorrhage risk feature sub-region; if the intracranial pressure value of the data point continues to rise and the cerebral oxygen saturation value slowly decreases, which is consistent with the parameter distribution characteristics of the edema risk feature sub-region, then it is mapped to the edema risk feature sub-region; if all parameter values ​​are within the clinical normal range and the fluctuations are gentle, then it is mapped to the circulatory stability feature sub-region; after the comparison is completed, a clear data point mapping result is generated.

[0084] Step 4.2: Based on the data point mapping results, extract the calculated statistical characteristics of the feature sub-regions, including the data distribution center, dispersion, and boundary morphology features. Simultaneously, combine the individual pathological characteristics of the target patient recorded preoperatively with the parameter weight configuration preset by clinical experts to construct a basic parameter set for dynamic weighted fusion calculation. Specifically, based on the obtained data point mapping results, firstly, extract the preset statistical characteristics of the corresponding feature sub-regions. The data distribution center reflects the core numerical level of normal parameter combinations within the sub-region; the dispersion reflects the allowable range of parameter fluctuations within the region; and the boundary morphology features clarify the critical division criteria between this region and other sub-regions. All statistical characteristics are derived from extensive clinical data training and calculation, possessing clear clinical reference value. Simultaneously, the system retrieves the complete individual pathological characteristic information recorded preoperatively by the target patient, including… This includes factors such as whether the patient has pre-existing conditions like hypertension or diabetes, the specific location and size of the intracranial lesion, the surgical procedure and intraoperative condition, and the patient's age and overall physical condition. These factors directly influence the patient's tolerance to changes in intracranial pressure and the high risk of postoperative complications, effectively compensating for the lack of individual pathological feature adaptation in existing technologies. Furthermore, it incorporates parameter weighting configurations preset by clinical experts based on the risk characteristics of different postoperative time windows. For example, in the early postoperative bleeding phase, intracranial pressure and cerebral perfusion pressure have higher preset weights than other parameters; in the mid-postoperative edema phase, cerebral oxygen saturation has an appropriately increased weight. By integrating the extracted sub-regional statistical characteristics, individual patient pathological characteristics, and preset parameter weighting configurations, redundant information is removed, data formats are standardized, and a complete and comprehensive dynamic weighted fusion calculation parameter set is constructed.

[0085] Step 4.3: Based on the basic parameter set, dynamically adjust the fusion weight coefficients of each monitoring parameter according to the relative position of the data point in the feature sub-region, the matching degree of individual pathological features, and the priority of clinical parameter weights. Perform weighted fusion calculation to obtain the fusion value of intracranial pressure status at the current moment. Specifically, this includes: based on the constructed basic parameter set, initiating the dynamic weight adjustment and weighted fusion calculation process to completely solve the problem of fixed fusion weights in existing technologies; firstly, adjusting the weights according to the relative position of the data point in the feature sub-region: if the data point is close to the distribution center of the feature sub-region, it indicates that the current parameter combination is in a stable state in that region, and the weights of each parameter are allocated according to the preset benchmark ratio; if the data point is close to the high-risk boundary of the feature sub-region, such as the intracranial pressure critical value boundary of the hemorrhage risk sub-region, then increase the weight coefficients of intracranial pressure and cerebral perfusion pressure, reduce the weights of secondary parameters, and highlight the key parameters. The impact of risk parameters is considered; secondly, the weights are adjusted based on the matching degree of individual pathological characteristics: if the patient has a history of hypertension before surgery, the fluctuation of their mean arterial pressure has a more significant impact on intracranial pressure, thus increasing the fusion weight of mean arterial pressure; if the patient has insufficient cerebral perfusion before surgery, the weights of cerebral perfusion pressure and cerebral oxygen saturation are increased to make the fusion calculation more in line with the individual physiological characteristics of the patient; finally, the final adjustment is made in combination with the priority of clinical parameter weights. In the priority rules preset by clinical experts, core parameters that are directly related to changes in intracranial pressure have a higher priority than indirectly related parameters. The weight coefficients are fine-tuned and optimized according to this priority order to ensure that key parameters play a leading role in the fusion calculation; after the weight adjustment is completed, the values ​​of each monitored parameter are multiplied by the corresponding weight coefficients, and then all the product results are summed to obtain the fusion value of intracranial pressure status at the current moment that can comprehensively reflect the synergistic influence of multiple parameters.

[0086] Step 4.4: Based on the current intracranial pressure fusion value and combined with the historical fusion value sequence, perform temporal smoothing and trend analysis to finally output the temporal fusion monitoring value characterizing the overall intracranial pressure status. Specifically, this includes: after obtaining the current intracranial pressure fusion value, immediately retrieving the intracranial pressure fusion values ​​from all historical moments during the patient's postoperative monitoring period to form a complete historical fusion value sequence, avoiding the biased monitoring results caused by relying solely on data from a single moment; firstly, perform temporal smoothing on the current and historical fusion value sequences, using a smoothing method based on the physiological parameter change patterns to remove abrupt abnormal fusion values ​​caused by instantaneous sensor fluctuations or minor interference in data transmission. For data gaps after removing abnormal values, based on... The system is supplemented by analyzing the changing trends of adjacent fusion values ​​to make the changes in the fusion value sequence more closely match the actual physiological fluctuations of intracranial pressure, thus improving data stability. Subsequently, trend analysis is performed, with the system calculating the amplitude and rate of change of recent fusion value sequences to determine the current trend of intracranial pressure fusion values: a sustained, slow upward trend in fusion values ​​suggests early cerebral edema; a sudden increase in fusion values ​​followed by a sustained high level suggests potential intracranial hemorrhage; and stable fusion values ​​within the normal range with minimal fluctuations indicate a stable intracranial pressure state. Time-series smoothing is used to eliminate data noise, and trend analysis captures potential risk changes, ultimately outputting time-series fusion monitoring values ​​that comprehensively, accurately, and stably characterize the overall intracranial pressure status.

[0087] In this embodiment of the invention, by employing technical means such as mapping real-time data points to corresponding feature sub-regions, extracting statistical characteristics of sub-regions and constructing a basic parameter set by combining individual pathological characteristics and clinically preset weights, dynamically adjusting the fusion weight coefficients and performing weighted calculations based on the spatial location of data points and the matching degree of individual pathology, and outputting intracranial pressure time-series fusion monitoring values ​​after completing temporal smoothing and trend analysis by combining historical fusion value sequences, the invention overcomes the technical problems of fixed weights, failure to adapt to individual pathological differences, failure to combine clinical characteristics of parameter distribution, and lack of temporal trend analysis in existing intracranial pressure multi-parameter fusion schemes, which lead to low accuracy, insufficient individual adaptability, and poor stability of fusion monitoring values. Thus, it achieves dynamic intelligent adaptation of fusion weights, improving the individual relevance, data stability, and clinical relevance of intracranial pressure comprehensive status monitoring values.

[0088] In a preferred embodiment of the present invention, step 5 above may include:

[0089] Step 5.1: Receive time-series fusion monitoring values, identify the current postoperative time window according to the postoperative time window division rules, and retrieve the corresponding preset threshold interval parameters based on the time window. Specifically, this includes: receiving time-series fusion monitoring values ​​in real time; the monitoring values ​​have integrated multi-parameter correlation features and individual patient differences, and can comprehensively characterize the overall intracranial pressure status; subsequently, initiate the postoperative time window identification process, and make a judgment based on the preset postoperative time window division rules, which are formulated based on the clinical postoperative complication high incidence patterns: 0-24 hours postoperatively is the first time window, during which the risk of intracranial hemorrhage is the highest, and the monitoring accuracy requirements are the most stringent; 24-48 hours postoperatively is the second time window, where cerebral edema is the main risk, and the slow change trend of parameters needs to be closely monitored; 48-72 hours postoperatively is the third time window, where the patient gradually recovers and stabilizes, the risk is relatively reduced, and the threshold interval can be appropriately relaxed; by calculating the postoperative monitoring time of the target patient, the specific time window to which the current moment belongs is determined, for example, 18 hours postoperatively is determined to be the first time window, and 30 hours postoperatively is determined to be the second time window; after identification, retrieve the preset threshold interval parameters corresponding to the time window from the clinical parameter database.

[0090] Step 5.2: Based on preset threshold interval parameters, construct a multidimensional geometric feature region. Each dimension of the multidimensional geometric feature region corresponds to a clinical risk assessment dimension. Specifically, after obtaining the preset threshold interval parameters for the current time window, initiate the construction process of the multidimensional geometric feature region. First, clarify the selection criteria for the clinical risk assessment dimensions. Combining the core clinical needs of intracranial pressure monitoring, determine the dimension types, including intracranial pressure risk dimension, cerebral perfusion pressure risk dimension, cerebral oxygen metabolism risk dimension, and circulatory stability risk dimension. Each dimension corresponds to a warning indicator for key postoperative complications. Subsequently, based on each clinical risk assessment dimension, ... Preset threshold range parameters are assigned to corresponding dimensions: intracranial pressure risk dimension corresponds to the intracranial pressure-related threshold range, cerebral perfusion pressure risk dimension corresponds to the cerebral perfusion pressure-related threshold range, cerebral oxygen metabolism risk dimension corresponds to the cerebral oxygen saturation-related threshold range, and circulatory stability risk dimension corresponds to the combined threshold range of mean arterial pressure and heart rate. The numerical boundary of each dimension is determined by the upper and lower limits of the threshold of that dimension. By integrating the boundary ranges of all dimensions, a closed multidimensional geometric feature region is constructed. The internal space of the region represents the state in which each risk dimension is within a safe range, the boundary of the region is the critical risk line of each dimension, and the outside of the region is the abnormal risk state.

[0091] Step 5.3 involves decomposing and mapping the time-series fusion monitoring values ​​according to clinical risk assessment dimensions to generate corresponding multidimensional geometric feature points. The coordinates of these multidimensional geometric feature points in each dimension are determined by the components of the time-series fusion monitoring values ​​in the corresponding risk dimensions. Specifically, this includes: after receiving the time-series fusion monitoring values, decomposing and mapping the monitoring values ​​based on the constructed clinical risk assessment dimension system. The decomposition process strictly follows the calculation logic of the time-series fusion monitoring values, extracting the components directly related to each clinical risk assessment dimension. For the intracranial pressure risk dimension, the numerical components in the fusion value primarily contributed by intracranial pressure parameters are extracted; for the cerebral perfusion pressure risk dimension, the contribution components corresponding to cerebral perfusion pressure parameters in the fusion value are extracted; and for the cerebral oxygen metabolism risk dimension... The system extracts components related to brain oxygen saturation parameters. For the circulatory stability risk dimension, it extracts components corresponding to the synergistic effect of mean arterial pressure and heart rate, ensuring that each component accurately reflects the actual state of the corresponding risk dimension. Subsequently, the components corresponding to each risk dimension are transformed into coordinate values ​​of multidimensional geometric feature points. The intracranial pressure risk dimension component serves as the first dimension of the feature point, the brain perfusion pressure risk dimension component serves as the second dimension, the brain oxygen metabolism risk dimension component serves as the third dimension, and the circulatory stability risk dimension component serves as the fourth dimension. The numerical range of each coordinate value is consistent with the boundary range of the corresponding dimension of the multidimensional geometric feature region, ultimately generating multidimensional geometric feature points that can comprehensively reflect the synergistic state of each risk dimension.

[0092] Step 5.4: Based on the multidimensional geometric feature region and multidimensional geometric feature points, a point-within-a-polygon detection algorithm is used to determine the spatial positional relationship of the multidimensional geometric feature points relative to the multidimensional geometric feature region, so as to obtain the judgment result, and generate a comparison result based on the judgment result. Specifically, this includes: First, acquiring the boundary coordinate data of the multidimensional geometric feature region and the coordinate data of the multidimensional geometric feature points. To ensure the accuracy of the position judgment, the boundary coordinate data of the multidimensional geometric feature region is first geometrically standardized, converting the coordinate values ​​of all boundary vertices into a unified relative coordinate system, eliminating the calculation error caused by the difference in the numerical range of different dimensions, generating a standardized polygon boundary vertex sequence, and clarifying the precise position of each dimension boundary; Next, the point-within-a-polygon detection algorithm is used to perform position judgment. Based on the standardized polygon boundary vertex sequence and the coordinate data of the multidimensional geometric feature points, a measurement line is sent from the multidimensional geometric feature points to a preset direction. The test ray is used to calculate the number of intersections between each ray and each edge of the polygon boundary, recording the position and direction of each intersection in detail to ensure the accuracy of the intersection count. Then, the spatial position of the multidimensional geometric feature point is determined based on the parity of the intersection count: if the number of intersections is odd, the feature point is determined to be inside the multidimensional geometric feature region; if the number of intersections is even, the feature point is determined to be outside the region. Simultaneously, the straight-line distances from the multidimensional geometric feature point to each edge of the polygon are calculated, and the minimum value is selected as the minimum distance from the feature point to the region boundary. The position determination result and the minimum distance are combined to form a complete spatial position determination result. Finally, a comparison result is generated based on the spatial position determination result: if the feature point is inside the region and the minimum distance is greater than the preset safety distance threshold, it is determined to be in a normal state; if the feature point is inside the region but the minimum distance is less than or equal to the safety distance threshold, it is determined to be in a critical state; if the feature point is outside the region, it is determined to be in an abnormal state.

[0093] In this embodiment of the invention, by employing technical means such as identifying the current time window according to the postoperative time window division rules and retrieving the corresponding threshold interval parameters, constructing a multidimensional geometric feature region corresponding to the clinical risk assessment dimension based on the threshold parameters, decomposing and mapping the time-series fusion monitoring values ​​into multidimensional geometric feature points, and using a point-in-polygon detection algorithm to determine the spatial positional relationship between the feature points and the geometric region and generate comparison results, the technical problems of static and uniform thresholds, failure to distinguish the risk differences of different postoperative time windows, and low accuracy and insufficient clinical adaptability of abnormal state determination caused by using only single-dimensional threshold comparison are overcome in the existing intracranial pressure monitoring threshold determination schemes. Thus, it achieves accurate comparison of intracranial pressure status by time window and multiple dimensions, and improves the time adaptability, dimensional correlation and clinical accuracy of abnormal state determination.

[0094] In a preferred embodiment of the present invention, step 5.4 above may include:

[0095] Step 5.41: Receive the boundary coordinate data of the multidimensional geometric feature region and the coordinate data of the multidimensional geometric feature points. Perform geometric standardization on the boundary coordinate data of the multidimensional geometric feature region to obtain a standardized sequence of polygon boundary vertices. Specifically, this includes: first, receiving the complete boundary coordinate data of the multidimensional geometric feature region and the coordinate data of the multidimensional geometric feature points. The boundary coordinate data includes the boundary vertex coordinates of each clinical risk assessment dimension, and the feature point coordinate data corresponds to the specific values ​​of the current intracranial pressure status in each risk dimension. Since the parameter value ranges differ between different clinical risk assessment dimensions, directly determining spatial location will be affected by the dimensional scale. Inconsistencies lead to errors and are one of the reasons for the insufficient accuracy of threshold judgment in existing technologies. To solve this problem, geometric standardization is performed on the boundary coordinate data of multi-dimensional geometric feature regions. During the process, the maximum and minimum values ​​of each risk assessment dimension are first extracted to determine the range of values ​​for each dimension. Then, the boundary coordinate values ​​of all dimensions are mapped to the same standardized value range according to a uniform ratio to eliminate scale differences between different dimensions. At the same time, the arrangement order of boundary vertices is standardized to ensure that the vertices are arranged in an orderly manner in a clockwise or counterclockwise direction to form a continuous closed boundary contour. After processing, a standardized polygon boundary vertex sequence is obtained.

[0096] Step 5.42: Based on the standardized polygon boundary vertex sequence and the coordinate data of multi-dimensional geometric feature points, a ray intersection counting algorithm is used to calculate the number of intersections between the test ray emanating from the multi-dimensional geometric feature points and the polygon boundary edges, obtaining the intersection count result. Specifically, this includes: after obtaining the standardized polygon boundary vertex sequence and the coordinate data of multi-dimensional geometric feature points, starting the execution flow of the ray intersection counting algorithm; first, starting from the multi-dimensional geometric feature points, a test ray is emitted in a preset fixed direction. The direction is chosen to be non-parallel to each risk assessment dimension to avoid counting deviations caused by the ray coinciding with the polygon boundary edges; then, adjacent vertices in the standardized polygon boundary vertex sequence are extracted one by one to form each boundary edge of the polygon, and the intersection of the test ray with each boundary edge is judged sequentially. During the judgment process, the system analyzes the spatial positional relationship between the ray and the boundary edge. If the ray passes through the boundary edge and the intersection point is not located at the endpoint of the boundary edge, a valid intersection is recorded; if the ray is parallel to the boundary edge or only touches the endpoint, it is not counted as a valid intersection; after completing the intersection judgment of all boundary edges and test rays in sequence, the total number of valid intersections is summarized to obtain the intersection count result.

[0097] Step 5.43: Based on the crossover count calculation results, determine whether the multidimensional geometric feature point is located within the multidimensional geometric feature region according to the parity of the crossover count. Simultaneously, calculate the minimum Euclidean distance from the multidimensional geometric feature point to the polygon boundary to obtain the spatial location judgment result. Specifically, this includes: Based on the obtained crossover count calculation results, firstly, perform a region-inside / outside judgment of the multidimensional geometric feature point. According to the core logic of the ray crossover point counting algorithm, if the crossover count is odd, it means that the test ray crosses the polygon boundary an odd number of times after starting from the feature point, determining that the multidimensional geometric feature point is located inside the multidimensional geometric feature region; if the crossover count is even, it means that the ray returns to the outside of the region after crossing the boundary an even number of times, determining that the feature point is located outside the multidimensional geometric feature region. At the same time, initiate the minimum Euclidean distance calculation process, calculating the straight-line distance from the multidimensional geometric feature point to each boundary edge of the polygon one by one. During this process, fully utilize standardized coordinate data to ensure the accuracy of the distance calculation. After the calculation is completed, select the minimum value from all the distance values ​​corresponding to the boundary edges as the minimum Euclidean distance from the multidimensional geometric feature point to the polygon boundary. Integrate the region-inside / outside judgment result with the minimum Euclidean distance value to form a complete spatial location judgment result.

[0098] Step 5.44: Based on the spatial location judgment result, and by comparing the minimum Euclidean distance with the preset critical distance threshold, the spatial location judgment result is mapped into a clinically interpretable state identifier, resulting in a comparison result that includes normal state, critical state, and abnormal state identifiers. Specifically, this includes: first, retrieving the preset critical distance threshold, which is formulated by clinical experts based on the risk characteristics and clinical intervention experience of different postoperative time windows, accurately reflecting the critical standards for risk warning at each stage; then, combining the obtained spatial location judgment result with the minimum Euclidean distance to perform state identifier mapping. If the feature point is located within the multidimensional geometric feature region and the minimum Euclidean distance is greater than the preset critical distance threshold, it indicates that the current intracranial pressure status is within the safe range in all risk dimensions and is far from the critical risk line, and is mapped as a normal state identifier. If the feature point is located within the region, but the minimum Euclidean distance is less than or equal to the preset critical distance threshold, it indicates that the current state is close to the risk threshold and may develop into an abnormal state if not monitored in time, and is mapped as a critical state identifier. If the feature point is located outside the multidimensional geometric feature region, it indicates that the current state has exceeded the safe range and there is a clear clinical risk, and is mapped as an abnormal state identifier. Finally, a comparison result containing normal state, critical state, or abnormal state identifiers is generated.

[0099] In this embodiment of the invention, by employing geometric standardization of the boundary coordinate data of multidimensional geometric feature regions, using a ray intersection counting algorithm to determine the spatial positional relationship between multidimensional geometric feature points and polygonal regions, calculating the minimum Euclidean distance from feature points to the region boundary, and mapping the spatial positional results to clinically interpretable status identifiers that include normal, critical, and abnormal conditions by comparing the minimum Euclidean distance with a preset critical distance threshold, the technical problems of existing intracranial pressure status comparison schemes—such as large positional judgment errors due to non-standardized boundary data, the inability to identify critical states only by determining whether thresholds are exceeded, and the lack of clinical interpretability of spatial positional judgment results—are overcome. This achieves improved accuracy in multidimensional spatial positional judgment, enables early identification of critical intracranial pressure states, and gives the comparison results clear clinical significance.

[0100] In a preferred embodiment of the present invention, step 6 above may include:

[0101] Step 6.1: Receive the comparison results, analyze the clinical status identifiers contained in the comparison results, and identify whether the current intracranial pressure status is abnormal. Specifically, this includes: continuously receiving completed comparison results that clearly contain clinical status identifiers for normal, borderline, or abnormal states, and possessing clear clinical interpretability; performing structured analysis on the received comparison results, extracting core status identifier information, and quickly distinguishing the current intracranial pressure status type through preset identifier recognition rules: if a normal status identifier is parsed, it is determined that no early warning process needs to be initiated, and only the status is recorded for subsequent trend analysis; if a borderline status identifier is parsed, it is marked as a potential risk status requiring key attention, triggering the subsequent risk tracking mechanism; if an abnormal status identifier is parsed, the current intracranial pressure status is directly determined to be abnormal, and subsequent early warning related processes are immediately initiated.

[0102] Step 6.2: Based on the identified abnormal state, retrieve the coordinate data of multidimensional geometric feature points and the boundary data of multidimensional geometric feature regions, and calculate the excess distance and direction of the multidimensional geometric feature points exceeding the boundary of the multidimensional geometric feature regions. Specifically, this includes: immediately initiating the data retrieval process after identifying the abnormal state, accurately retrieving the coordinate data of the multidimensional geometric feature points and the corresponding boundary data of the multidimensional geometric feature regions from the database at the current moment, ensuring that the two types of data are completely matched in terms of time dimension and dimensional system; subsequently, performing the excess distance calculation, using the coordinates of each dimension of the multidimensional geometric feature points and the boundary coordinates of the multidimensional geometric feature regions as the basis, calculating the minimum distance from the feature point to the region boundary. The short straight-line distance, which is the excess distance, directly reflects the degree to which the abnormal state deviates from the safe range. Simultaneously, the direction of excess is analyzed: by comparing the coordinates of each dimension of the feature point with the threshold range of the corresponding dimension of the region boundary, it is determined which specific clinical risk assessment dimension(s) exceed the limit. For example, if the coordinate value of the feature point in the intracranial pressure risk dimension exceeds the upper limit of the region boundary, the direction of excess is the intracranial pressure risk dimension; if the coordinate value in the cerebral oxygen metabolism risk dimension is lower than the lower limit of the region boundary, the direction of excess is the cerebral oxygen metabolism risk dimension. This clarifies the specific degree and dimension of excess in the abnormal state, solving the problem in existing technologies where only the abnormality is known but the degree of abnormality cannot be quantified or the direction of excess located.

[0103] Step 6.3: Based on the distance and direction of the excess, combined with the individual pathological characteristics of the target patient and the clinical risk level of the postoperative time window, assess the clinical severity of intracranial pressure abnormality to obtain a quantitative index of the excess degree. Specifically, this includes: initiating the clinical severity assessment process based on the obtained distance and direction of the excess; firstly, retrieving the complete individual pathological characteristic information of the target patient entered preoperatively, including whether there are underlying diseases such as hypertension and diabetes before surgery, the location and size of the intracranial lesion, the surgical method and intraoperative situation, the patient's age and basic physical condition, etc., which directly affect the patient's tolerance to intracranial pressure abnormality and the probability of complications; at the same time, determining the postoperative time window to which the current moment belongs, and retrieving the corresponding clinical risk level of the time window. The risk level is categorized as follows: 0-24 hours post-surgery is high-risk, 24-48 hours is medium-risk, and 48-72 hours is low-risk. Different risk levels correspond to different assessment weights. During the assessment, the distance exceeding the risk level is multiplied by the risk level weight to obtain a base quantitative value. Then, a correction coefficient is assigned based on the importance of the risk dimension corresponding to the direction of the exceedance; for example, the correction coefficient is higher when the intracranial pressure risk dimension exceeds the standard. Finally, adjustments are made based on the degree of matching with individual pathological characteristics. If the patient has a history of hypertension and the direction of the exceedance is the intracranial pressure dimension, the quantitative value is further increased; otherwise, it is appropriately decreased. Through multi-factor comprehensive calculation, a quantitative index of the degree of exceedance that comprehensively reflects the clinical severity of intracranial pressure abnormalities is obtained.

[0104] Step 6.4: Based on the quantitative indicators of the degree of exceedance, and according to the preset warning grading rule base, the quantitative indicators of the degree of exceedance are mapped to the corresponding warning levels, generating graded warning information containing warning levels, risk descriptions, and suggested treatment measures. Specifically, this includes: retrieving the preset warning grading rule base, which was developed by clinical experts based on extensive clinical case data, evidence-based medicine, and postoperative complication intervention experience. This rule base contains the correspondence between quantitative indicators of the degree of exceedance and warning levels, with warning levels divided into Level 1, Level 2, and Level 3 warnings, corresponding to mild, moderate, and severe abnormalities, respectively; comparing the obtained quantitative indicators of the degree of exceedance with the threshold ranges in the rule base: if the quantitative indicator is in the mild abnormality range, it is mapped to Level 1 warning; if it is in the moderate abnormality range... If the abnormality is within a certain range, it is mapped to a Level 2 warning; if it is within a severely abnormal range, it is mapped to a Level 3 warning. Subsequently, corresponding graded warning information is generated, with the warning level clearly indicating the current risk level. The risk description section details the risk dimension, degree of abnormality, and possible clinical complications corresponding to the direction of the deviation. For example, a Level 1 warning may be described as a slight deviation of the intracranial pressure risk dimension from the safe range, with a possible risk of early cerebral edema. The recommended treatment section provides targeted clinical intervention suggestions based on the warning level and the direction of deviation. For example, a Level 1 warning suggests increasing the monitoring frequency and recording parameter changes once per hour; a Level 2 warning suggests repeating a head computed tomography scan and adjusting the dosage of dehydrating drugs; and a Level 3 warning suggests immediately notifying the doctor to prepare for emergency treatment and surgical intervention.

[0105] Step 6.5: Based on the tiered early warning information, the tiered early warning information is transmitted in real time to the clinical monitoring terminal through the medical data communication interface, and the timestamp, warning level, and patient identification information of the early warning event are recorded. Specifically, this includes: based on the generated tiered early warning information, initiating the medical data transmission process, and transmitting the early warning information in real time to designated clinical monitoring terminals through the hospital's internal medical dedicated data communication interface, including the central monitoring screen at the nurse station, the mobile terminal of the attending physician, and the handheld terminal of the ward nursing staff, ensuring that relevant medical staff can receive the early warning notification as soon as possible, avoiding intervention delays caused by information transmission delays; at the same time, the early warning event recording module is activated to automatically record the key information of the early warning event, with the timestamp accurate to the second, recording the specific time the early warning was triggered; the warning level clearly indicates the current risk level of the early warning; the patient identification information includes the patient's name, age, hospital number, etc., ensuring that the early warning event is accurately associated with the target patient. In addition, the risk description and suggested treatment measures in the early warning information are also recorded to form a complete early warning event file.

[0106] In this embodiment of the invention, by employing the technical means of identifying abnormal states using clinical status identifiers from analytical comparison results, calculating the distance and direction of multidimensional geometric feature points exceeding the region, quantifying the severity of abnormalities by combining individual pathological characteristics with postoperative time window risk levels, generating graded early warning information containing early warning levels and treatment suggestions based on an early warning grading rule base, and pushing early warning information and recording relevant event data through a medical communication interface, the technical problems of existing intracranial pressure early warning schemes—such as lack of graded early warnings, failure to adapt to individual pathological differences and postoperative time window risks, lack of quantification of specific location and degree of abnormalities, lack of traceability of early warning information, and insufficient clinical guidance—are overcome. This achieves precise graded early warning of intracranial pressure abnormalities, improves the individual adaptability and clinical relevance of early warnings, provides clear risk descriptions and treatment guidelines for medical staff, and ensures that early warning events are traceable throughout the entire process, thus helping clinicians to take timely differentiated intervention measures and reduce the risk of postoperative complications.

[0107] like Figure 2 As shown, embodiments of the present invention also provide a dynamic multi-parameter fusion monitoring system for intracranial pressure in patients, comprising:

[0108] The acquisition module is used to acquire raw monitoring data of multiple parameters of the target patient during the postoperative monitoring period;

[0109] The synchronization module is used to perform time-series alignment processing on the original monitoring data of multiple parameters to unify the timestamps of each parameter data and obtain time-synchronized multi-parameter time-series data.

[0110] The mapping module is used to construct a multi-dimensional feature space based on time-synchronized multi-parameter time-series data to characterize the correlation between various monitoring parameters, and to generate a continuous parameter mapping surface within the multi-dimensional feature space; by adaptively dividing the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained.

[0111] The fusion module is used to map real-time acquired time-series data points to corresponding feature sub-regions, and then perform dynamic weighted fusion calculations by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status.

[0112] The comparison module is used to construct a multi-dimensional geometric feature region based on the temporal fusion monitoring value, first construct the preset threshold interval of the corresponding postoperative time window as a multi-dimensional geometric feature region, convert the temporal fusion monitoring value at each time into multi-dimensional geometric feature points, and determine whether the geometric feature points are within the preset multi-dimensional geometric feature region, and complete the comparison between the fusion monitoring value and the preset threshold interval of the corresponding time window to obtain the comparison result.

[0113] The processing module is used to generate and output graded early warning information corresponding to the degree of exceedance when the fused monitoring value exceeds the preset threshold range based on the comparison results.

[0114] The monitoring system according to embodiments of the present invention can correspond to the execution of the methods described in the embodiments of the present invention, and the above and other operations and / or functions of each module of the monitoring system are respectively for implementing Figure 1 The corresponding process of the method in the illustrated embodiment will not be described in detail here for the sake of brevity.

[0115] This application also provides a computing device. This computing device can utilize a server.

[0116] like Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 700 includes a bus 701, a processor 702, a communication interface 703, and a memory 704. The processor 702, the memory 704, and the communication interface 703 communicate with each other via the bus 701.

[0117] The 701 bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0118] The processor 702 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0119] Communication interface 703 is used for external communication. Memory 704 may include volatile memory, such as random access memory (RAM). Memory 704 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD). Executable code is stored in memory 704, and processor 702 executes this executable code to perform the aforementioned dynamic multi-parameter fusion monitoring method for intracranial pressure in patients.

[0120] Specifically, in implementing the patient intracranial pressure dynamic multi-parameter fusion monitoring system described in the above embodiments, and where each module or unit of the patient intracranial pressure dynamic multi-parameter fusion monitoring system described in the above embodiments is implemented by software, the software or program code required to execute the functions of each module / unit in the patient intracranial pressure dynamic multi-parameter fusion monitoring system described in the above embodiments can be partially or entirely stored in the memory 704. The processor 702 executes the program code corresponding to each unit stored in the memory 704 to execute the aforementioned patient intracranial pressure dynamic multi-parameter fusion monitoring method.

[0121] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium capable of being stored by a computing device, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the above-described method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients.

[0122] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0123] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0124] When the computer program product is executed by a computer, the computer performs any of the aforementioned methods of the dynamic multi-parameter fusion monitoring method for intracranial pressure in patients. The computer program product can be a software installation package; when any of the aforementioned methods of the dynamic multi-parameter fusion monitoring method for intracranial pressure in patients needs to be used, the computer program product can be downloaded and executed on the computer.

[0125] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for monitoring intracranial pressure of a patient dynamically and multi-parameters fusion, characterized in that, The method includes: Acquire raw multi-parameter monitoring data of the target patient during the postoperative monitoring period; Based on the original monitoring data of multiple parameters, time-series alignment processing is performed to unify the timestamps of each parameter data, so as to obtain time-synchronized multi-parameter time-series data. Based on time-synchronized multi-parameter time-series data, a multi-dimensional feature space is constructed to characterize the correlation between various monitoring parameters, and a continuous parameter mapping surface is generated within the multi-dimensional feature space. By adaptively partitioning the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained, including: Receive time-synchronized multi-parameter time-series data, and construct a multi-dimensional feature space based on the physiological significance and clinical relevance of each monitoring parameter. Each dimension corresponds to a monitoring parameter, so that each monitoring parameter forms a spatial distribution point set in the multi-dimensional feature space. Based on the spatial distribution point set formed in the multidimensional feature space, a nonlinear mapping algorithm is used to perform surface fitting on the spatial distribution point set to obtain a continuous parametric mapping surface. Based on the parametric mapping surface, the geometric features and data distribution characteristics of the parametric mapping surface are analyzed. According to the surface curvature change gradient, data density distribution gradient and clinical risk distribution characteristics, the parametric mapping surface is adaptively divided into regions to obtain multiple feature sub-regions with differentiated data distribution characteristics. After mapping the real-time acquired time-series data points to the corresponding feature sub-regions, dynamic weighted fusion calculation is performed by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status. Based on the temporal fusion monitoring values, the preset threshold interval of the corresponding postoperative time window is first constructed as a multidimensional geometric feature region. The temporal fusion monitoring values ​​at each time point are converted into multidimensional geometric feature points, and it is determined whether the geometric feature points are within the preset multidimensional geometric feature region. The fusion monitoring values ​​are then compared with the preset threshold interval of the corresponding time window to obtain the comparison results. Based on the comparison results, when the fused monitoring value exceeds the preset threshold range, a graded early warning information corresponding to the degree of exceedance is generated and output.

2. The patient intracranial pressure dynamic multi-parameter fusion monitoring method according to claim 1, characterized in that, Based on the original multi-parameter monitoring data, time-series alignment processing is performed to unify the timestamps of each parameter, resulting in time-synchronized multi-parameter time-series data, including: Analyze the raw monitoring data of multiple parameters to identify the original sampling frequency of the raw time series data stream corresponding to different physiological parameters; Based on the original sampling frequency, and using a unified target frequency dynamically set according to clinical monitoring needs as a benchmark, the original time series data streams of each parameter are resampled using an adaptive interpolation algorithm to align the sampling times of all parameters to the same time series. Based on a unified time series, a time-synchronized multi-parameter time series data is constructed, which includes synchronized data columns for each parameter.

3. The method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients according to claim 2, characterized in that, After mapping the real-time acquired time-series data points to corresponding feature sub-regions, dynamic weighted fusion calculation is performed by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights. This ultimately yields time-series fusion monitoring values ​​used to characterize the overall intracranial pressure status, including: The system receives the current time data point from the time-synchronized multi-parameter time series data in real time. Based on multiple feature sub-regions with differentiated data distribution characteristics, it maps the current time data point to the corresponding feature sub-region to obtain the data point mapping result. Based on the data point mapping results, the statistical characteristics of the feature sub-regions that have been calculated are extracted, including the data distribution center, dispersion and boundary morphology features. At the same time, combined with the individual pathological characteristics of the target patient entered before the operation and the parameter weight configuration preset by the clinical experts, a basic parameter set for dynamic weighted fusion calculation is constructed. Based on the basic parameter set, the fusion weight coefficients of each monitoring parameter are dynamically adjusted according to the relative position of the data points in the feature sub-region, the degree of matching of individual pathological features, and the priority of clinical parameter weights. Weighted fusion calculation is then performed to obtain the fusion value of intracranial pressure status at the current moment. Based on the current intracranial pressure status fusion value, combined with the fusion value sequence of historical time points, temporal smoothing and trend analysis are performed, and finally, a temporal fusion monitoring value is output to characterize the overall intracranial pressure status.

4. The method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients according to claim 3, characterized in that, Based on time-series fusion monitoring values, a preset threshold interval for the corresponding postoperative time window is first constructed as a multidimensional geometric feature region. The time-series fusion monitoring values ​​at each time point are then converted into multidimensional geometric feature points, and it is determined whether the geometric feature points are within the preset multidimensional geometric feature region. The fusion monitoring values ​​are then compared with the preset threshold interval for the corresponding time window to obtain the comparison results, including: Receive time-series fusion monitoring values, identify the postoperative time window to which the current moment belongs according to the postoperative time window division rules, and retrieve the corresponding preset threshold interval parameters based on the time window; Based on preset threshold range parameters, a multidimensional geometric feature region is constructed, where each dimension of the multidimensional geometric feature region corresponds to a clinical risk assessment dimension. The time-series fusion monitoring values ​​are decomposed and mapped according to the clinical risk assessment dimensions to generate corresponding multi-dimensional geometric feature points. The coordinate values ​​of the multi-dimensional geometric feature points in each dimension are determined by the components of the time-series fusion monitoring values ​​in the corresponding risk dimensions. Based on multidimensional geometric feature regions and multidimensional geometric feature points, an algorithm for detecting whether a point is inside a polygon is used to determine the spatial positional relationship of the multidimensional geometric feature points relative to the multidimensional geometric feature regions, so as to obtain the judgment result, and generate the comparison result based on the judgment result.

5. The method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients according to claim 4, characterized in that, Based on multidimensional geometric feature regions and multidimensional geometric feature points, an algorithm for detecting whether a point is inside a polygon is used to determine the spatial positional relationship of the multidimensional geometric feature points relative to the multidimensional geometric feature regions, thereby obtaining a judgment result. Based on the judgment result, a comparison result is generated, including: Receive boundary coordinate data of multidimensional geometric feature regions and coordinate data of multidimensional geometric feature points, perform geometric normalization processing on the boundary coordinate data of multidimensional geometric feature regions, and obtain a normalized polygon boundary vertex sequence; Based on the standardized polygon boundary vertex sequence and coordinate data of multidimensional geometric feature points, a ray intersection counting algorithm is used to calculate the number of intersections between the test ray emanating from the multidimensional geometric feature point and the polygon boundary edge, and the result of the intersection count is obtained. Based on the crossover count, the parity of the crossover count determines whether the multidimensional geometric feature point is located within the multidimensional geometric feature region. At the same time, the minimum Euclidean distance from the multidimensional geometric feature point to the polygon boundary is calculated to obtain the spatial location judgment result. Based on the spatial location judgment results, and by comparing the minimum Euclidean distance with the preset critical distance threshold, the spatial location judgment results are mapped to clinically interpretable state identifiers, resulting in comparison results that include normal state, critical state, and abnormal state identifiers.

6. The method for dynamic multi-parameter fusion monitoring of intracranial pressure in patients according to claim 5, characterized in that, Based on the comparison results, when the fused monitoring value exceeds a preset threshold range, a graded early warning message corresponding to the degree of exceedance is generated and output, including: Receive the comparison results, analyze the clinical status identifiers contained in the comparison results, and identify whether the current intracranial pressure status is abnormal. Based on the identified abnormal states, retrieve the coordinate data of multidimensional geometric feature points and the boundary data of multidimensional geometric feature regions, and calculate the distance and direction of the multidimensional geometric feature points exceeding the boundary of the multidimensional geometric feature regions; Based on the distance and direction of the excess, combined with the individual pathological characteristics of the target patient and the clinical risk level of the postoperative time window, the clinical severity of intracranial pressure abnormality is assessed to obtain a quantitative index of the degree of excess. Based on the quantitative indicators of the degree of exceedance, and according to the preset warning classification rule base, the quantitative indicators of the degree of exceedance are mapped to the corresponding warning levels, generating classified warning information that includes the warning level, risk description and suggested handling measures. Based on the graded early warning information, the graded early warning information is transmitted to the clinical monitoring terminal in real time through the medical data communication interface, and the timestamp of the early warning event, the early warning level and the patient identification information are recorded.

7. A dynamic multi-parameter fusion monitoring system for intracranial pressure in patients, wherein the system implements the method as described in any one of claims 1 to 6, characterized in that, include: The acquisition module is used to acquire raw monitoring data of multiple parameters of the target patient during the postoperative monitoring period; The synchronization module is used to perform time-series alignment processing on the original monitoring data of multiple parameters to unify the timestamps of each parameter data and obtain time-synchronized multi-parameter time-series data. The mapping module is used to construct a multi-dimensional feature space based on time-synchronized multi-parameter time-series data to characterize the correlation between various monitoring parameters, and to generate a continuous parameter mapping surface within the multi-dimensional feature space; by adaptively dividing the parameter mapping surface, multiple feature sub-regions with differentiated data distribution characteristics are obtained. The fusion module is used to map real-time acquired time-series data points to corresponding feature sub-regions, and then perform dynamic weighted fusion calculations by combining the statistical characteristics of the feature sub-regions, the individual pathological characteristics of the target patient, and the preset clinical parameter weights to finally obtain the time-series fusion monitoring value used to characterize the overall intracranial pressure status. The comparison module is used to construct a multi-dimensional geometric feature region based on the temporal fusion monitoring value, first construct the preset threshold interval of the corresponding postoperative time window as a multi-dimensional geometric feature region, convert the temporal fusion monitoring value at each time into multi-dimensional geometric feature points, and determine whether the geometric feature points are within the preset multi-dimensional geometric feature region, and complete the comparison between the fusion monitoring value and the preset threshold interval of the corresponding time window to obtain the comparison result. The processing module is used to generate and output graded early warning information corresponding to the degree of exceedance when the fused monitoring value exceeds the preset threshold range based on the comparison results.

8. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images