Emergency cardiovascular patient early warning system

Abstract

The invention relates to the technical field of cardiovascular monitoring, in particular to an emergency cardiovascular patient early warning system which comprises a signal layering analysis module, an anomaly recognition module, an early warning task scheduling module, a signal verification module and a recovery and backup module. According to the method, the physiological signals are analyzed hierarchically, the internal relation between the signal characteristics of each layer can be deeply excavated, the accurate identification and abnormity judgment capability of the complex cardiovascular signals is improved, each parameter is independently judged according to the hierarchical relation of the signals, the missing report and the false report are effectively reduced, and the accuracy and the reliability of abnormity monitoring are enhanced; a processing strategy can be dynamically adjusted under multi-source signal interaction, the real-time performance of signal processing and anomaly detection is ensured, a periodic verification and recovery mechanism ensures stable operation of the system under a fault or anomaly condition, the recovery capability of the system is optimized, the reliability of the whole monitoring system is improved, and more stable and accurate data support is provided for emergency medical treatment.

Description

Technical Field

[0001] This invention relates to the field of cardiovascular monitoring technology, and in particular to an early warning system for emergency cardiovascular patients. Background Technology

[0002] The field of cardiovascular monitoring technology falls under the category of medical engineering and physiological parameter measurement technology. Its core content includes the acquisition, analysis, and monitoring of physiological signals of the human cardiovascular system. By measuring parameters such as heart rate, blood pressure, pulse waveform, and electrocardiogram signals, it achieves real-time monitoring of the circulatory system's state. This field involves key components such as physiological signal acquisition sensors, signal amplification and conversion circuits, physiological data analysis algorithms, and monitoring terminal equipment. It aims to achieve quantitative measurement and dynamic observation of cardiovascular activity, providing real-time data support and health status assessment for critically ill patients.

[0003] Traditional emergency cardiovascular patient early warning systems refer to systems used in emergency medicine scenarios to monitor patients' cardiovascular parameters and provide abnormal alerts based on preset thresholds. These systems collect electrocardiogram signals, pulse signals, and blood pressure fluctuation data through electrodes or photoelectric sensors. The data parameters that can be used for judgment are obtained through signal amplification and analog-to-digital conversion circuits. The current physiological data is then compared with the reference range using preset rules. When the detection result exceeds the normal range, an audible and visual alarm or a control terminal alert is triggered. Traditional systems mostly use a single-parameter threshold judgment method for abnormal identification, relying on fixed numerical thresholds and manual intervention to achieve the early warning function.

[0004] Current technologies in emergency cardiovascular patient early warning systems primarily rely on collecting electrocardiogram (ECG), pulse, and blood pressure fluctuation data, combined with fixed threshold values ​​and manual intervention, to determine anomalies. This approach depends on a pre-set single parameter threshold and fails to effectively consider the dynamic correlation and interaction between multiple physiological parameters. Consequently, it cannot accurately reflect the patient's cardiovascular health status in practical applications, especially when multi-source data is complex and subject to frequent false negatives or missed positives due to threshold judgment errors. Furthermore, existing systems have weak capabilities in processing and recovering abnormal signals, failing to perform effective real-time verification and dynamic recovery for abnormal nodes. Therefore, they cannot guarantee the accuracy and reliability of data in the event of system failure, impacting the timeliness and accuracy of clinical decision-making. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an early warning system for emergency cardiovascular patients.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an emergency cardiovascular patient early warning system includes:

[0007] The signal layering and analysis module acquires multi-source physiological signals collected by cardiovascular monitoring equipment, extracts key feature points of the signals, analyzes the correlation patterns between feature points, and generates a cardiovascular signal layering and analysis template.

[0008] Based on the cardiovascular signal hierarchical analysis template, the anomaly identification module performs independent anomaly determination on signal nodes at each level, records the node numbers and logical relationships in the anomaly determination path, and establishes an anomaly signal determination mapping table.

[0009] The early warning task scheduling module collects the processing requirements of each level of signal node according to the abnormal signal determination mapping table, analyzes the performance parameters and load capacity of the processing unit, allocates signals to the appropriate processing unit, and generates a signal processing allocation scheme.

[0010] Based on the signal processing allocation scheme, the signal verification module performs periodic verification operations on the signals in the processing unit, records the abnormal node numbers and verification timestamps in the verification results, and generates a list of cardiovascular signal abnormality early warning records.

[0011] As a further embodiment of the present invention, the cardiovascular signal hierarchical analysis template includes a set of key feature points, a feature point association pattern map, and a hierarchical node mapping table. The abnormal signal determination mapping table includes a node abnormal determination path, an abnormal level index, and a set of logical relationship types. The signal processing allocation scheme includes a list of processing unit performance parameters, a signal node processing position mapping, and a processing capacity allocation result. The cardiovascular signal abnormality early warning record list includes a set of abnormal node numbers, a verification timestamp sequence, and a verification result status marker.

[0012] As a further aspect of the present invention, the signal layering analysis module includes:

[0013] The feature point extraction submodule acquires multi-source physiological signals collected by cardiovascular monitoring equipment, performs feature point annotation operations on them, records the index positions of key feature points, and classifies the signals by level by comparing the correlation patterns of feature points to obtain the distribution results of key feature points;

[0014] The correlation pattern parsing submodule extracts the correlation patterns between feature points based on the distribution results of the key feature points, truncates the feature points based on the logical intervals of the signal, constructs a feature point correlation pattern map, and reorganizes the feature points in combination with signal information to obtain a feature point correlation pattern set.

[0015] The hierarchical template generation submodule rearranges the feature point set according to the feature point association pattern set, logically splices multiple feature points at the same level, and integrates the node information of each level to obtain a cardiovascular signal hierarchical analysis template.

[0016] As a further aspect of the present invention, the anomaly identification module includes:

[0017] The anomaly detection submodule, based on the cardiovascular signal hierarchical analysis template and combined with the hierarchical labels of signal nodes, compares and sorts the nodes at each level according to their hierarchical label priority values, using the following formula:

[0018] ;

[0019] The priority evaluation value of the compute nodes is used to independently determine the anomalies of the nodes in order from top to bottom, and the anomaly determination index value is obtained.

[0020] Where P represents the priority evaluation value of the node. The weight representing the label at level i of node i. The signal strength representing node i, The highest level label priority of the node is represented by n, where n represents the total number of nodes.

[0021] The path recording submodule identifies the set of node numbers in the anomaly determination path based on the anomaly determination index value, records the connection direction of each pair of adjacent nodes, and integrates the anomaly edge information by combining the node connection relationship in the path to generate anomaly determination path map data.

[0022] The mapping table generation submodule collects the node numbers and adjacent node pairs in the connecting edges based on the anomaly determination path graph data, stores the upstream and downstream relationship types and connection directions between each node according to the adjacent structural relationship, and obtains the anomaly signal determination mapping table.

[0023] As a further aspect of the present invention, the early warning task scheduling module includes:

[0024] Based on the abnormal signal determination mapping table, the demand analysis submodule collects the processing demand parameters of each level of signal node, records the node number and demand type in the processing demand, and generates the processing demand distribution result by combining the processing demand type.

[0025] The performance evaluation submodule analyzes the performance parameters and load capacity of the processing unit based on the processing demand distribution results, extracts the processing unit's computing speed, response time, and available resource information, and generates a list of processing unit performance parameters.

[0026] The processing allocation submodule allocates signal nodes to suitable processing units according to the processing unit performance parameter list, records the correspondence between processing unit numbers and signal node numbers, integrates processing allocation information, and generates a signal processing allocation scheme.

[0027] As a further aspect of the present invention, the signal verification module includes:

[0028] Based on the signal processing allocation scheme, the verification task scheduling submodule collects the signal node number information in the processing unit, extracts the processing position information of each signal node in sequence and establishes a signal node mapping set, calculates the index value of the processing position information, marks the processing unit identifier and verification cycle parameter to which the signal node belongs, and obtains the signal node number value.

[0029] The verification execution submodule performs verification operations on the signal nodes in the processing unit according to the signal node number value, extracts the abnormal node numbers from the verification results, records the verification timestamp, compares it with the verification benchmark value, filters out abnormal node combinations that meet the conditions, and establishes an abnormal node number set.

[0030] The log generation submodule queries the original signal node identifier based on the set of abnormal node numbers and the combination of abnormal nodes corresponding to the numbers, integrates the signal node identifier and the verification timestamp information, and generates a list of cardiovascular signal abnormality warning records.

[0031] As a further aspect of the present invention, the step of extracting the abnormal node number from the verification result refers to verifying the signal node based on the signal node number value, selecting the signal node number that exceeds the preset deviation threshold from the verification output as the abnormal node number, and recording the abnormal node number and the corresponding verification timestamp in chronological order to obtain the abnormal node number.

[0032] The process of selecting anomalous node combinations that meet the criteria involves comparing the anomalous node number data based on the verification benchmark value, aggregating the anomalous node numbers that exceed a set threshold number of anomalous nodes within the same verification timestamp, and forming anomalous node combination that includes the corresponding verification timestamp.

[0033] As a further aspect of the present invention, the system also includes a recovery and backup module:

[0034] The recovery and backup module, based on the cardiovascular signal abnormality early warning record list, counts the processing units to which abnormal nodes belong, assigns abnormal nodes to the corresponding processing units, analyzes the recovery correspondence between processing units and signal nodes, and generates a cardiovascular signal recovery and backup table.

[0035] The cardiovascular signal recovery and backup table includes processing unit identifiers, abnormal node grouping results, and recovery attribution mapping results.

[0036] As a further aspect of the present invention, the recovery and backup module includes:

[0037] The unit extraction submodule collects the processing unit corresponding to each abnormal node according to the cardiovascular signal abnormality early warning record list, records the abnormal node number associated with the processing unit and the number of its corresponding signal node sets, determines the matching index between the abnormal node and the processing unit, and obtains the unit number value to which the abnormal node belongs.

[0038] The node classification submodule classifies the corresponding signal nodes into each processing unit based on the abnormal node's unit number value, and extracts the node number list of each unit to obtain the node affiliation strength value of the processing unit.

[0039] The structure restoration generation submodule integrates the processing unit and its subordinate signal node numbers through structural mapping based on the node affiliation strength value of the processing unit, outputs the processing unit index, the corresponding abnormal node number and the total number of nodes, determines the affiliation of the unit and the signal node, and generates a cardiovascular signal recovery and backup table.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] This invention utilizes a hierarchical approach to physiological signal analysis, enabling in-depth exploration of the intrinsic connections between signal features at different levels. This significantly enhances the accuracy of identifying and judging anomalies in complex cardiovascular signals. By independently determining various parameters based on signal hierarchy, it effectively reduces false alarms and missed detections, improving the accuracy and reliability of anomaly monitoring. Furthermore, it dynamically adjusts processing strategies under multi-source signal interaction, ensuring real-time signal processing and anomaly detection. The periodic signal verification and recovery mechanism not only ensures stable system operation under fault or abnormal conditions but also accurately locates and repairs abnormal nodes when faults occur, thereby optimizing the system's recovery capability, improving the overall reliability of the monitoring system, and providing more stable and accurate data support for emergency medicine. Attached Figure Description

[0042] Figure 1 This is a system flowchart of the present invention;

[0043] Figure 2 This is a flowchart of the signal layering analysis module in this invention;

[0044] Figure 3 This is a flowchart of the anomaly identification module in this invention;

[0045] Figure 4 This is a flowchart of the early warning task scheduling module in this invention;

[0046] Figure 5 This is a flowchart of the signal verification module in this invention;

[0047] Figure 6This is a flowchart of the recovery and backup module in this invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Please see Figure 1 An emergency cardiovascular patient early warning system includes:

[0051] The signal layering and analysis module acquires multi-source physiological signals collected by cardiovascular monitoring equipment, extracts key feature points of the signals, analyzes the correlation patterns between feature points, and generates a cardiovascular signal layering and analysis template.

[0052] The anomaly identification module is based on the cardiovascular signal hierarchical analysis template. It performs independent anomaly determination on signal nodes at each level, records the node numbers and logical relationships in the anomaly determination path, and establishes an anomaly signal determination mapping table.

[0053] The early warning task scheduling module collects the processing requirements of each level of signal nodes based on the abnormal signal determination mapping table, analyzes the performance parameters and load capacity of the processing units, allocates signals to the appropriate processing units, and generates a signal processing allocation scheme.

[0054] Based on the signal processing and allocation scheme, the signal verification module performs periodic verification operations on the signals in the processing unit, records the abnormal node numbers and verification timestamps in the verification results, and generates a list of cardiovascular signal abnormality early warning records.

[0055] The recovery and backup module, based on the list of abnormal cardiovascular signal warning records, counts the processing units to which abnormal nodes belong, assigns abnormal nodes to the corresponding processing units, analyzes the recovery correspondence between processing units and signal nodes, and generates a cardiovascular signal recovery and backup table.

[0056] The cardiovascular signal hierarchical analysis template includes a set of key feature points, a feature point association pattern map, and a hierarchical node mapping table. The abnormal signal judgment mapping table includes node abnormal judgment paths, abnormal level indexes, and a set of logical relationship types. The signal processing allocation scheme includes a list of processing unit performance parameters, signal node processing location mapping, and processing capacity allocation results. The cardiovascular signal abnormality early warning record list includes a set of abnormal node numbers, a verification timestamp sequence, and a verification result status marker. The cardiovascular signal recovery and backup table includes processing unit identifiers, abnormal node grouping results, and recovery attribution mapping results.

[0057] Please see Figure 2 The signal layering analysis module includes:

[0058] The feature point extraction submodule acquires multi-source physiological signals collected by cardiovascular monitoring equipment, performs feature point annotation operations on them, records the index positions of key feature points, and classifies the signals by level by comparing the correlation patterns of feature points to obtain the distribution results of key feature points;

[0059] Data acquisition is performed by synchronously receiving analog signals from front-end sensors in real time via an analog-to-digital converter interface configured with a 500Hz sampling rate. This operation involves the parallel digitization of the potential changes in a three-lead ECG and the intensity changes in a dual-wavelength photoplethysmography (PPG) pulse wave. The acquired raw discrete digital sequences then enter the digital signal processing flow. A finite impulse response (FIR) bandpass filter bank with passband cutoff frequencies set to 0.5Hz and 40Hz is applied to filter out low-frequency baseline drift and suppress power line interference and high-frequency electromyographic noise. The filtered ECG sequence then enters the feature recognition stage. An algorithm based on the sum of squares of the first derivative of the signal combined with an adaptive sliding window threshold judgment is used to identify the R-wave peak position in the QRS complex. Simultaneously, the program performs a local maximum search on the PPG signal within a fixed 300ms time window after detecting the R-wave peak trigger point to locate the systolic peak. Feature point annotation is performed to record key indices. The ECG R-wave peak point is marked with an index of 1250 in the current sample buffer, and the corresponding PPG systolic peak point within the time window is marked with an index of 1400. By comparing the correlation patterns of feature points, the index difference between two feature points on the discrete time axis is calculated, revealing that the R-wave peak point leads the PPG peak point by 150 sampling points. This time difference reflects the conduction delay from cardiac electrical excitation to the arrival of the peripheral mechanical pulse. Based on this sequential relationship, the time-leading ECGR wave feature point is classified into the first-level signal node category, and the time-lagging PPG peak feature point is classified into the second-level signal node category, thus obtaining the distribution results of key feature points reflecting physiological time sequence.

[0060] The correlation pattern parsing submodule extracts the correlation patterns between feature points based on the distribution results of key feature points, truncates the feature points based on the logical intervals in the signal, constructs a feature point correlation pattern map, and reorganizes the feature points in combination with signal information to obtain a set of feature point correlation patterns.

[0061] Based on the first-level and second-level node time series classification established from the key feature point distribution results, a cross-level association pattern is defined to characterize the conduction time interval characteristics of electro-mechanical coupling. Data truncation is performed based on the logical index interval of feature points in the continuous signal stream. This operation relies on a circular buffer mechanism. The starting boundary of the truncation window is set at the position of the first 50 sample points before the index of the first-level R-wave feature points, and the ending boundary is set at the position of the last 50 sample points after the index of the second-level PPG peak feature points. Through pointer operations, a total of 250 sample-length signal segments containing the complete propagation process from cardiac electroexcitation to mechanical pulsation are extracted from the buffer. The feature point association pattern map is constructed using a directed graph data structure. The source node instantiation object represents the first-level R-wave event, and the target node instantiation object represents the second-level PPG peak event. A one-way connection edge is established between the two nodes, and the calculated difference between 150 sample intervals is assigned as the weight attribute of the connection edge to quantify the time distance between the nodes. The feature points are recombined with signal information. The morphological analysis subroutine is called to measure the QRS group width in the ECG truncation segment and quantize it into a 90ms value, which is then assigned to the attribute field of the source node. The original amplitude of the PPG truncation period is normalized to obtain a dimensionless normalized amplitude, which is then mapped to a quantized voltage value of 1.2V according to a preset voltage mapping rule and written into the attribute field of the target node. This completes the data filling work of the map nodes, and finally, the feature point association pattern set is obtained.

[0062] The hierarchical template generation submodule rearranges the feature point set according to the feature point association pattern set through logical order, logically splices multiple feature points at the same level, and integrates the node information of each level to obtain a cardiovascular signal hierarchical analysis template.

[0063] Based on the feature point association pattern set data, following the inherent anatomical logical order of cardiac physiological activity from the initiation of sinoatrial node electrical excitation to the peripheral vascular mechanical perfusion response, a rearrangement operation is performed on the feature point set to establish the temporal logical priority of first-level electrical signal feature points over second-level mechanical signal feature points. A timestamp sorting algorithm is used to globally align the order of all identified feature point events. Multiple feature points at the same level are logically concatenated, connecting multiple first-level R-wave feature points identified within a continuous monitoring period into a unidirectional linked list structure representing the rhythm of cardiac electrical activity according to their timestamp order. The same linked list construction operation is performed for second-level PPG feature points. By integrating the node information of each level, the first-level linked list is traversed to collect the precise time values, discretized amplitudes, and waveform width quantization information of all nodes. Similarly, the corresponding parameter information of all nodes in the second-level linked list is collected. The scattered information elements are mapped to a predefined hierarchical data model to form a structured hierarchical data definition. This definition includes attribute descriptors of nodes at each level, connection relationship descriptors between levels, and overall temporal constraints. Finally, a cardiovascular signal hierarchical analysis template is obtained for subsequent analysis.

[0064] Please see Figure 3 The anomaly detection module includes:

[0065] The anomaly detection submodule, based on the cardiovascular signal hierarchical analysis template and combined with the hierarchical labels of signal nodes, compares and sorts nodes at each level according to their hierarchical label priority values, using the following formula:

[0066] ;

[0067] The priority evaluation value of the compute nodes is used to independently determine the anomalies of the nodes in order from top to bottom, and the anomaly determination index value is obtained.

[0068] Where P represents the priority evaluation value of the node. The weight representing the label at level i of node i. The signal strength representing node i, The highest level label priority of the node is represented by n, where n represents the total number of nodes.

[0069] Based on the cardiovascular signal hierarchical analysis template, the hierarchical structure defined in the template is read. The first level represents cardiac electrophysiological activity, and the second level represents peripheral hemodynamic activity. The priority value of the hierarchical label for the first level is set to 10 based on physiological criticality, and the priority value for the hierarchical label for the second level is set to 5. Nodes in each level are compared and sorted according to their hierarchical label priority values. The first level with higher priority values ​​is selected for detailed evaluation, and the total number of nodes in the first level within the current time window is identified. The key nodes are the R wave, T wave, and P wave. Hierarchical label weights are assigned to each node based on its clinical diagnostic importance. Set the weight of R-wave (node ​​1) T-wave (node ​​2) weight P-wave (node ​​3) weight Measure the normalized signal strength of each node relative to the baseline. R-wave intensity was measured T-wave intensity P-wave intensity Confirm the highest level label priority of the node in the current processing level. The value is 10. Here, a formula is introduced to calculate the priority evaluation value P of the node. The formula is: In the formula, P represents the priority evaluation value of the node, which is used to quantify the overall anomaly risk level of the signal node at the current level. The weight of the node i-th level label reflects the relative importance of the node in clinical diagnosis. The value ranges from 0 to 1 and the sum of the weights of the same level is 1. The signal strength representing node i is a normalized amplitude or energy index used to reflect the significance of node activity. The highest level label priority of the node is used to amplify the evaluation value according to the physiological criticality of the level; n represents the total number of nodes, which refers to the number of nodes of the same level currently participating in the evaluation. The symbol indicates that the calculation results within the parentheses are summed. The symbol represents the square root operation of the summation result. The weighted summation square root method can highlight the impact of high-weight and high-intensity nodes on the overall risk and avoid noise interference from low-intensity nodes.

[0070] Substituting the above actual parameters into the formula for calculation: First, calculate the squared term of the weighted intensity for each node, where node 1 is... Node 2 is Node 3 is ;

[0071] Next, sum the squared terms:

[0072] ;

[0073] Then perform the square root operation on the sum:

[0074] ;

[0075] Finally, multiply by the highest level priority to obtain the final evaluation value:

[0076] ;

[0077] The calculation result of 5.1575 indicates the weighted value of the comprehensive activity intensity and importance of the current first-level node. This result is compared with the preset normal baseline value range [2.0, 4.5]. Since P=5.1575 exceeds the upper limit of the baseline value of 4.5, it is determined that the first-level node has potential anomalies. The nodes are independently judged for anomalies in the order from the top to the bottom. Similar calculations and judgments are continued for the second level, and the identifiers of the first-level nodes judged to be abnormal are recorded to obtain the anomaly judgment index value.

[0078] The path recording submodule identifies the set of node numbers in the anomaly determination path based on the anomaly determination index value, records the connection direction of each pair of adjacent nodes, and integrates the anomaly edge information by combining the node connection relationship in the path to generate anomaly determination path graph data.

[0079] Based on the anomaly determination index value, the list of first-level anomaly nodes identified in the index is read and parsed. The key node numbers N_R (representing R wave) and N_T (representing T wave) that caused the first-level anomaly are locked. Each pair of adjacent nodes is numbered and their connection direction is recorded. Based on the chronological order of the events, the connection direction from N_R to N_T is determined to be a forward temporal connection. Combining the node connection relationship in the path, the cross-level causal link mapping table defined in the hierarchical parsing template is called to identify a predefined driver-response relationship between the first-level node N_R and the subsequent second-level PPG peak node N_P. By integrating abnormal edge information and calculating the time difference between the occurrence of N_R and N_P, when the actual transmission time interval is detected to exceed the preset physiological normal range threshold (e.g., exceeding the 250ms setting), the directed edge connecting N_R and N_P is marked as an abnormal connection edge with the "transmission delay" attribute. A graph structure data containing nodes N_R, N_T, N_P and their normal connection edges and the connection edges marked as abnormal is constructed, and finally, an abnormal judgment path map data describing the abnormal propagation link is generated.

[0080] The mapping table generation submodule collects the node numbers and adjacent node pairs in the connecting edges based on the anomaly judgment path graph data, stores the upstream and downstream relationship types and connection directions between each node according to the adjacent structure relationship, and obtains the anomaly signal judgment mapping table.

[0081] Based on the anomaly determination path graph data, a graph data parsing operation is performed. All connection edge information in the graph structure is traversed to filter records marked with anomaly attributes. The starting node number N_R and the ending node number N_P connected to the anomaly edge with the "conduction delay" attribute are extracted. According to the adjacent structure relationship database, it is confirmed that N_R belongs to the first level of electrical signal category and N_P belongs to the second level of mechanical signal category. The two constitute a "electric-mechanical coupling" physical adjacent structure. The upstream and downstream relationship type and connection direction data between each node are stored. N_R is defined as the upstream driving node role and N_P is defined as the downstream response node role. The connection direction is recorded as from N_R to N_P. The relationship type between the two is classified and stored as "causal-delay type" anomaly category. A structured data entry containing a unique node number identifier, the level code, the upstream and downstream role definition, the connection direction indication, and the anomaly relationship type description is established. All such entries are summarized to obtain a complete anomaly signal determination mapping table.

[0082] Please see Figure 4 The early warning task scheduling module includes:

[0083] The requirements analysis submodule collects the processing requirements parameters of each level of signal nodes based on the abnormal signal judgment mapping table, records the node number and requirement type in the processing requirements, and generates the processing requirement distribution results by combining the processing requirement type.

[0084] Based on the anomaly signal determination mapping table, the "causal-delayed" anomaly entries for nodes N_R and N_P are read. For the upstream node N_R, belonging to the first level, it is determined that high-precision waveform morphology analysis is required to screen for subtle arrhythmia features. A query of the pre-defined processing requirement database reveals that the computational load parameter required for this specific analysis algorithm is 50 MFLOPs. For the downstream node N_P, belonging to the second level, it is determined that trend tracking analysis is required to assess long-term blood perfusion. A query of the database reveals that the required computational load parameter is 10 MFLOPs. The correspondence between node numbers and requirement types in the processing requirements is recorded. Based on the relative proportion of computational load and storage requirements, node N_R is marked as a "computation-intensive" requirement category, and node N_P is marked as a "storage-intensive" requirement category because it needs to cache longer historical data segments for trend fitting. Combining the processing requirement type information, a distribution view data containing the computational load values ​​and storage type requirements of each node is generated, resulting in the final processing requirement distribution result.

[0085] The performance evaluation submodule analyzes the performance parameters and load capacity of the processing unit based on the distribution of processing requirements, extracts the processing unit's computing speed, response time, and available resource information, and generates a list of processing unit performance parameters.

[0086] Based on the distribution of processing demands, a status query request is initiated through the underlying hardware resource management interface to perform real-time status polling on the two currently running independent processing units PU_A (Dedicated Digital Signal Processor architecture) and PU_B (General Purpose Central Processing Unit core architecture). Key performance indicators of the processing units are extracted, including processing speed, interrupt response time, and currently available storage resources. Measurements taken from the underlying driver show that PU_A's specific instruction set clock speed is 2.5 GHz, its average interrupt service routine response time is 5 microseconds, and the currently available on-chip high-speed static random access memory (SRAM) is 512 MB. Simultaneously, PU_B's clock speed is measured at 3.0 GHz, its average task context switching response time is 20 microseconds, and its currently available dynamic random access memory (DRAM) system memory is 2 GB. The collected hardware metrics data are organized into a structured parameter list. This list records in detail the key performance metrics of the currently available processing units PU_A and PU_B, including the clock speed, response latency to external task requests, and the amount of memory available for dynamic allocation at the current moment. Finally, a processing unit performance parameter list is generated.

[0087] The processing and allocation submodule allocates signal nodes to suitable processing units based on the performance parameter list of processing units, records the correspondence between processing unit numbers and signal node numbers, integrates processing and allocation information, and generates a signal processing and allocation scheme.

[0088] Based on the list of processing unit performance parameters, and comparing the real-time performance status data of the processing units with the previously generated processing demand distribution data, a resource matching strategy algorithm is executed. Given that node N_R is marked as a "computation-intensive" requirement (computational load index 50 MFLOPs) and has high timeliness requirements for waveform analysis, the algorithm prioritizes its allocation to DSP processing unit PU_A, which has a lower interrupt response time (measured value 5μs) and a hardware architecture adept at specific vector operations such as Fast Fourier Transform. Given that node N_R is marked as a "storage-intensive" requirement and has a significant dependence on large-capacity data cache space, the algorithm allocates it to general-purpose CPU processing unit PU_B, which has larger available general-purpose storage resources (measured value 2GB). The exact correspondence between processing unit numbers and signal node numbers is recorded, resource mapping pairs (PU_A, N_R) and (PU_B, N_R) are established, and all processing allocation information is integrated to form a sequence of scheduling instruction sets containing the unique identifiers of all nodes to be processed and their target processing unit physical addresses, thus generating the final signal processing allocation scheme.

[0089] Please see Figure 5 The signal verification module includes:

[0090] The verification task scheduling submodule, based on the signal processing allocation scheme, collects the signal node number information in the processing unit, sequentially extracts the processing position information of each signal node, and establishes a signal node mapping set using the formula:

[0091] ;

[0092] Calculate the index value of the processing location information, mark the processing unit identifier and verification cycle parameter to which the signal node belongs, and obtain the signal node number value;

[0093] Where M represents the index value for processing location information. This represents the numerical value of the processing position information recorded by the j-th signal node in the processing unit. This represents the reference position information value recorded by the j-th signal node in the signal node mapping set. This represents the weight parameter value of the processing location information recorded for the j-th signal node in the signal node mapping set. The value of the processing position information deviation amplification factor for the j-th signal node;

[0094] Based on the signal processing allocation scheme, the instructions in the scheme regarding the allocation of node N_R to processing unit PU_A are read. The real-time input buffer of processing unit PU_A is accessed, and the processing position information of each signal node is extracted sequentially and a signal node mapping set is established. The current write pointer position value of node N_R in the PU_A buffer is extracted. The value is 1024, and the reference synchronization position value corresponding to node N_R is retrieved from the pre-established ideal synchronization mapping set. The value is 1020, which represents the weight parameter value for obtaining the processing location information of this node. Set to 1.0 (indicating standard importance), and obtain the value of the deviation amplification factor for the processing position information of this node. Set to 10.0 (to amplify small synchronization deviations on the critical path), the index value M for processing location information is calculated using the following formula: ;

[0095] In the formula, M represents the index value of the processed position information, which is used to quantify the degree of synchronization deviation between the actual position of the signal node in the processing unit and the ideal reference position;

[0096] This represents the actual processing position information recorded by the j-th signal node in the processing unit, such as the current pointer address of the buffer;

[0097] This represents the reference position information value recorded by the j-th signal node in the signal node mapping set, such as the expected synchronization point address;

[0098] The value of the weight parameter representing the processing position information recorded by the j-th signal node in the signal node mapping set is used to adjust the weight of the influence of different node deviations on the final feature value.

[0099] The value of the processing position information deviation amplification factor for the j-th signal node is used to nonlinearly amplify the deviation of a specific key node to improve sensitivity.

[0100] The symbol indicates the absolute value operation, ensuring that the direction of the deviation does not affect the amplitude calculation;

[0101] The exponentiation, or square root operation, is used to smooth the data distribution of eigenvalues.

[0102] Substitute the above actual parameters into the formula for calculation: First, calculate the absolute deviation between the actual position and the reference position. Next, multiply the deviation by the amplification factor. Then divide by the weight parameter. Finally, the square root of the result is obtained. ;

[0103] The calculation result 6.324 indicates the position synchronization deviation characteristic value of node N_R in processing unit PU_A. The processing unit identifier and verification period parameter to which the signal node belongs are marked. The identifier PU_A and a high-frequency verification period parameter of 10ms are bound to the data structure of node N_R to obtain the signal node number value.

[0104] The verification execution submodule performs verification operations on the signal nodes in the processing unit based on the signal node number value, extracts the abnormal node numbers from the verification results, records the verification timestamp, compares it with the verification benchmark value, filters out the abnormal node combinations that meet the conditions, and establishes an abnormal node number set.

[0105] Extracting the abnormal node number from the verification result refers to verifying the signal node based on the signal node number value, selecting the signal node number that exceeds the preset deviation threshold from the verification output as the abnormal node number, and recording the abnormal node number and the corresponding verification timestamp in chronological order to obtain the abnormal node number.

[0106] Filtering out abnormal node combinations that meet the criteria involves comparing the abnormal node number data based on the verification benchmark value, aggregating the abnormal node numbers that exceed the set abnormal number threshold within the same verification timestamp, and forming an abnormal node combination that includes the corresponding verification timestamp.

[0107] Based on the signal node number, the signal node N_R data block in the internal storage space of the processing unit PU_A is locked. A cyclic redundancy check (CRC) calculation task is initiated for the N_R data block. A new checksum is generated using a preset polynomial algorithm, and the calculation result is compared bit-by-bit with the original CRC checksum carried in the header field of the data block. Abnormal node numbers are extracted from the check results. When the comparison result shows a CRC value mismatch, it is confirmed that the data integrity of node N_R is compromised, and its number N_R is extracted as the abnormal node number. Simultaneously, a check timestamp is recorded, using a high-precision clock source to obtain the current system's accurate time as the check timestamp. The previously calculated mapping characteristic value (6.324) is compared with the preset maximum allowable synchronization deviation benchmark value (5.0), confirming that the actual deviation value exceeds the allowable range. Abnormal node combinations that meet the criteria are filtered. Within a 10ms time window centered on the recorded timestamp, the abnormal event log is searched, confirming that a total of 5 nodes (including N_R) have experienced similar check errors or deviation exceeding the limit. Since this number exceeds the set threshold of 3 abnormal nodes, these 5 nodes are aggregated into a related abnormal group, and an abnormal node number set is established.

[0108] The log generation submodule queries the original signal node identifier based on the set of abnormal node numbers and the combination of abnormal nodes corresponding to the numbers, integrates the signal node identifier and the verification timestamp information, and generates a list of cardiovascular signal abnormality warning records.

[0109] Based on the set of abnormal node numbers, the system iterates through the abnormal combination records containing node N_R. Using node number N_R as the primary index key, it queries the system's global configuration database and retrieves the original business identifier string corresponding to N_R, which is "ECG_LeadII_RPeak_ID45A". The system integrates the signal node identifier and verification timestamp information, combining the retrieved original identifier string, the record's verification timestamp string "2024-05-20-T10:30:15.123Z", and the specific abnormal type description string "CRC check failed and synchronization deviation exceeded". A detailed warning record is generated, containing the event ID, timestamp, original identifier of the involved node, abnormal category code, and a detailed description. This record is appended to the currently active activity log file of the system for persistent storage via a file input / output interface, ultimately generating a list of cardiovascular signal abnormality warning records.

[0110] Please see Figure 6 The recovery and backup module includes:

[0111] The unit extraction submodule collects the processing unit corresponding to each abnormal node based on the cardiovascular signal abnormality early warning record list, records the abnormal node number associated with the processing unit and the number of its corresponding signal node sets, determines the matching index between the abnormal node and the processing unit, and obtains the unit number value to which the abnormal node belongs.

[0112] Based on the cardiovascular signal anomaly warning record list data, the details of the anomaly record item with the identifier "ECG_LeadII_RPeak_ID45A" are parsed and read. By querying the historical scheduling record time series database, it is confirmed that the processing unit PU_A was responsible for performing the processing task at the exact time when the anomaly occurred. The number of the anomaly node associated with the processing unit and the number of its corresponding signal node sets are recorded. By scanning the current running status context data structure of PU_A, it is found that PU_A is currently associated with a total of 3 different anomaly node numbers, including N_R. At the same time, by querying the buffer management register, it is confirmed that a total of 150 signal node set data packets awaiting processing reside in the input queue of PU_A. The matching index relationship between the anomaly node and the processing unit is determined. A clear attribution index entry is established in memory, pointing from the unique identifier of the anomaly node N_R to the physical identifier of the processing unit PU_A, and the unit number value of the anomaly node is obtained.

[0113] The node classification submodule classifies the corresponding signal nodes into each processing unit based on the unit number value of the abnormal node, and extracts the node number list of each unit to obtain the node affiliation strength value of the processing unit.

[0114] Based on the unit number values ​​of the abnormal nodes, the processing unit PU_A is identified as a physical unit with potential operational risks. All 150 signal nodes currently residing in the input buffer queue of PU_A are uniformly classified and marked as the "risk group to which PU_A belongs". The node number list to which each unit belongs is extracted. The risk group is traversed to generate a list containing the complete and unique identifiers of these 150 nodes. The node affiliation strength value index of the processing unit is obtained. By performing a division operation, the ratio of the number of currently confirmed abnormal nodes (value is 3) to the total number of nodes to be processed in the unit (value is 150) is calculated to obtain the node affiliation strength value of the processing unit. The calculation result is 0.02. This value quantifies the proportion of the current impact of known abnormal events on PU_A in the overall load.

[0115] The structure restoration generation submodule integrates the processing unit with its subordinate signal node numbers by performing structural mapping based on the node affiliation strength value of the processing unit. It outputs the processing unit index, the corresponding abnormal node number and the total number of nodes, determines the affiliation of the unit and the signal node, and generates a cardiovascular signal restoration and backup table.

[0116] Based on the node affiliation strength value data of the processing unit, the calculated PU_A affiliation strength value of 0.02 is compared with the preset risk level threshold. It is determined that the unit is at a low risk level but needs to be kept under watch. A dedicated backup list data structure is created in memory. This structure uses the physical index of PU_A as the core field, attaches the main node number N_R that caused the anomaly as the associated event field, and creates a pointer array to associate a list of all 150 signal node numbers belonging to the unit. The processing unit index, the corresponding abnormal node number, and the total number of nodes are output. The PU_A processing unit itself and its currently associated 150 node data packets are clearly identified as the target object set of a complete backup operation. After determining the affiliation relationship between the unit and the signal nodes, a dynamic rule is added to the system's global backup strategy configuration table, specifying that a differential backup operation based on the current memory snapshot state should be performed on PU_A immediately, generating the final cardiovascular signal recovery and backup table.

[0117] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An early warning system for emergency cardiovascular patients, characterized in that, The system includes: The signal layering and analysis module acquires multi-source physiological signals collected by cardiovascular monitoring equipment, extracts key feature points of the signals, analyzes the correlation patterns between feature points, and generates a cardiovascular signal layering and analysis template. Based on the cardiovascular signal hierarchical analysis template, the anomaly identification module performs independent anomaly determination on signal nodes at each level, records the node numbers and logical relationships in the anomaly determination path, and establishes an anomaly signal determination mapping table. The early warning task scheduling module collects the processing requirements of each level of signal node according to the abnormal signal determination mapping table, analyzes the performance parameters and load capacity of the processing unit, allocates signals to the appropriate processing unit, and generates a signal processing allocation scheme. Based on the signal processing allocation scheme, the signal verification module performs periodic verification operations on the signals in the processing unit, records the abnormal node numbers and verification timestamps in the verification results, and generates a list of cardiovascular signal abnormality early warning records.

2. The emergency cardiovascular patient early warning system according to claim 1, characterized in that, The cardiovascular signal hierarchical analysis template includes a set of key feature points, a feature point association pattern map, and a hierarchical node mapping table. The abnormal signal determination mapping table includes a node abnormal determination path, an abnormal level index, and a set of logical relationship types. The signal processing allocation scheme includes a list of processing unit performance parameters, a signal node processing position mapping, and a processing capacity allocation result. The cardiovascular signal abnormality early warning record list includes a set of abnormal node numbers, a verification timestamp sequence, and a verification result status marker.

3. The emergency cardiovascular patient early warning system according to claim 1, characterized in that, The signal layering analysis module includes: The feature point extraction submodule acquires multi-source physiological signals collected by cardiovascular monitoring equipment, performs feature point annotation operations on them, records the index positions of key feature points, and classifies the signals by level by comparing the correlation patterns of feature points to obtain the distribution results of key feature points; The correlation pattern parsing submodule extracts the correlation patterns between feature points based on the distribution results of the key feature points, truncates the feature points based on the logical intervals of the signal, constructs a feature point correlation pattern map, and reorganizes the feature points in combination with signal information to obtain a feature point correlation pattern set. The hierarchical template generation submodule rearranges the feature point set according to the feature point association pattern set, logically splices multiple feature points at the same level, and integrates the node information of each level to obtain a cardiovascular signal hierarchical analysis template.

4. The emergency cardiovascular patient early warning system according to claim 3, characterized in that, The anomaly detection module includes: The anomaly detection submodule, based on the cardiovascular signal hierarchical analysis template and combined with the hierarchical labels of signal nodes, compares and sorts the nodes at each level according to their hierarchical label priority values, using the following formula: ； The priority evaluation value of the compute nodes is used to independently determine the anomalies of the nodes in order from top to bottom, and the anomaly determination index value is obtained. Where P represents the priority evaluation value of the node. The weight representing the label at level i of node i. The signal strength representing node i, The highest level label priority of the node is represented by n, where n represents the total number of nodes. The path recording submodule identifies the set of node numbers in the anomaly determination path based on the anomaly determination index value, records the connection direction of each pair of adjacent nodes, and integrates the anomaly edge information by combining the node connection relationship in the path to generate anomaly determination path map data. The mapping table generation submodule collects the node numbers and adjacent node pairs in the connecting edges based on the anomaly determination path graph data, stores the upstream and downstream relationship types and connection directions between each node according to the adjacent structural relationship, and obtains the anomaly signal determination mapping table.

5. The emergency cardiovascular patient early warning system according to claim 4, characterized in that, The early warning task scheduling module includes: Based on the abnormal signal determination mapping table, the demand analysis submodule collects the processing demand parameters of each level of signal node, records the node number and demand type in the processing demand, and generates the processing demand distribution result by combining the processing demand type. The performance evaluation submodule analyzes the performance parameters and load capacity of the processing unit based on the processing demand distribution results, extracts the processing unit's computing speed, response time, and available resource information, and generates a list of processing unit performance parameters. The processing allocation submodule allocates signal nodes to suitable processing units according to the processing unit performance parameter list, records the correspondence between processing unit numbers and signal node numbers, integrates processing allocation information, and generates a signal processing allocation scheme.

6. The emergency cardiovascular patient early warning system according to claim 5, characterized in that, The signal verification module includes: Based on the signal processing allocation scheme, the verification task scheduling submodule collects the signal node number information in the processing unit, extracts the processing position information of each signal node in sequence and establishes a signal node mapping set, calculates the index value of the processing position information, marks the processing unit identifier and verification cycle parameter to which the signal node belongs, and obtains the signal node number value. The verification execution submodule performs verification operations on the signal nodes in the processing unit according to the signal node number value, extracts the abnormal node numbers from the verification results, records the verification timestamp, compares it with the verification benchmark value, filters out abnormal node combinations that meet the conditions, and establishes an abnormal node number set. The log generation submodule queries the original signal node identifier based on the set of abnormal node numbers and the combination of abnormal nodes corresponding to the numbers, integrates the signal node identifier and the verification timestamp information, and generates a list of cardiovascular signal abnormality warning records.

7. The emergency cardiovascular patient early warning system according to claim 6, characterized in that, The extraction of abnormal node numbers from the verification results refers to verifying the signal nodes based on their signal node numbers, selecting signal node numbers that exceed a preset deviation threshold from the verification output as abnormal node numbers, and recording the abnormal node numbers and their corresponding verification timestamps in chronological order to obtain the abnormal node numbers. The process of selecting anomalous node combinations that meet the criteria involves comparing the anomalous node number data based on the verification benchmark value, aggregating the anomalous node numbers that exceed a set threshold number of anomalous nodes within the same verification timestamp, and forming anomalous node combination that includes the corresponding verification timestamp.

8. The emergency cardiovascular patient early warning system according to claim 1, characterized in that, The system also includes a recovery and backup module: The recovery and backup module, based on the cardiovascular signal abnormality early warning record list, counts the processing units to which abnormal nodes belong, assigns abnormal nodes to the corresponding processing units, analyzes the recovery correspondence between processing units and signal nodes, and generates a cardiovascular signal recovery and backup table. The cardiovascular signal recovery and backup table includes processing unit identifiers, abnormal node grouping results, and recovery attribution mapping results.

9. The emergency cardiovascular patient early warning system according to claim 8, characterized in that, The recovery and backup module includes: The unit extraction submodule collects the processing unit corresponding to each abnormal node according to the cardiovascular signal abnormality early warning record list, records the abnormal node number associated with the processing unit and the number of its corresponding signal node sets, determines the matching index between the abnormal node and the processing unit, and obtains the unit number value to which the abnormal node belongs. The node classification submodule classifies the corresponding signal nodes into each processing unit based on the abnormal node's unit number value, and extracts the node number list of each unit to obtain the node affiliation strength value of the processing unit. The structure restoration generation submodule integrates the processing unit and its subordinate signal node numbers through structural mapping based on the node affiliation strength value of the processing unit, outputs the processing unit index, the corresponding abnormal node number and the total number of nodes, determines the affiliation of the unit and the signal node, and generates a cardiovascular signal recovery and backup table.

Images