Hl7-based crrt online acquisition and intelligent monitoring system

Abstract

This invention relates to the fields of medical informatics and data communication technology, specifically a CRRT online acquisition and intelligent monitoring system based on HL7. The system includes: a protocol cleaning and state mapping step: parsing the private protocol binary stream based on a virtual device state machine model to extract the device operating status and monitoring value vectors; a differential semantic analysis step: calculating the rate of change and deviation vector of the monitoring value vector relative to the historical time window; an adaptive message encoding step: selectively generating incremental maintenance messages or full semantic messages based on the comparison results of the rate of change and semantic fluctuation threshold; and a hybrid channel transmission step: dynamically allocating transmission channels and error correction coding strategies according to the message type to execute transmission. This invention solves the transmission contradiction between the real-time nature of high-frequency vital sign data and the redundancy of standardized protocols, balancing standard compatibility and real-time alarm capabilities under limited network conditions.

Description

Technical Field

[0001] This invention relates to the field of medical informatics and data communication technology, specifically to a CRRT online acquisition and intelligent monitoring system based on HL7. Background Technology

[0002] In the current intensive care medical scenario, continuous renal replacement therapy devices generate high-frequency and proprietary vital sign monitoring data streams in real time. These data are usually in a compact binary format, and the protocol definitions of different manufacturers are significantly different, making it difficult to interact directly with the hospital information system.

[0003] To achieve data interconnection, existing solutions often convert all raw data into standard formats such as HL7 for transmission. Standard protocols suffer from severe header redundancy and character bloat, resulting in extremely high network bandwidth utilization. In environments with limited or fluctuating network resources, this high throughput requirement can easily lead to congestion and packet loss, causing delays in critical alarm signals. Furthermore, traditional unified transmission mechanisms fail to distinguish between the high reliability requirements of control signaling and the low latency requirements of waveform data, and heterogeneous protocol parsing often faces a black box state, resulting in data timing disorder, rigid processing flow, and poor cross-system compatibility, failing to meet the millisecond-level response requirements of clinical emergency situations.

[0004] Therefore, how to resolve the contradiction between the real-time performance and bandwidth limitations of high-frequency data transmission while ensuring compatibility with standard protocols, and how to ensure data integrity and timing synchronization under harsh network conditions, are urgent technical problems that need to be solved. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a CRRT online data acquisition and intelligent monitoring system based on HL7. Specifically, the technical solution of this invention includes:

[0006] The protocol cleaning and state mapping module is used to obtain the private protocol binary stream of the source device and parse the private protocol binary stream based on the preset virtual device state machine model to extract the device operation state vector and monitoring value vector within the current time window.

[0007] The differential semantic analysis module is used to calculate the rate of change and deviation vector of the monitoring value vector in the current time window relative to the monitoring value vector in the historical time window.

[0008] The adaptive message encoding module is used to selectively generate incremental maintenance messages or full semantic messages in standard protocol format based on the comparison results between the rate of change generated by the differential semantic analysis module and the preset semantic fluctuation threshold.

[0009] The hybrid channel transmission module is used to dynamically allocate transmission channels and error correction coding strategies for incremental maintenance messages or full semantic messages based on the message type generated by the adaptive message encoding module, so as to perform data transmission. The adaptive message encoding module generates messages including: when the rate of change is lower than the semantic fluctuation threshold, generating an incremental maintenance message containing only the state maintenance identifier; when the rate of change exceeds the semantic fluctuation threshold, generating a full semantic message containing the complete monitoring parameter segment and attaching a high-priority quality of service mark.

[0010] Preferably, the hybrid channel transmission module is specifically used to: construct a two-layer transmission architecture including a control flow channel and a data flow channel; map non-time-sequence control data containing patient identity information and medical order configuration information to the control flow channel based on the TCP protocol for transmission; map incremental maintenance messages or full semantic messages generated by the adaptive message encoding module to the data flow channel based on the UDP protocol, and apply a forward error correction coding algorithm to the messages in the data flow channel to generate redundant check packets.

[0011] Preferably, the protocol cleaning and state mapping module is specifically used for: maintaining a virtual shadow object locally on the acquisition gateway that is synchronized with the physical state of the source device; updating the state attributes of the virtual shadow object in real time, and mapping the heterogeneous data fields in the private protocol binary stream to fields to be filled in the standard protocol format; wherein the standard protocol format is one of HL7V2, HL7V3 or FHIR standards.

[0012] Preferably, the differential semantic analysis module calculates the deviation vector by: performing a first-order difference operation on the time-series waveform data in the monitored numerical vector to obtain the instantaneous rate of change; identifying whether there is a preset alarm feature code in the monitored numerical vector; and, in response to the identification of the alarm feature code, forcibly setting the priority of the deviation vector to the highest level to trigger the generation of the full semantic message.

[0013] Preferably, the system further includes: a server-side semantic reassembly module, used to receive messages sent by the hybrid channel transmission module, and to perform vector synthesis based on incremental maintenance messages and the previous full semantic messages cached locally, so as to reconstruct the complete standard protocol message at the current moment; wherein, the server-side semantic reassembly module is also used to perform correction and recovery of data packets lost during UDP transmission using forward error correction coding algorithm and redundancy check packets.

[0014] Preferably, the system further includes: a clock synchronization and alignment module, used to embed an absolute timestamp sequence based on a unified network clock into the message header when the adaptive message encoding module generates a message; and a server-side semantic reassembly module to perform timing alignment and jitter smoothing on data streams from different source devices based on the absolute timestamp sequence.

[0015] Preferably, the system further includes: a congestion awareness and threshold adjustment module, used to monitor the packet loss rate and round-trip latency of the data flow channel; and dynamically adjust the semantic fluctuation threshold in the adaptive message encoding module based on the mapping relationship between the packet loss rate and the preset network congestion model; wherein, dynamically adjusting the semantic fluctuation threshold includes: increasing the semantic fluctuation threshold when the packet loss rate increases to reduce the sending frequency of full semantic messages, thereby reducing bandwidth usage.

[0016] Preferably, after the server-side semantic reconstructing module reconstructs the complete standard protocol message, it is also used to: distribute the reconstructed complete standard protocol message to the hospital information system or electronic medical record system to achieve cross-system medical data interconnection and persistent storage.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. This invention, through a protocol cleaning and state mapping module, maintains a virtual shadow object locally on the acquisition gateway, synchronized with the physical state of the source device. Utilizing a two-level mapping mechanism of protocol-shadow-semantic mapping tables, it eliminates the black-box state of heterogeneous protocol parsing. By real-time parsing and updating the attributes of the virtual shadow object with the binary stream of the private protocol, and then constructing standard protocol messages based on the shadow object, it decouples the binary parsing and standard protocol construction processes, ensuring a logical closed loop for data flow. This achieves accurate mapping of private protocol data to standard formats such as HL7V2, HL7V3, or FHIR, solving the problem of traditional private protocol modes struggling to achieve cross-system interconnection.

[0019] 2. This invention achieves intelligent traffic shaping through the collaborative work of a differential semantic analysis module and an adaptive message encoding module. By calculating the weighted Euclidean distance of the monitored numerical vectors to obtain the instantaneous semantic change rate, and combining it with the bitmask matching mechanism of the alarm feature code, the system can send only incremental maintenance messages containing the status maintenance identifier during the stable period of the patient's condition, which greatly saves network bandwidth. When data mutation occurs or critical alarm features are identified, it can forcibly trigger the transmission of high-priority full semantic messages. This mechanism effectively solves the transmission contradiction between the real-time nature of high-frequency vital sign data and the redundancy of standardized protocols, ensuring that bandwidth saving and real-time clinical alarms are balanced in a limited network environment.

[0020] 3. This invention constructs a two-layer transmission architecture that separates control flow and data flow through a hybrid channel transmission module, and introduces a forward error correction coding algorithm based on Cauchy matrix. By mapping non-time-sequential control data with high reliability requirements to the TCP channel and mapping waveform data that is sensitive to delay to the UDP channel with additional redundant check packets, the mathematical property of maximum distance divisible code is utilized, so that the receiving end can recover lost data through matrix operations without requesting retransmission when the network packet loss rate is high. This design avoids the transmission of large amounts of waveform data from blocking the issuance of critical medical orders, ensures deterministic low-latency transmission of CRRT alarm signals, and improves the stability of the monitoring system.

[0021] 4. This invention endows the system with adaptive capabilities and timing accuracy to the network environment through congestion perception and threshold adjustment modules and clock synchronization and alignment modules. By monitoring packet loss rate and round-trip delay in real time and dynamically adjusting semantic fluctuation thresholds, the system can automatically reduce the frequency of full packet transmission to reduce bandwidth consumption and ensure the delivery of critical alarms when the network is congested. At the same time, by embedding an absolute timestamp sequence based on a unified network clock in the packet header and sorting and aligning it on the server side using a jitter smoothing buffer, the data timing disorder caused by network transmission delay jitter is eliminated, providing an accurate timing basis for subsequent cross-device and cross-system clinical big data analysis. Attached Figure Description

[0022] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0023] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0025] Example 1:

[0026] Please see Figure 1 The CRRT online acquisition and intelligent monitoring system based on HL7 includes: a protocol cleaning and state mapping module, which is used to acquire the private protocol binary stream of the source device and parse the private protocol binary stream based on the preset virtual device state machine model to extract the device operating state vector and monitoring value vector within the current time window; and a differential semantic analysis module, which is used to calculate the rate of change and deviation vector of the monitoring value vector within the current time window relative to the monitoring value vector of the historical time window.

[0027] The adaptive message encoding module is used to selectively generate incremental maintenance messages or full semantic messages in standard protocol format based on the comparison results between the rate of change generated by the differential semantic analysis module and the preset semantic fluctuation threshold.

[0028] The hybrid channel transmission module is used to dynamically allocate transmission channels and error correction coding strategies for incremental maintenance messages or full semantic messages based on the message type generated by the adaptive message encoding module, so as to perform data transmission. The adaptive message encoding module generates messages including: when the rate of change is lower than the semantic fluctuation threshold, generating an incremental maintenance message containing only the state maintenance identifier; when the rate of change exceeds the semantic fluctuation threshold, generating a full semantic message containing the complete monitoring parameter segment and attaching a high-priority quality of service mark.

[0029] This embodiment describes in detail the core processing flow of the system. The system is mainly deployed between the bedside acquisition gateway and the central server, aiming to solve the transmission contradiction between the real-time nature of high-frequency vital sign data and the redundancy of standardized protocols. The protocol cleaning and state mapping module reads the private protocol binary stream of the CRRT source device in real time through the serial port or network interface, and uses the virtual device state machine model running in the gateway memory. This model has a pre-set protocol feature mapping table containing the frame header and frame tail definitions, and a data field definition table containing byte offset, data type, and scaling factor.

[0030] To implement the parsing function, the virtual device state machine model defines a finite set of states, including SYNC_SEARCH, HEADER_LATCH, PAYLOAD_READ, and VALIDATION. The module is initially in the SYNC_SEARCH state, scanning the binary stream byte by byte until it matches the frame header feature word in the mapping table, and then transitions to the HEADER_LATCH state.

[0031] After reading the length of bytes defined by the protocol, the state machine transitions to the PAYLOAD_READ state to extract the byte sequence at the specified offset and applies a scaling factor for physical quantity conversion. It then enters the VALIDATION state to perform CRC or checksum comparison. If the check passes, the parsing result is output; otherwise, the data is discarded and the state is reset to SYNC_SEARCH. This parses the binary stream into a device operating state vector S containing the device's operating mode and a monitoring value vector V containing sensor sampling data. To accurately quantify the fluctuation of the data and eliminate numerical scale bias caused by differences in dimensions between different monitoring parameters, such as venous pressure (mmHg) and flow rate (ml / min), the differential semantic analysis module calculates the instantaneous semantic change rate using the square of the weighted Euclidean distance, as shown in the following formula:

[0032] in, These represent the monitoring value vectors for the current time window and the previous time window, respectively. This vector consists of continuously measured sub-vectors. With discrete state subvectors It is pieced together; for Each element in has been subjected to max-min standardization, and The original unsigned integer format is preserved without normalization to retain bitmask characteristics;

[0033] for Each floating-point element in All have been normalized to the minimum and mapped to the dimensionless interval [0, 1], and their calculation formula is:

[0034] in, To prevent the minimum value of division by zero, such as , and The device physical parameter specification table is pre-loaded into the gateway memory. This table clearly defines the theoretical physical range of each monitoring parameter, such as setting the venous pressure range as [-50, 300] mmHg. The source is a pre-defined diagonal weight matrix. To eliminate the subjectivity of manual setting and ensure clinical safety, its diagonal elements... Automatic generation of alarm priorities based on parameters:

[0035] The system maps each monitoring parameter to a priority index according to the IEC60601-1-8 standard. These correspond to low, medium, and high priorities, respectively, and their weights are calculated using the Softmax function. The calculation formula is as follows:

[0036] in, The traversal index for the summation operation has a range of values. to ; To monitor the total number of dimensions of the numerical vector, i.e., the total number of monitoring parameters supported by the system; thereby ensuring that high-risk parameters, such as venous pressure, have a significantly higher weight than secondary parameters, It can keenly reflect clinical critical conditions;

[0037] The source is the system clock, and its physical meaning is the time interval between two time windows, measured in seconds. The calculation also incorporates... For example, 0.001s, to avoid division by zero overflow caused by high-frequency sampling;

[0038] Meanwhile, the deviation vector is specifically defined as This is used to preserve the original differences in each dimension for subsequent analysis; based on this, the adaptive message encoding module uses the calculated... With the preset semantic fluctuation threshold Make a judgment;

[0039] To ensure It can adapt to the sensor noise floor levels of different devices. This threshold is not a fixed constant, but is obtained by initiating a self-calibration process: the system performs a self-calibration process before the device starts up. For example, if the system is in learning mode for 30 seconds, it only collects data and does not send alarms, calculating the time period within that period. Statistical mean and standard deviation and set This allows for the construction of a baseline threshold covering 99.7% of random noise; in response to The system generates an incremental maintenance message containing only a status maintenance identifier and a timestamp; in response to The system generates a full semantic message containing complete parameter values ​​at the current moment and marks it as high priority.

[0040] The hybrid channel transmission module performs transmission tasks according to message type. By performing semantic cleaning and differential coding at the edge, the system achieves traffic shaping. During the stable period of CRRT treatment, which lasts for tens of hours, only byte-level heartbeat packets are sent, which greatly saves network bandwidth. At the moment of sudden data change or alarm or change in condition, the system can use the idle bandwidth resources to send complete standard HL7 data with millisecond-level low latency. Thus, in a limited network environment, it can simultaneously ensure standard compatibility and real-time alarm capability.

[0041] Example 2:

[0042] The hybrid channel transmission module is specifically used to: construct a two-layer transmission architecture that includes a control flow channel and a data flow channel; map non-time-sequence control data containing patient identity information and medical order configuration information to the TCP-based control flow channel for transmission; map incremental maintenance messages or full semantic messages generated by the adaptive message encoding module to the UDP-based data flow channel, and apply forward error correction coding algorithm to the messages in the data flow channel to generate redundant check packets.

[0043] This embodiment further elaborates on the underlying dual-layer transmission architecture of the hybrid channel transmission module, aiming to resolve the contradiction between the high reliability requirements of control signaling and the low latency requirements of waveform data in medical data. The system constructs a control flow channel. In response to the non-time-sequence control data such as patient identity information and medical order configuration information to be transmitted, the system maps it to a TCP-based channel and uses TCP's handshake and retransmission mechanism to ensure the integrity and order of the data.

[0044] Simultaneously, the system constructs a data flow channel. In response to whether the data to be transmitted is an incremental maintenance message or a full semantic message, the system maps it to a channel based on the UDP protocol. During this process, after sending N raw data packets, the module applies a forward error correction coding algorithm, specifically the Reed-Solomon erasure coding algorithm based on the Vandermonde matrix, to generate K redundant check packets and send them together.

[0045] The values ​​of parameters N and K are determined by a preset redundancy strategy. For example, setting N=10 and K=4 means that 4 check packets are added for every 10 original data packets sent, allowing any 4 packets to be lost during transmission without affecting data recovery. To ensure the feasibility of the algorithm, the erasure coding algorithm operates in a finite field. The above operations are performed using primitive polynomials, for example... Construct a multiplication table and a logarithm table.

[0046] To clarify the construction method of the encoding matrix and ensure that the maximum distance separable (MDS) property is satisfied to guarantee the invertibility of any submatrix, the encoding matrix... Specifically constructed as a Cauchy matrix: defining two in disjoint sets and For example, take for , for Each element in the matrix The calculation formula is:

[0047] in, This represents the row index of the encoding matrix, corresponding to the set of redundant packets. The element index in the text; The column index of the encoding matrix corresponds to the original data packet set. The element index in the text; For addition over a field, i.e., the XOR operation, this structure mathematically guarantees that all subarrays are nonsingular; during the encoding phase, the module performs byte slicing operations: traversal Each raw data packet is analyzed, and its maximum byte length is extracted and defined as follows: And perform zero-padding alignment on short packets. For each byte offset, extract N data packets. The bytes at each location constitute the information vector. and utilize pre-generated Encoding matrix The vector is multiplied on the left to calculate the byte values ​​of the K redundant packets at the corresponding positions in parallel, and finally concatenated to generate a redundancy check packet;

[0048] By separating the control flow and data flow, this system avoids the transmission of large amounts of waveform data from blocking the issuance of critical medical orders. In particular, the introduction of UDP in conjunction with the FEC mechanism in the data flow channel enables the receiving end to recover lost data through redundant packets without requesting retransmission when the network packet loss rate is high, thereby ensuring the deterministic low latency of CRRT alarm signal transmission.

[0049] Example 3:

[0050] The protocol cleaning and state mapping module is specifically used for: maintaining virtual shadow objects that are synchronized with the physical state of the source device on the local acquisition gateway; updating the state attributes of the virtual shadow objects in real time; and mapping heterogeneous data fields in the private protocol binary stream to fields to be filled in the standard protocol format; wherein the standard protocol format is one of HL7V2, HL7V3 or FHIR standards.

[0051] This embodiment refines the execution logic of the protocol cleaning and state mapping module, aiming to eliminate the black-box state of heterogeneous protocol parsing. The module constructs and maintains a virtual shadow object in the memory of the acquisition gateway that is completely synchronized with the physical state of the source CRRT device. This object exists as a class instance with attribute KV pairs. When the binary stream of the private protocol arrives, the module does not directly perform string concatenation, but uses a two-level mapping mechanism to update the attribute values ​​of the virtual shadow object in real time. The two-level mapping mechanism performs operations based on a preset protocol-shadow-semantic mapping table.

[0052] To ensure compatibility between continuous numerical values ​​and discrete states and to guarantee the traceability of the mapping path, the mapping table supports an expanded seven-tuple structure: for continuous numerical fields, such as pressure and flow rate, the structure {Byte_Offset, Bit_Mask, Bit_Shift, Linear_Coeff, Linear_Intercept, Shadow_Attr_Key, HL7_Target_Path} is adopted.

[0053] The specific analysis steps are as follows: Physical quantity extraction and shadow update: The module locates the target byte in the binary stream based on Byte_Offset, performs bitwise operations (RawData&Bit_Mask) >> Bit_Shift to extract the original payload, and substitutes the payload into the linear equation:

[0054] Restore the physical quantity and immediately call the update method ShadowObj.Update(Shadow_Attr_Key, Value_{phy}) of the virtual shadow object, for example, store the calculated venous pressure in the attribute with the key name Venous_Pressure;

[0055] Standard semantic mapping: When a message needs to be generated, the module traverses the mapping table, reads the current value of the corresponding Shadow_Attr_Key from the virtual shadow object according to the guidance of HL7_Target_Path, and fills it into the corresponding leaf node of the standard protocol tree;

[0056] For discrete status fields, such as device operating mode and alarm code, the structure {Byte_Offset, Bit_Mask, Enum_Lookup_Table, Shadow_Attr_Key, HL7_Target_Path} is adopted. Enum_Lookup_Table is a preset hash dictionary, such as {0x01: "CVVH", 0x02: "HVHF"}. After the module extracts the payload, it performs key-value matching, updates the obtained standard semantic string to the shadow object, and finally maps it to the observation result segment of HL7V2, the XML of HL7V3, the structure element, or the JSON resource of FHIR. By introducing Shadow_Attr_Key as an intermediate anchor point, the system decouples the binary parsing and HL7 construction process, ensuring the logical closed loop of data flow.

[0057] Example 4:

[0058] The differential semantic analysis module calculates the deviation vector, including: performing a first-order difference operation on the time-series waveform data in the monitored numerical vector to obtain the instantaneous rate of change; identifying whether there is a preset alarm feature code in the monitored numerical vector; and in response to the identification of the alarm feature code, forcibly setting the priority of the deviation vector to the highest level to trigger the generation of the full semantic message.

[0059] This embodiment details the process of calculating the deviation vector in the differential semantic analysis module and clarifies the internal data structure of the vector to avoid type confusion; it also monitors the numerical vector. Logically divided into continuous measurement sub-vectors With discrete state subvectors ; Module pair For time-series waveform data, such as blood pressure and flow velocity, perform a first-order difference operation, and the calculation formula is as follows: To obtain the instantaneous rate of change;

[0060] Meanwhile, in order to identify alarm signature codes, the module... A bitmask matching operation is performed on a specific dimension. The specific steps are as follows: traverse the preset alarm feature mask table, which contains multiple sets of... For each entry, the module calculates:

[0061] like If so, it is determined that an alarm feature code has been identified, such as a high transmembrane pressure alarm or an air monitoring alarm.

[0062] Based on this, in response to the identification of the alarm signature code, regardless of the current situation... If the rate of change of the numerical value exceeds the threshold, the system executes a priority reversal strategy, forcibly reversing the deviation vector. The priority flag is set to the highest level; this mechanism ensures clinical safety, especially in the case of a sudden kink in the tubing that causes the value to drop to zero instantly or remain unchanged in a state of apparent death. A single differential calculation may fail, but the introduction of a bitmask-based alarm feature code to force the trigger condition ensures that all critical alarms can unconditionally trigger the generation and transmission of full semantic messages, eliminating the risk of missed alarms.

[0063] Example 5:

[0064] The system also includes a server-side semantic reassembly module, which receives messages sent by the hybrid channel transmission module and performs vector synthesis based on incremental maintenance messages and the previous full semantic messages cached locally to reconstruct the complete standard protocol message at the current moment; the server-side semantic reassembly module is also used to perform correction and recovery of data packets lost during UDP transmission using forward error correction coding algorithms and redundant check packets.

[0065] This embodiment supplements the server-side processing logic, especially detailing the erasure and recovery algorithm based on RS codes to meet reproducibility requirements; the module receives messages sent by the hybrid channel transmission module; in response to receiving a full semantic message, the module directly parses and updates the local current state cache; in response to receiving an incremental maintenance message, the module reads the previous full semantic message in the local cache, and combines it with the timestamp in the incremental message to perform a vector synthesis operation, that is, adopting the zero-order hold-all (ZOH) strategy, directly inheriting the numerical attributes of the previous full semantic message, only updating the timestamp attribute, and reconstructing the complete state at the current moment;

[0066] During this period, to address potential packet loss during UDP transmission, the module performs deterministic algebraic recovery using the received redundant check packets. The specific steps are as follows: Receive set construction: Assuming the sender uses... The strategy involves N original packets and K redundant packets. The set of valid packets collected by the receiver within the timeout window is: ,like Then it cannot be recovered, if Then select the first N packets to form the recovery vector. And record the index set corresponding to these N packages. The original package index is Redundant package index is ;

[0067] Decoding matrix construction: in a finite field Above the construction Decoding matrix ;for The Line, corresponding :like If the received packet is the original packet, then this line represents the identity matrix. The Okay; if If the received packet is redundant, then this line of code is the encoding matrix. That is, the Cauchy matrix, the corresponding row, i.e., the first row. OK;

[0068] Inverse matrix solution and recovery: using Gauss-Jordan elimination method in Above calculation inverse matrix And perform matrix multiplication. This allows for the simultaneous calculation of the original N data packets. ;

[0069] This module utilizes the mathematical properties of maximum distance separable (MDS) codes, enabling the receiver to recover the full data through matrix inversion operations once it receives any N packets, thus achieving highly reliable transmission over unreliable UDP channels.

[0070] Example 6:

[0071] The system also includes: a clock synchronization and alignment module, which is used to embed an absolute timestamp sequence based on a unified network clock into the message header when the adaptive message encoding module generates messages; and a server-side semantic reassembly module, which performs timing alignment and jitter smoothing on data streams from different source devices based on the absolute timestamp sequence.

[0072] This embodiment introduces a clock synchronization and alignment module to solve the clock inconsistency problem in the distributed acquisition system. At the moment the adaptive message encoding module generates a message, the module inserts an absolute timestamp sequence based on the unified network clock of NTP or PTP into the message header.

[0073] When processing messages from different beds and devices, the server-side semantic reconstruction module does not rely on arrival time but strictly sorts and aligns messages based on the absolute timestamp sequence within the messages. Simultaneously, the module utilizes a jitter smoothing buffer to eliminate waveform distortion caused by network transmission latency jitter. Specifically, the jitter smoothing buffer employs a circular queue structure and sets a preset playback delay threshold. For example, 200ms, for arrival times of And carries a timestamp The module temporarily stores the message in a queue until the local system time. Only then will the message be popped up and submitted to the upper layer for processing, thereby absorbing random latency fluctuations during network transmission;

[0074] This mechanism ensures strict synchronization of multi-source heterogeneous data on the timeline, eliminates data timing disorder caused by network latency differences, and provides an accurate timing basis for subsequent analysis of clinical big data such as the relationship between blood pressure drop and CRRT ultrafiltration rate at the same time.

[0075] The system also includes: a congestion awareness and threshold adjustment module, used to monitor the packet loss rate and round-trip latency of the data flow channel; and dynamically adjusts the semantic fluctuation threshold in the adaptive message encoding module based on the mapping relationship between the packet loss rate and the preset network congestion model. The dynamic adjustment of the semantic fluctuation threshold includes: increasing the semantic fluctuation threshold when the packet loss rate increases to reduce the sending frequency of full semantic messages, thereby reducing bandwidth usage.

[0076] This embodiment introduces a congestion awareness and threshold adjustment module, giving the system the ability to adapt to the network environment; the module monitors the packet loss rate and round-trip latency of the data flow channel in real time; based on the mapping relationship between packet loss rate, round-trip latency and a preset network congestion model, it calculates the dynamic semantic fluctuation threshold, as shown in the following formula:

[0077] in, This is the dynamically adjusted semantic fluctuation threshold; The source is a preset value, and its physical meaning is a basic semantic fluctuation threshold; to ensure the objectivity of its value, this basic threshold... This is not arbitrarily assigned, but rather derived from the startup self-calibration process described in Example 1 above: that is, in the learning mode during the initial startup of the device, the system calculates the semantic change rate. Statistical mean and standard deviation and set This serves as a benchmark reference when the network conditions are good.

[0078] The source is a preset constant, and its physical meaning is a weighted adjustment coefficient, for example, setting... This indicates that the system is more sensitive to packet loss than to latency;

[0079] The formula for calculating the packet loss rate factor is as follows: In order to be able to collect statistics The system encapsulates a 32-bit monotonic sequence number (SequenceID) in the header of each UDP packet. The receiving end counts the number of valid received packets by detecting discontinuities in the sequence numbers and transmits the count via a TCP control channel using a structure. Periodic feedback in the form of;

[0080] To eliminate parsing errors caused by byte order differences across different hardware platforms, this structure is strictly defined at the network transport layer as a big-endian binary stream: the first 4 bytes are stored... That is, an unsigned 32-bit integer, with the last 4 bytes stored. That is, an unsigned 32-bit integer with a fixed total length of 8 bytes and no alignment padding bytes. The sending end converts the structure data into a big-endian network byte stream for encapsulation, and the receiving end restores the network byte stream to host byte order for parsing after receiving it.

[0081] The sending end performs precise statistics by combining local sending logs and feedback data: The sending end maintains a timestamp index log that records the sequence IDs of sent messages. In response to receiving a feedback structure, the sending end records the local system time when the current feedback is received and extracts the maximum sequence number that has been sent locally at that moment. The local maximum sequence number sent when the last feedback was received The difference between the two is calculated as the actual total number of transmissions. ,Right now ;

[0082] Simultaneously calculate the two feedbacks The difference is used as This allows for the accurate calculation of the packet loss rate within the feedback cycle. And when season ; The formula for calculating the latency degradation factor is as follows:

[0083] in, The average round-trip time currently monitored. This serves as the baseline latency when the network is idle, such as 10ms; to adapt to different network environments, During the handshake phase of system connection establishment, the following data is dynamically acquired: the system sends a set of packets, for example 20, ICMP probe packets or application layer heartbeat packets, and the minimum round-trip time is calculated as the value for this session. ;

[0084] To clarify the interaction interfaces and calling relationships between modules, the congestion awareness and threshold adjustment module and the adaptive packet encoding module interact decoupled through a configuration center based on shared memory; the congestion awareness and threshold adjustment module, acting as a producer, calculates and outputs data at preset intervals, such as 1 second. The value is written to a shared variable protected by a read-write lock; the adaptive message encoding module, acting as a consumer, reads the current value of this shared variable before each message generation as the threshold for this judgment; if the shared variable has not yet been initialized, a preset basic semantic fluctuation threshold is used. Make a judgment; based on this, respond to the packet loss rate. or delay The system automatically increases the height. ;

[0085] This means that only data with larger fluctuations can trigger the sending of full semantic messages, while minor fluctuations will be filtered out. This application-layer proactive congestion control mechanism significantly reduces the sending frequency of full messages by sacrificing the accuracy of minor data, thereby significantly reducing bandwidth usage and ensuring that the most critical alarm information and significant changes in the condition can still be delivered in harsh network environments, thus maximizing system resilience.

[0086] After the server-side semantic reconstructing module reconstructs the complete standard protocol message, it is also used to distribute the reconstructed complete standard protocol message to the hospital information system or electronic medical record system to achieve cross-system medical data interconnection and persistent storage.

[0087] This embodiment describes the final distribution stage of data processing; the server-side semantic reconstructing module completes data reconstruction, error correction, and time sequence alignment; the module distributes the generated standard HL7 messages to the hospital's internal Hospital Information System (HIS) or Electronic Medical Record System (EMR) through a message queue or Web service interface; this process realizes cross-system medical data interconnection and persistently stores high-value data during CRRT treatment, breaking down information silos in CRRT equipment and achieving a full-link digital closed loop from bedside equipment to the hospital's core database, effectively supporting subsequent medical quality control and scientific research analysis.

[0088] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An HL7-based CRRT online acquisition and intelligent monitoring system, characterized in that, include: The protocol cleaning and state mapping module is used to obtain the private protocol binary stream of the source device and parse the private protocol binary stream based on the preset virtual device state machine model to extract the device operation state vector and monitoring value vector within the current time window. The differential semantic analysis module is used to calculate the rate of change and deviation vector of the monitoring value vector in the current time window relative to the monitoring value vector in the historical time window. The adaptive message encoding module is used to selectively generate incremental maintenance messages or full semantic messages in standard protocol format based on the comparison results between the rate of change generated by the differential semantic analysis module and the preset semantic fluctuation threshold. The hybrid channel transmission module is used to dynamically allocate transmission channels and error correction coding strategies for incremental maintenance messages or full semantic messages based on the message type generated by the adaptive message encoding module, so as to perform data transmission; wherein, the adaptive message encoding module generates messages including: when the rate of change is lower than the semantic fluctuation threshold, generating an incremental maintenance message containing only a state maintenance identifier; when the rate of change exceeds the semantic fluctuation threshold, generating a full semantic message containing complete monitoring parameter segments and attaching a high-priority quality of service flag.

2. The HL7 based CRRT online collection and intelligent monitoring system according to claim 1, wherein, The hybrid channel transmission module is specifically used to: construct a two-layer transmission architecture including a control flow channel and a data flow channel; map non-time-series control data containing patient identity information and medical order configuration information to the control flow channel based on the TCP protocol for transmission; map incremental maintenance messages or full semantic messages generated by the adaptive message encoding module to the data flow channel based on the UDP protocol, and apply a forward error correction encoding algorithm to the messages in the data flow channel to generate redundant check packets.

3. The HL7 based CRRT online collection and intelligent monitoring system according to claim 1, wherein, The protocol cleaning and state mapping module is specifically used for: maintaining a virtual shadow object locally on the acquisition gateway that is synchronized with the physical state of the source device; updating the state attributes of the virtual shadow object in real time, and mapping the heterogeneous data fields in the private protocol binary stream to fields to be filled in the standard protocol format; wherein, the standard protocol format is one of HL7V2, HL7V3 or FHIR standards.

4. The HL7 based CRRT online collection and intelligent monitoring system according to claim 1, wherein, The differential semantic analysis module calculates the deviation vector, including: performing a first-order difference operation on the time-series waveform data in the monitored numerical vector to obtain the instantaneous rate of change; identifying whether there is a preset alarm feature code in the monitored numerical vector; and in response to the identification of the alarm feature code, forcibly setting the priority of the deviation vector to the highest level to trigger the generation of the full semantic message.

5. The HL7 based CRRT online collection and intelligent monitoring system according to claim 1, wherein, Also includes: The server-side semantic reassembly module is used to receive messages sent by the hybrid channel transmission module and perform vector synthesis based on the incremental maintenance message and the previous full semantic message cached locally to reconstruct the complete standard protocol message at the current moment; wherein, the server-side semantic reassembly module is also used to perform correction and recovery of data packets lost during UDP transmission using forward error correction coding algorithm and redundancy check packets.

6. The CRRT online data acquisition and intelligent monitoring system based on HL7 according to claim 5, characterized in that, Also includes: The clock synchronization and alignment module is used to embed an absolute timestamp sequence based on a unified network clock into the message header when the adaptive message encoding module generates a message; The server-side semantic reassembly module performs time-series alignment and jitter smoothing on data streams from different source devices based on the absolute timestamp sequence.

7. The HL7-based CRRT online data acquisition and intelligent monitoring system according to claim 2, characterized in that, Also includes: The congestion awareness and threshold adjustment module is used to monitor the packet loss rate and round-trip latency of the data stream channel. Based on the mapping relationship between packet loss rate and preset network congestion model, the semantic fluctuation threshold in the adaptive message encoding module is dynamically adjusted; wherein, the dynamic adjustment of the semantic fluctuation threshold includes: when the packet loss rate increases, increasing the semantic fluctuation threshold to reduce the sending frequency of full semantic messages, thereby reducing bandwidth usage.

8. The CRRT online data acquisition and intelligent monitoring system based on HL7 according to claim 5, characterized in that, After the server-side semantic reconstructing module reconstructs the complete standard protocol message, it is also used to: distribute the reconstructed complete standard protocol message to the hospital information system or electronic medical record system to achieve cross-system medical data interconnection and persistent storage.

Images