A medical data accurate routing and closed loop transmission method based on patient risk state

Abstract

The application relates to the technical field of medical Internet of Things data transmission, and discloses a medical data accurate routing and closed-loop transmission method based on a patient risk state. The method collects patient-side multi-source medical data and network state data, forms a medical semantic unit, and determines a transmission control risk level and a medical emergency degree; the medical semantic unit is recombined into a core risk package and an extended evidence package which are bound to each other; candidate paths and transmission plan combinations are subjected to safety level, encryption capability, bearing capacity and dependency relationship hard constraint screening, and path selection is carried out in the feasible path with emergency degree modulation; queue scheduling and resource allocation are carried out according to a basic path cost, a waiting time and a core risk package mark; and the proportion of real-time uploading of the extended evidence is adjusted according to the time delay, the packet loss rate and the sending success rate of the core risk package. The method can improve the timeliness, safety and stability of key medical data transmission.

Description

Technical Field

[0001] This invention belongs to the field of medical Internet of Things (IoT) data transmission technology, specifically relating to a method for precise routing and closed-loop transmission of medical data based on patient risk status. Background Technology

[0002] With the widespread application of wearable monitoring devices, home healthcare gateways, hospital edge servers, and medical information platforms, physiological monitoring data, behavioral event data, device status data, and continuous waveform data generated on the patient side need to be continuously transmitted to the hospital via wireless LAN, cellular network, or medical IoT. Different medical data vary in terms of data volume, timeliness, privacy level, and integrity requirements. Key status fields typically have smaller data volumes and require timely delivery, while their corresponding continuous waveforms, raw monitoring segments, and historical comparison data have larger data volumes. When network congestion or bandwidth limitations occur, using the same transmission method for all medical data can easily consume real-time transmission resources, increasing queuing delays and packet loss risks for critical data.

[0003] Patent CN119155321A discloses an intelligent medical data management system. This system adjusts the transmission order based on the real-time priority of medical data and calculates a comprehensive transmission score based on transmission path latency, packet loss rate, and bandwidth utilization to select the appropriate transmission path. Simultaneously, it adjusts relevant transmission parameters based on transmission latency fluctuations. This technology can improve the timeliness of medical data transmission to a certain extent.

[0004] However, the aforementioned existing technologies primarily treat various diagnostic and treatment data or transmission tasks as independent scheduling objects, failing to establish a hierarchical relationship and transmission order constraints between key informational data and large-volume extended evidence data within the same monitoring event. Extended evidence data may still compete with key informational data for real-time bandwidth. Existing path selection methods typically employ a comprehensive scoring approach to handle indicators such as latency, packet loss rate, and bandwidth. When security level, encryption capability, minimum bandwidth, or trusted caching capability are also considered as compensable scoring items, a path that does not fully meet the security or carrying requirements of medical data may be selected.

[0005] Furthermore, if the path cost is modulated based on the urgency of data transmission during the path selection phase, and then directly used for queue scheduling, the impact of urgency can easily be superimposed at different control levels. Existing technologies also lack a closed-loop control mechanism that adjusts the real-time upload ratio of extended evidence data based on the actual transmission delay, packet loss rate, and transmission success rate of critical data.

[0006] Therefore, it is necessary to provide a precise routing and closed-loop transmission method for medical data, which involves dependency-based packetization of key information data and extended evidence data, setting security level, encryption capability, minimum carrying capacity, trusted caching capability, and sending order as hard constraints for path filtering, and dynamically adjusting the usage of extended evidence data on real-time transmission resources based on the actual transmission effect of key data, so as to improve the timeliness, security, and stability of key medical data transmission. Summary of the Invention

[0007] This invention addresses the problems that arise when medical data of varying importance and volume are transmitted using the same method. These problems include critical medical data being overwhelmed by large-capacity extended evidence data, transmission delays in critical data, conflicts between security requirements and real-time optimization, and a lack of dynamic adjustment of subsequent resource allocation based on the actual transmission performance of critical data.

[0008] To achieve the above technical objectives, the technical solution adopted by this invention is: 1. A method for precise routing and closed-loop transmission of medical data based on patient risk status, characterized by comprising the following steps:

[0009] S1 collects multi-source medical data and network status data from the patient side, forms candidate medical semantic units based on the medical semantic transmission rules configured at the hospital, and determines the transmission control risk level and medical urgency.

[0010] S2, the candidate medical semantic units are reorganized into mutually bound core risk packages and extended evidence packages, and a dependency relationship is established between the extended evidence packages and the core risk packages. The reorganized core risk packages and extended evidence packages are collectively referred to as medical semantic transmission units.

[0011] S3. Based on the medical IoT network topology, candidate paths and corresponding transmission plan combinations are generated, and hard constraints are applied to filter them according to security level, encryption capability, minimum carrying capacity and the aforementioned dependency relationship.

[0012] S4, among the combinations filtered by hard constraints, modulates the delay, queuing, packet loss and congestion costs according to the medical urgency to select the target path;

[0013] S5 performs queue scheduling and resource allocation based on the unmodulated basic cost, waiting time and core risk packet marking of the target path, so that the core risk packet obtains transmission resources no less than the preset lower limit, and sends the extended evidence packet after the dependency conditions are met.

[0014] S6 records the transmission delay, queuing delay, packet loss rate, and transmission success rate of core risk packets, and adjusts the proportion of real-time upload of extended evidence and transmission control parameters for the next closed-loop cycle accordingly.

[0015] The beneficial effects of this invention are as follows:

[0016] 1. This invention divides the same medical semantic unit into a core risk package and an extended evidence package that are bound to each other, and limits the extended evidence package to be transmitted under the core risk package. This can reduce the real-time bandwidth occupation of large amounts of continuous waveforms, historical fragments or original records while preserving the ability to supplement subsequent evidence, thereby shortening the queuing time and transmission delay of the core risk package.

[0017] 2. Before calculating the path cost, this invention performs hard constraint screening on candidate paths and transmission plan combinations based on path security level, encryption capability, minimum bandwidth, trusted caching capability, and data packet dependency. This ensures that the medical urgency is adjusted only within feasible paths that meet security and carrying capacity conditions, thus avoiding the selection of paths with insufficient security or transmission capacity in pursuit of low latency, thereby improving the security and reliability of medical data transmission.

[0018] 3. In the queue scheduling stage, this invention comprehensively considers basic transmission priority, medical urgency, waiting time, immediate upload flag and core risk packet flag, and sets a minimum transmission resource ratio for core risk packets. This can prioritize the transmission of core risk packets in the event of congestion or limited resources. At the same time, waiting time compensation reduces the long-term retention of ordinary medical data and queue starvation, and improves the rationality and stability of transmission resource allocation.

[0019] 4. Based on the actual end-to-end latency, queuing latency, packet loss rate, and transmission success rate of the core risk package, this invention provides feedback adjustment to the real-time upload ratio of extended evidence and related transmission control parameters in the next closed-loop cycle. When the transmission service quality of the core risk package deteriorates, it can proactively reduce the occupation of real-time resources by extended evidence data, thereby forming a closed-loop control constrained by the core data transmission effect, improving the timeliness and adaptability of continuous transmission of critical medical data under complex network conditions. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention;

[0021] Figure 2 A comparison chart of average end-to-end latency for high-risk core data under different network congestion conditions;

[0022] Figure 3 A comparison chart of on-time delivery rates of core risk packets under different network congestion conditions;

[0023] Figure 4 A comparison chart of average queuing latency for core risk packages under mixed business loads;

[0024] Figure 5 A comparison chart showing the improvement in overall transmission performance due to closed-loop updates. Detailed Implementation

[0025] Without departing from the concept of this invention, any equivalent adjustments made by those skilled in the art to the data source, sensor model, sampling frequency, prediction window length, kernel function type, model parameters, and threshold settings can be applied to the implementation scenarios of this invention.

[0026] Reference Figures 1-5 As can be seen, this invention discloses a method for precise routing and closed-loop transmission of medical data based on patient risk status, including the following steps:

[0027] S1 collects multi-source medical data and network status data from the patient side, forms candidate medical semantic units based on the medical semantic transmission rules configured at the hospital, and determines the transmission control risk level and medical urgency.

[0028] Step S1 is used to establish the unified input required for subsequent packet subpackaging and routing. This step sequentially completes patient data configuration, medical semantic transmission rule configuration, multi-source medical data and network status data acquisition, data preprocessing and quality assessment, dynamic time window determination, candidate medical semantic unit formation, and transmission control risk level and medical urgency calculation.

[0029] S1.1, Establish monitoring indicator collection and configuration. The hospital side targets the first... Establish data collection configuration items for monitoring indicators:

[0030] ;

[0031] In the formula, The name of the monitoring indicator; The standard sampling period; Individualized reference ranges for patients or baseline ranges configured at the hospital level; Transmission control risk boundary; For identifying the data acquisition device; To allow for sampling time deviation; Privacy level; This refers to the timeliness level. The reference range and risk boundaries are pre-configured by the hospital or updated non-diagnostically based on historical patient monitoring data.

[0032] S1.2, Establish medical semantic transmission rules. The hospital establishes the m-th type of medical semantic transmission rule based on the monitoring task type:

[0033] ;

[0034] In the formula, It is a medical semantic type; For the set of participating indicators; A set of key participation indicators; For the set of participating behavioral events; For reference observation time; Minimum effective observation time; The risk boundary is close to the upper limit of the time forecast; and These are the lower and upper limits of the dynamic time window, respectively. The level of attention for hospitals; and These represent the privacy and timeliness levels of the corresponding medical semantics. For example, glucose metabolism medical semantics can be associated with blood glucose, diet, medication, and exercise data, while blood pressure medical semantics can be associated with systolic blood pressure, diastolic blood pressure, heart rate, medication records, and symptom reports.

[0035] S1.3, Collect multi-source medical data and network status data from the patient side. The patient-side device collects physiological monitoring data, behavioral event data, and device status data within the t-th time period; the gateway, access nodes, and hospital edge server synchronously collect link latency, available bandwidth, packet loss rate, node load, node queue length, and link queue length. The system performs unified timestamps, duplicate data filtering, obvious error value removal, data format standardization, and data source marking on the data. For short-term missing data, local interpolation or the most recent valid value is only performed when the timestamp deviation and device status meet the conditions; if the conditions are not met, the missing data marker is retained to avoid misinterpreting device abnormalities as changes in patient status.

[0036] S1.4, Calculate the data quality score and network congestion status. The data quality score for the r-th monitoring indicator can be determined by the following formula:

[0037] ;

[0038] In the formula, For the number of valid samples, To determine the expected number of samples that should be obtained according to the standard sampling period, This represents the average sampling time deviation. Rate the equipment status. Score the connection status. to As a weight for quality evaluation, This means that the calculation results are mapped to the range of 0 to 1. The higher the data integrity, the smaller the sampling time deviation, and the more stable the device and connection status, the higher the data quality score.

[0039] Network congestion status can be determined by the following formula:

[0040] ;

[0041] In the formula, For link or node load, To normalize the network packet loss rate, and These are the normalized node queue length and the link queue length, respectively. To normalize the available bandwidth, to These are network state weights. Network congestion states are influenced by factors such as increased network load, packet loss rate, queue length, or decreased available bandwidth. The corresponding increase.

[0042] S1.5, Estimating the time it takes for key participation indicators to approach the risk boundary. The system determines the normalized risk boundary distance δ based on the current representative value, individualized reference range, and risk boundary, and calculates the approach trend quantity, approach rate, and estimated time to approach the risk boundary:

[0043] ;

[0044] In the formula, For the first Class of monitoring indicators in the first The normalized distance from the representative value of the time period to the nearest risk boundary; The time interval between adjacent time periods; To prevent the use of pre-defined positive numbers with a denominator of zero; This means that the calculation results will be limited to a given upper and lower limit. When the monitoring indicator has reached or exceeded the risk boundary, the time to approach the risk boundary will be set to 0; when the approach rate is lower than the preset rate threshold, this time will be set to the upper limit of the prediction. Indicates the first The risk boundary of key participation indicators is close to the trend volume at the current moment. Indicates the first Key participation indicators at the current moment The risk boundary approaches the rate. This indicates the expected timeframe for approaching the risk boundary.

[0045] S1.6, forming a dynamic time window. For key participating indicators of the same medical semantics, the shortest proximity time is selected, and the proximity trends of each participating indicator are aggregated by maximizing the value or using weighted summation. Then, the base time window and the dynamic time window are calculated:

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] In the formula, Indicates the first Medical semantics at the present moment The risk boundary is approaching time. Indicates the first Medical semantics at the present moment The overall risk boundary is close to the trend volume. Indicates participation in the The complete set of monitoring indicators for medical semantic construction. Indicates the first The trend weight of the participation indicator in the overall risk boundary is close to the trend quantity. Indicates the first Medical semantics at the present moment The basic time window, This indicates that the hospital side is the first Reference observation time for medical-like semantic configuration; Indicates the formation of the first Minimum observation time required for a class of effective medical semantic units; , and These represent the combined weights of the reference observation time, the minimum observation time, and the time of approach to the risk boundary within the base time window, respectively. Indicates the first The minimum time window allowed for medical-like semantics; Indicates the first The maximum time window allowed for medical-like semantics; This represents the interval cutoff function. For the previous closed-loop cycle Transmission control of risk status in medical semantics. This is a coefficient used to adjust the degree of time window extension due to network congestion. This is an adjustment coefficient for the degree of time window shortening in relation to transmission control risk states. This is an adjustment coefficient for the shortening of the time window as the risk boundary approaches its trend. When patient-side data is stable and network congestion is high, the time window is appropriately extended to aggregate low-urgency data; when the transmission control risk status increases or key indicators rapidly approach the risk boundary, the time window is shortened to generate data to be transmitted more quickly.

[0051] S1.7, Forming Candidate Medical Semantic Units. Within a dynamic time window, the system, based on time relevance, indicator relevance, and event relevance, selects physiological monitoring data, behavioral event data, and device status data that satisfy the corresponding medical semantic rules to form candidate medical semantic units. Data quality score of candidate medical semantic units The weighted average of the quality scores of each participating indicator can be used; for scenarios where the quality of any key indicator is required to reach a threshold, the minimum value of the quality score of the key participating indicator can also be used.

[0052] S1.8, Calculate the transmission trigger prompt score. The system determines the transmission trigger prompt score based on single-indicator threshold prompts, short-term trend prompts, multi-indicator joint prompts, historical state offset prompts, and preset rule trigger prompts:

[0053] ;

[0054] In the formula, Provide a score for a single indicator threshold. Score for short-term trend indications. For multi-indicator joint prompt scoring, Score the historical state offset indicator. Pre-set rules for hospitals to trigger scoring prompts. to The corresponding weights are used. This score only indicates the degree to which the timeliness of the corresponding data transmission needs to be improved.

[0055] S1.9, Determine the transmission control risk level. The system first calculates the original risk level based on the risk status of the previous period, the current alert score, the trend strength of the indicators, and the level of concern. Then, it performs confidence correction using data quality, the synergistic support of multiple indicators, and the support of historical consistency.

[0056] ;

[0057] ;

[0058] ;

[0059] In the formula, Indicates the first The original transmission control risk level of each candidate medical semantic unit. Indicates the first Medical semantics in the previous closed-loop cycle Transmission control risk status, Indicates the first The transmission of candidate medical semantic units triggers a scoring prompt. This represents a normalization function used to map calculation results to a preset numerical range, preferably to... , Indicates the first Risk confidence correction factors corresponding to each candidate medical semantic unit. Indicates the first Data quality scores for each candidate medical semantic unit. Indicates the first The current transmission control risk level of each candidate medical semantic unit after confidence correction. The trend strength of indicators for candidate medical semantic units; The level of attention for hospitals; To provide support for the coordinated performance of multiple indicators; To support the consistency between current changes and historical patient monitoring patterns; This is the lower limit of the confidence correction factor; to and to Preset weights are used. Confidence correction can reduce the amplifying effect of equipment malfunctions, low-quality sampling, or isolated outliers on the transmission control risk level.

[0060] S1.10, Determine the medical urgency level. The system calculates the medical urgency level based on the current transmission control risk level, transmission trigger prompt score, timeliness level, indicator trend strength, and hospital-side attention level:

[0061] ;

[0062] In the formula, to Weighted according to the urgency of medical needs. (Medical urgency level) The dimensionless parameter is between 0 and 1 and is only used for subsequent data packetization, candidate path cost modulation, queue scheduling, and transmission resource allocation. For a single resource allocation process, only one of bandwidth, transmission time slot, or queue service share is selected as the current resource type, and resources of the same type are allocated according to the score ratio. When the data quality score is lower than the quality threshold and the collaborative support and historical consistency support are both low, the system generates a summary packet or re-collection prompt packet with quality markers, instead of directly improving the real-time upload level of complete extended evidence.

[0063] Step S1 ultimately outputs candidate medical semantic units. Data quality score Transmission trigger prompt rating Transmission control risk level Medical urgency Privacy level, timeliness level, current network congestion status, and dynamic time window are available for use in steps S2 to S6.

[0064] S2, the candidate medical semantic units are reorganized into mutually bound core risk packages and extended evidence packages, and a dependency relationship is established between the extended evidence packages and the core risk packages;

[0065] Step S2 is used to separate small-volume key fields from large-volume evidence data in the same candidate medical semantic unit, so that core data can enter the real-time transmission process first, while retaining the ability of the receiving end to obtain complete evidence as needed.

[0066] S2.1, Determine the packet triggering condition. When the transmission control risk level of the candidate medical semantic unit... or medical emergency When the corresponding threshold is reached, dependency subpackaging is performed on the unit; when , When both the transmission trigger prompt score and the data score are below the corresponding threshold, the data is summarized, compressed, or batch aggregated, without being forced to form a complete extended evidence package.

[0067] S2.2, Form the core risk package. The core risk package includes key indicator fields, data generation time, transmission control risk level, medical urgency, data quality score, privacy level, timeliness level, and context summary. Key indicator fields prioritize the values, directions of change, rates of change, and corresponding device identifiers of indicators that trigger transmission control rules, enabling the receiving end to obtain the key content of this transmission task without waiting for large volumes of raw data.

[0068] S2.3, Form an extended evidence package. The extended evidence package includes at least one of the following: original monitoring segments corresponding to the core risk package, continuous waveforms, historical comparison data, and context data from adjacent time periods. The system can set the maximum segment size, compression method, cache validity period, and maximum allowed completion time of the extended evidence package according to the data type.

[0069] ;

[0070] ;

[0071] ;

[0072] In the formula, This is a key indicator field. For the time when the data was generated, For context summary, This is the original monitoring segment. It is a continuous waveform. For historical comparison data, For adjacent time periods; For medical semantic unit identification, Identify the core risk package to which it belongs. As a dependency state, This is the maximum allowed completion time.

[0073] S2.4, Establish Dependencies. The core risk package and the extended evidence package use the same medical semantic unit identifier. The extended evidence package records the identifier of the core risk package it belongs to. The basic dependency condition is: the extended evidence package must not enter the real-time transmission queue before the core risk package; preferably, the corresponding extended evidence package is allowed to enter the real-time transmission queue only after receiving successful feedback on the transmission of the core risk package or after the hospital platform generates a supplementary evidence request. If the core risk package is not successfully transmitted and no supplementary evidence request is generated, the extended evidence package is stored in a trusted cache node or a delayed upload queue that meets privacy level requirements.

[0074] S2.5, Set transmission markers. The core risk packet is configured for immediate or real-time upload; the extended evidence packet can be configured for real-time supplementation, trusted caching, asynchronous upload, or batch upload based on medical urgency, data volume, network congestion status, and hospital-side concern level. This forms a medical semantic transmission unit, which outputs the data packet type, source node, target node, privacy level, encryption requirements, minimum bandwidth, minimum cache space, maximum allowed completion time, and dependency identifier to step S3.

[0075] S3. Based on the medical IoT network topology, candidate paths and corresponding transmission plan combinations are generated, and hard constraints are applied to filter them according to security level, encryption capability, minimum carrying capacity and the aforementioned dependency relationship.

[0076] Step S3 is used to establish a compliant feasible domain that cannot be relaxed by medical urgency before calculating the soft path cost. The system first establishes or updates the medical IoT network topology model. The node attributes in the topology include at least node type, node security level, encryption capability, trusted caching capability, available cache balance, and queue status. The link attributes include at least link security level, available bandwidth, basic latency, packet loss rate, and congestion status.

[0077] S3.1, Generate candidate paths and transmission plans. Based on the source and target nodes of the medical semantic transmission units, generate a set of candidate paths in the current network topology using K-shortest path, constrained shortest path, or other repeatable path search methods, and generate a corresponding transmission plan for each path:

[0078] ;

[0079] ;

[0080] In the formula, This indicates one of the following upload types: immediate upload, real-time upload, trusted caching, asynchronous upload, or batch upload. Indicates the time slot or order in which items are planned to enter the queue; Indicates the trusted cache nodes to be used; Indicates the encryption method; This indicates the maximum allowed completion time. Each candidate is composed of a path and a transmission plan, avoiding scoring based solely on the path while ignoring the sending order and buffering arrangements.

[0081] S3.2, Determine the path security level. The path security level of a candidate path is the lowest value of the security levels of all nodes and all links in the path:

[0082] ;

[0083] in, Indicates the first The medical semantic transmission unit is transmitted via the first The path security level when transmitting through candidate paths. Represents a node The node security level, Indicates communication link The link security level is determined by this process, which ensures that any low-security node or low-security link in the path will lower the security level of the entire path, thus avoiding the need to compensate for low-security links through the scores of other high-security nodes.

[0084] S3.3, implement hard constraints for security and encryption screening. When the path security level is lower than the privacy level of the medical semantic transmission unit, the corresponding combination is eliminated; when the data packet requires encrypted upload but the candidate path cannot provide the corresponding encryption method, key negotiation method, or trusted execution conditions, the corresponding combination is eliminated. Medical urgency is not subject to the relaxation of the above conditions.

[0085] S3.4, Perform minimum carrying capacity screening. For core risk packets that are uploaded immediately or in real time, the remaining bandwidth of the bottleneck of the candidate path should not be less than the minimum bandwidth, and the transmission should be completed within the maximum allowable time. For extended evidence packets that are uploaded in trusted cache, asynchronously, or in batches, there should be trusted cache nodes in the candidate path that meet the security level requirements, and the remaining trusted cache space should not be less than the minimum cache space.

[0086] S3.5, Perform dependency order filtering. When a candidate path and transmission plan combination would cause the extended evidence packet to enter the real-time transmission queue before its parent core risk packet, the combination should be eliminated. When using the dependency method of "transmitting extended evidence only after the core packet is successfully sent", the transmission plan should also include the trigger condition for successful feedback of the core risk packet or the trigger condition for supplementary evidence request.

[0087] S3.6, execute a fallback if no feasible path is found. When none of the combinations meet the hard constraints, generate local alarm records and retransmission tasks for the core risk package, and temporarily store it in a trusted cache node without reducing the security level; the extended evidence package is added to the delayed upload queue. When the network status recovers, cache resources are released, or a new compliant path appears, candidate path generation and hard constraint screening are re-executed.

[0088] S4, among the combinations filtered by hard constraints, modulates the delay, queuing, packet loss and congestion costs according to the medical urgency to select the target path;

[0089] Step S4 calculates the network performance cost only for the candidate paths and transmission plan combinations selected in step S3. By setting security conditions, encryption capabilities, minimum bandwidth, trusted caching capabilities, and dependency order as hard constraints, paths that do not meet security or bearer requirements can be avoided by using soft scoring to compensate for medical urgency.

[0090] S4.1, Determine network performance metrics. Calculate the normalized basic transmission delay for each candidate path. Normalized queuing delay Normalized packet loss risk and normalized path congestion The basic transmission delay is determined by the propagation delay, transmission delay, and node processing delay of each link in the path; the queuing delay is determined by the node queue length, link queue length, and service rate; the packet loss risk is determined by the historical packet loss rate, the current bit error rate, and the retransmission rate; and the congestion level is determined by the bandwidth utilization rate, queue utilization rate, and data arrival rate.

[0091] S4.2, Determine the bearer mismatch penalty. The bearer mismatch penalty can be calculated based on the difference between the data volume and the remaining path bandwidth and trusted cache availability.

[0092]

[0093] In the formula, To normalize the data volume, To normalize the remaining bandwidth matching amount, To normalize the remaining amount of trusted cache, and Weights are assigned to bandwidth mismatch and cache mismatch. For core risk packets uploaded immediately, caching can be disabled; for extended evidence packets that use caching, asynchronous uploads, or batch uploads, the effectiveness of caching can be increased.

[0094] S4.3, Determine the cache match level. Cache match level. The available cache space, expected cache duration, and subsequent forwarding capability of trusted cache nodes that meet privacy level requirements in the path are determined; when the medical semantic transmission unit is an immediately uploaded core risk packet that does not allow caching, then... It equals 0.

[0095] S4.4, Calculate path cost. Medical urgency modulates latency, queuing, packet loss, and congestion costs separately, allowing different network performance terms to adjust independently based on corresponding feedback, rather than applying the same common multiplier to all network penalty terms.

[0096] ;

[0097] In the formula, For the first The medical semantic transmission unit is transmitted via the first The path cost when transmitting along multiple candidate paths; For medical emergencies; To carry the penalty for mismatch; For cache matching degree; to Weights for each cost item; , , and These are the modulation coefficients for the medical urgency level on the basic transmission delay, queuing delay, packet loss risk, and congestion level, respectively. Each input indicator is mapped to the same numerical range before calculation.

[0098] S4.5, Select the target path. Compare the path costs of the same medical semantic transmission unit on each candidate path, and determine the candidate path with the lowest path cost and its transmission plan as the target combination. When two paths have the same cost or the difference is less than the preset tolerance, the path with a higher security level, larger remaining bandwidth at the bottleneck, or larger trusted cache reserve is given priority.

[0099] S5 performs queue scheduling and resource allocation based on the unmodulated basic cost, waiting time and core risk packet marking of the target path, so that the core risk packet obtains transmission resources no less than the preset lower limit, and sends the extended evidence packet after the dependency conditions are met.

[0100] Step S5 addresses the issue of multiple medical semantic transmission units competing for the same node queue or link queue after the target path is determined. To avoid repeatedly amplifying medical urgency at the path and queue levels, queue scheduling uses the target path base cost that does not include an urgency modulation term.

[0101] S5.1, Calculate the unmodulated base cost. Calculate for the target path selected in step S4:

[0102] ;

[0103] In the formula, Represented as the first The target path selected by each medical semantic transmission unit; Used to reflect the underlying transport burden of the path itself, excluding , , and Modulation of medical urgency.

[0104] S5.2, Calculate the basic transmission priority. The basic transmission priority can be determined by the following formula:

[0105] ;

[0106] In the formula, Based on transmission priority, For privacy level, Score the data quality. To normalize the data volume, to The weight is used for data volume. Data volume is used as a negative term to reduce the likelihood of large, expanding data continuously consuming real-time resources, all other things being equal.

[0107] S5.3, Calculate the queue scheduling score. The system calculates the score based on the basic transmission priority, medical urgency, immediate upload flag, core risk packet flag, waiting time, and basic cost of the target path:

[0108] ;

[0109] In the formula, Upload tags immediately; Mark the core risk package; This is the normalized waiting time accumulated since the data packet entered the sending queue; to Assign queue scheduling weights. It increases with waiting time, which can prevent ordinary medical data from being starved for a long time; As a negative factor, it can prevent tasks with a high path load from being continuously placed at the front of the queue.

[0110] S5.4, Determine the queue order. The system arranges data packets in the same node queue or link queue from high to low according to the scheduling score, and inserts data packets with both the immediate upload flag and the core risk packet flag before all non-immediate upload data packets. To avoid corrupting data frames being transmitted, the priority insertion operation takes effect after the current uninterruptible data segment has been transmitted.

[0111] S5.5, Allocate transmission resources. The scheduling score is non-negatively processed, and then the resource allocation is determined based on the proportion of non-negative scheduling scores.

[0112] ;

[0113] ;

[0114] ;

[0115] In the formula, Indicates the first Scheduling score of each medical semantic transmission unit after nonnegation processing The minimum positive number is preset. The number of medical semantic transmission units currently participating in resource allocation; The bandwidth, transmission slot, or service share allocated to the i-th medical semantic transmission unit; This represents the minimum resource allocation for the core risk package. When the resources allocated to the core risk package according to the scoring percentage fall below this minimum, the difference will be allocated from the resources of the extended evidence package, asynchronously uploaded data, or batch uploaded data.

[0116] S5.6, Sending is performed according to dependency conditions. First, the core risk package is sent. After receiving successful feedback on the core risk package's transmission, the extended evidence package corresponding to the same medical semantic unit identifier is added to the real-time sending queue. The priority of each evidence fragment in the extended evidence package is determined based on the time distance, event correlation, and data quality score between each evidence fragment and the key indicator fields of the core risk package. Data volumes corresponding to the real-time upload ratio are selected and added to the real-time sending queue in descending order of priority. The remaining evidence fragments are stored in a trusted cache node or a delayed upload queue, or added to the real-time sending queue after a supplementary evidence request is generated on the hospital platform. If the core risk package is not successfully sent and there is no supplementary evidence request, the extended evidence package remains in the trusted cache node or delayed upload queue. The extended evidence package is sent only when network congestion decreases, the core risk package is successfully sent, or real-time resources are available.

[0117] S6 records the transmission delay, queuing delay, packet loss rate, and transmission success rate of core risk packets, and adjusts the proportion of real-time upload of extended evidence and transmission control parameters for the next closed-loop cycle accordingly.

[0118] Step S6 records the actual transmission effect of each medical semantic transmission unit with a preset time period as the closed-loop update cycle, and updates the network congestion status, risk status, dynamic time window, path weight, queue parameters, core resource lower limit and extended evidence real-time upload ratio for the next cycle based on the service quality of the core risk package and the feedback from the receiving end.

[0119] S6.1 Record transmission feedback. The feedback record for each medical semantic transmission unit can be represented as follows:

[0120] ;

[0121] In the formula, The message has been successfully sent. This is the actual end-to-end delay; This represents the actual packet loss rate. This refers to the actual queuing delay or queue status. This represents the actual congestion state. This represents the number of retransmissions. A feedback flag is triggered for transmission; Request to mark as supplementary evidence; To view the confirmation marker. Feedback is bound to the original data packet according to the medical semantic unit identifier. Late feedback is returned to its corresponding original closed-loop cycle or merged into the current cycle according to preset rules.

[0122] S6.2, Update network congestion status. Calculate the network congestion status for the next period based on the average actual congestion status, average queue status, and average packet loss rate of the current period:

[0123]

[0124] In the formula, , and These represent the average congestion state, average queue state, and average packet loss rate for the current period. to For feedback weights.

[0125] S6.3, Update the transmission control risk status. Aggregate the transmission trigger feedback flags corresponding to similar medical semantics into... The requests for supplementary evidence and the confirmation markers for viewing are aggregated into The transmission control risk status can be updated using the following formula:

[0126] ;

[0127] In the formula, To ensure the stability of transmission control for this type of medical semantics within the current cycle, to To update the weights. It is used to adjust the level of attention or expand the proportion of real-time evidence uploaded, and is not directly used as a result of judging the validity of medical abnormalities.

[0128] S6.4, Update Path and Queue Parameters. The system adjusts the weights of latency, queuing, packet loss, and congestion costs, as well as the urgency modulation coefficient, based on the average end-to-end latency, average queuing latency, average packet loss rate, and average congestion status of high-emergency core risk packets. It also adjusts the waiting time weight, basic path cost weight, and minimum resource ratio for core risk packets based on the average waiting time and core risk packet transmission success rate of similar data. Initial values, update step sizes, and upper and lower limits are set for each parameter. When the target conditions are met for multiple consecutive cycles, the update step size is reduced or the current value is maintained to reduce closed-loop oscillations. The minimum resource ratio for core risk packets can be updated using the following formula:

[0129] ;

[0130] In the formula, The average waiting time for the high-emergency medical core risk package. To improve the success rate of sending core risk packets, and For adjustment coefficients, and These are the upper and lower limits of the minimum resource ratio for the core risk package, respectively.

[0131] S6.5, Update the real-time upload ratio of extended evidence. For the m-th type of medical semantics, update the real-time upload ratio of extended evidence for the next cycle based on the amount of attention and feedback from the hospital platform and the actual service quality of the core risk package:

[0132] ;

[0133] In the formula, and These represent the real-time upload ratios of extended evidence for the current period and the next period, respectively. To supplement evidence requests and view the amount of attention feedback obtained from aggregated confirmation markers; Normalized actual end-to-end latency for high-emergency medical core risk packages; Normalize its actual packet loss rate; To improve the success rate of sending core risk packages; to This is the adjustment coefficient; and To extend the upper and lower limits of the real-time evidence upload ratio, when the latency or packet loss of core risk packets worsens and the transmission success rate decreases, reducing the control priority of the extended evidence real-time upload ratio is higher than the control priority of increasing this ratio based on feedback, thereby prioritizing the release of real-time transmission resources.

[0134] S6.6, Enter the next closed-loop cycle. Substitute the updated network congestion status, transmission control risk status, and the trend of key participation indicators for the next cycle into the dynamic time window calculation process in step S1 to obtain the dynamic time window for the next cycle; then repeat the process of candidate medical semantic unit formation, dependency packetization, hard constraint path screening, path cost calculation, queue scheduling, and hierarchical transmission.

[0135] The following example uses "blood glucose change medical semantics" in the scenario of continuous home monitoring of patients. The complete process of this invention from multi-source data acquisition, candidate medical semantic unit formation, dependency subpackaging, path and transmission plan hard constraint screening, target path selection, queue scheduling, resource allocation, dependency release and closed-loop update is reproduced according to steps S1 to S6.

[0136] In this embodiment, the reference range, risk boundary, attention level, timeliness level, and various control thresholds are transmission control parameters pre-configured by the hospital or updated non-diagnostologically based on historical monitoring data. They are only used to determine the aggregation, packetization, routing, queue scheduling, and transmission resource allocation methods for medical data, and are not used to output disease diagnosis conclusions, condition judgment results, or treatment plans. The values ​​in this embodiment are used to illustrate the deterministic execution process of the algorithm, do not constitute clinical thresholds, and do not represent actual hospital network test results.

[0137] On the patient side, continuous glucose monitoring devices, wearable activity monitoring devices, and mobile terminals are installed; the home gateway is used to aggregate glucose data, activity data, food intake records, medication records, and device status data. On the hospital side, a hospital edge server, trusted cache nodes, a hospital data platform, and a doctor's terminal are installed. The home gateway can send medical semantic transmission units to the hospital data platform via three network paths.

[0138] The medical semantic transmission rules and control parameters adopted in this embodiment are shown in Table 1:

[0139] Table 1

[0140]

[0141] At the current time of 08:30, the continuous glucose monitoring device recorded a representative blood glucose value of 8.7 mmol / L at the previous valid time of 08:25 and a representative blood glucose value of 9.2 mmol / L at the current time; the mobile terminal recorded an eating event at 08:18 and a medication event at 08:22; the wearable device recorded the activity intensity as low intensity; the blood glucose monitoring device's self-test status is normal, the remaining battery power is 72%, and the home gateway connection status is normal.

[0142] The system standardizes data timestamps, eliminates duplicate records, and marks the data sources of blood glucose devices, wearable devices, and mobile terminals. The current dynamic window was expected to collect 4 blood glucose samples, and 4 valid samples were actually collected, with an average sampling time deviation of 0.3 minutes, a device status score of 0.95, and a connection status score of 0.92. After quality evaluation, the [number missing]... The data quality score for each candidate medical semantic unit is:

[0143] ;

[0144] in, For the first The data quality score for each candidate medical semantic unit indicates that the higher the value, the better the data integrity, time consistency and device reliability.

[0145] The home gateway synchronously obtains the current network status: link load, node queue length, and link queue length have all increased, and available bandwidth has decreased. After normalizing the various network status indicators, the current network congestion status is obtained:

[0146] ;

[0147] in, This represents the network congestion status in the current closed-loop cycle, with a value ranging from 0 to 1. A larger value indicates a more congested network.

[0148] Since this embodiment uses the upper risk boundary, the normalized risk boundary distances at the previous and current times are respectively:

[0149] ;

[0150] ;

[0151] In the formula, and The previous time and the current time are respectively the first and second times. Representative values ​​of key participation indicators; The upper risk boundary configured for the hospital side; This represents the upper limit of the individualized reference range; and These are the normalized distances from the previous time step and the current time step to the risk boundary, respectively.

[0152] The risk boundary is close to the trend quantity:

[0153] ;

[0154] The time interval between two adjacent valid moments is min, therefore the approach rate to the risk boundary is:

[0155] ;

[0156] Set a positive number to prevent the denominator from being zero. The estimated time to approach the risk boundary is:

[0157] ;

[0158] Due to the set of key participation indicators It contains only blood sugar, therefore:

[0159] ;

[0160] For each monitoring indicator in the indicator set, the near-trend value of the risk boundary is aggregated using the maximum value method. In this embodiment, the near-trend value of the blood glucose indicator is the largest, therefore:

[0161] ;

[0162] Set the base time window combined weights , , ,but:

[0163] ;

[0164] Substitute the values:

[0165] ;

[0166] The transmission control risk status of the previous closed-loop cycle was The window adjustment factor is set to , , The dynamic time window is then:

[0167] First, calculate the window adjustment factor:

[0168] ;

[0169] Recalculate the dynamic time window:

[0170] ;

[0171] Substituting the values ​​into the equation:

[0172] ;

[0173] ;

[0174] Therefore, the system extracts data from approximately 17.89 minutes prior to the current time 08:30, with the actual data extraction interval being approximately 08:12:07 to 08:30:00.

[0175] The system extracts blood glucose sequences, food intake records, medication records, activity status, and device status from the dynamic time window, and performs time-related, indicator-related, and event-related matching. The current dataset contains the key indicator "blood glucose," and the actual observation time is greater than the minimum effective observation time of 10 minutes, thus forming a complete candidate medical semantic unit.

[0176] ;

[0177] The system obtains individual indicator threshold prompt scores respectively. Short-term trend indicator score Multi-indicator joint prompt scoring Historical state offset prompts and scores and hospital-side rule trigger prompts and scoring Set weights , , , , ,but:

[0178] ;

[0179] ;

[0180] in, This only indicates the degree to which the current medical semantic unit needs to improve the timeliness of transmission.

[0181] This embodiment takes the trend strength of the indicator. Hospital-side attention level .set up , , , The original transmission control risk level is:

[0182] ;

[0183] ;

[0184] Multi-indicator synergistic support degree Historical consistency support degree ,set up , , as well as The confidence correction factor is:

[0185] ;

[0186] ;

[0187] The revised transmission control risk level is:

[0188] ;

[0189] set up , , , , ,but:

[0190] ;

[0191] ;

[0192] Step S1 outputs candidate medical semantic units Data quality score Transmission trigger prompt rating Transmission control risk level Medical urgency Privacy level 4, timeliness level 0.90, current network congestion status 0.65, and dynamic time window 17.89 min.

[0193] Hospital-side preset risk level subcontracting threshold Medical urgency level subcontracting threshold .because:

[0194] ;

[0195] and:

[0196] ;

[0197] The system analyzes candidate medical semantic units. Execute dependency subpackages.

[0198] Core risk package Includes: current representative blood glucose value 9.2 mmol / L, previous representative value 8.7 mmol / L, direction of change, risk boundary distance 0.3636, approach rate 0.04545 min^-1, data generation time, , , Privacy level, timeliness level, device identifier, and context summary. The core risk package size is 2.4kB, the maximum allowed completion time is 0.50s, and the upload type is set to "Upload Now".

[0199] It can be represented as:

[0200] ;

[0201] Extended evidence package This includes complete blood glucose sequences, food and medication records, activity sequences, device status, historical comparison data from the previous similar time period, and adjacent context within a dynamic time window. The extended evidence package size is 160kB, the maximum allowed completion time is 30 seconds, and the upload type is initially set to "trusted cache followed by real-time supplementation as needed".

[0202] It can be represented as:

[0203] ;

[0204] The core risk package and the extended evidence package use the same The expanded evidence package records the to which it belongs. The initial dependency status is set to 0, indicating that the extended evidence package has not yet been granted permission to enter the real-time sending queue; when the core risk package is successfully sent or the hospital platform generates a request for supplementary evidence, the dependency status is updated to 1.

[0205] The system generates three acyclic candidate paths in the current medical IoT topology. The attributes of each path are shown in Table 2 below.

[0206] Table 2

[0207]

[0208] The path security level is determined by the following formula:

[0209] ;

[0210] therefore:

[0211] ;

[0212] Because the current privacy level of medical semantic transmission units is ,path satisfy:

[0213] ;

[0214] Therefore, the path Its corresponding transmission plan is directly excluded and not included in the path cost calculation; even if the path It has high bandwidth and low base latency, and does not allow compensation through medical urgency.

[0215] path and path Both support TLS 1.3 and AES-256 encryption methods required by the core risk package, and the remaining bandwidth of the bottleneck is higher than 0.30 Mbit / s. Therefore, the core risk package upload plan for both of them passes the hard constraints of security, encryption and minimum bandwidth.

[0216] For the extended evidence package, path The trusted cache space is 100kB, which is less than the 160kB required for expanding the evidence packet, therefore the path... The corresponding "trusted caching and on-demand replenishment" plan has been removed; path The trusted cache has a remaining capacity of 500kB, which meets the caching requirements. Path The corresponding plan also includes the dependency condition that "the extended evidence package can only enter the real-time queue after the core risk package is successfully reported or a supplementary evidence request is triggered," thus creating a hard constraint on the dependency order.

[0217] The hard constraint screening results show that the feasible path for the core risk package is... and The feasible plan for expanding the evidence package is the path. A trusted caching plan.

[0218] The core risk package is in the path and path The network performance metrics were normalized, as shown in Table 3:

[0219] Table 3

[0220]

[0221] in, Based on transmission delay, To delay queuing, To mitigate the risk of packet loss, To indicate the degree of congestion, To accommodate the mismatch penalty, To determine the cache matching level. Since the current object is a core risk package that cannot be cached and must be uploaded immediately, let... .

[0222] set up , , , , , and set , , , The path cost is:

[0223] For ease of calculation, the path cost is decomposed into four network performance sub-costs:

[0224] ;

[0225] ;

[0226] ;

[0227] ;

[0228] The total cost of the path is:

[0229] ;

[0230] For path :

[0231] ;

[0232] ;

[0233] ;

[0234] For path :

[0235] ;

[0236] ;

[0237] ;

[0238] because:

[0239] ;

[0240] The system will path Its core risk package and its immediate upload plan were identified as the target portfolio. Although the path While its basic transmission latency is slightly higher, its queuing latency, packet loss risk, and congestion level are lower, thus achieving a lower overall path cost after modulating the medical urgency level.

[0241] Target path The unmodulated basic cost is:

[0242] ;

[0243] ;

[0244] in, This indicates the target path selected in step S4. .

[0245] The basic transmission priority of the current core risk packet is taken as follows Upload tags immediately Core risk package marked Normalization waiting time .set up , , , , , ,but:

[0246] ;

[0247] ;

[0248] Currently, there are three non-immediately uploaded extended or batch data packets in the same sending queue, with scheduling scores of 0.90, 0.80, and 0.70, respectively. Although some of these non-immediately uploaded data packets have scores higher than 0.6518, this invention first inserts the current core risk packet before all non-immediately uploaded data packets based on the immediate upload flag and the core risk packet flag, and sends the core risk packet first after the current uninterrupted data segment is sent.

[0249] Without implementing a minimum adjustment to core resources, the current core risk package receives the following resource proportions based on its scoring percentage:

[0250] ;

[0251] That is, 21.36%, which is lower than the minimum resource ratio of 25% for the core risk package. Therefore, the system proportionally reclaims 3.64 percentage points from the resources of the three non-core data packages to increase the current resource ratio of the core risk package to:

[0252] ;

[0253] After the adjustment, the resource ratios of the three non-core data packets are 28.125%, 25.000%, and 21.875% respectively, with the sum of all resource ratios being 100%.

[0254] Target path The bottleneck has a remaining bandwidth of 1.2 Mbit / s, therefore the minimum bandwidth allocated to the current core risk packet is:

[0255] ;

[0256] The core risk packet is 2.4kB in size, or 19.2kbit, and its pure transmission time is approximately:

[0257] ;

[0258] System via path Send the core risk packet. Once the current uninterruptible data segment is complete, the core risk packet enters the queue; the actual queuing delay is 85ms, the data transmission time is 64ms, and the total time for path propagation, processing, encryption, and acknowledgment is approximately 225ms. Therefore, the actual end-to-end delay is:

[0259] ;

[0260] Since 374ms is less than the maximum allowed completion time of 500ms, and the core risk packet has arrived completely at the hospital platform, the successful transmission status is set to 1. The system updates the dependency status of the extended evidence packet from 0 to 1 and allows it to enter the subsequent transmission process.

[0261] The hospital platform generates a request for supplementary evidence after receiving the core risk package. Currently, the real-time upload rate for extended evidence is 40%. Therefore, of the 160kB extended evidence package, 64kB is added to the real-time transmission queue, and the remaining 96kB is stored in the specified path. Trusted edge cache nodes are used to asynchronously resume transmission when network resources are idle or in subsequent cycles.

[0262] The current feedback record for the medical semantic unit includes at least: core risk packet transmission success status 1, actual end-to-end latency 374ms, actual queuing latency 85ms, actual packet loss rate 0, retransmission count 0, supplementary evidence request flag 1, and view confirmation flag 1. The system will link the feedback record with... Binding.

[0263] Within this closed-loop cycle, after aggregating similar high-urgent-risk core risk packages, the normalized high-urgent-risk core package latency is obtained. Normalized packet loss rate Success rate of sending core risk packages and the amount of feedback from hospitals .

[0264] Current period average congestion status Average queue state Average packet loss rate ,set up , , The network congestion state in the next cycle will be:

[0265] ;

[0266] ;

[0267] The current period's extended evidence real-time upload ratio is ,set up , , , ,and , ,but:

[0268] First, calculate the update volume of the real-time upload ratio of extended evidence:

[0269] ;

[0270] Then perform upper and lower limit truncation:

[0271] ;

[0272] Substituting the values ​​into the equation:

[0273] ;

[0274] ;

[0275] Therefore, the proportion of extended evidence uploaded in real time for the next closed-loop cycle was reduced from 40% to 29.8%. Although there was feedback from the hospital regarding supplementary evidence, the overall latency and success rate of the high-urgency core package in this cycle had not yet reached their optimal state. Therefore, the system prioritized reducing the real-time resource consumption of extended evidence to ensure the transmission resources of the core risk package for the next cycle.

[0276] Assuming the key indicators show a slowdown in change in the next cycle, the expected time to approach the risk boundary is updated to 15 minutes, the overall trend indicator is updated to 0.12, and the risk status of similar medical semantic transmission control in the previous cycle is updated to 0.54. Therefore, the base time window for the next cycle is:

[0277] ;

[0278] The next dynamic time window is:

[0279] ;

[0280] ;

[0281] Compared to the current period of 17.89 minutes, the next period window is slightly longer because key indicators are approaching a weakening trend and network congestion is decreasing; however, the window is still constrained by transmission control risk status and has not returned to the reference observation time of 30 minutes.

[0282] The system will , The updated transmission control risk status, dynamic time window, path parameters, and queue parameters are written into the next closed-loop cycle configuration, and steps S1 to S6 are re-executed.

[0283] Through the complete execution process of this embodiment, the system first forms medical semantic units within a dynamic time window of 17.89 minutes. Subsequently, the 2.4kB core risk package was separated from the 160kB extended evidence package; path The path was removed due to insufficient safety level and hard constraints. Although it can transmit core risk packets, the combined costs of queuing, packet loss, and congestion are higher than those of the original path. Therefore, the path The core risk packet is selected as the target path; during the queue scheduling phase, a resource limit of 25% is set for the core risk packet to keep its actual end-to-end latency within 374ms; only after the core risk packet is successfully sent can the extended evidence packet obtain real-time transmission permission; closed-loop feedback further reduces the proportion of real-time upload of extended evidence in the next cycle from 40% to 29.8%;

[0284] To illustrate the technical effects of this invention, a simulation scenario for continuous monitoring and transmission in the medical Internet of Things (IoT) is constructed based on preset parameters. The network includes patient-side monitoring devices, home or bedside gateways, access nodes, hospital edge servers, in-hospital medical data platforms, and doctor terminals. Figure 1 The illustrated process sequentially performs the following steps on medical data: medical semantic unit formation, reassembly of core risk packets and extended evidence packets, path hard constraint screening, target path selection, queue scheduling, hierarchical transmission, and feedback updates. Simulations are set up with mild, moderate, and severe congestion states and compared with fixed-window ordinary routing, risk-aware routing, and semantic packet static scheduling methods. The simulation data is theoretical verification data constructed based on preset conditions to illustrate the performance change trends under different transmission strategies.

[0285] like Figure 2 As shown, the average end-to-end latency of core risk packets increases with increasing network congestion levels. The average end-to-end latency of this invention under light, moderate, and heavy congestion conditions is 220ms, 310ms, and 430ms, respectively. Under heavy congestion conditions, it is approximately 60.2% lower than the 1080ms of the fixed-window ordinary routing method, indicating that relying on packet segmentation, hard constraint filtering, and urgency-modulated path selection can reduce the transmission latency of core risk packets.

[0286] like Figure 3 As shown, under severe congestion, the on-time delivery rate of the core risk packets of this invention is 89.6%, which is higher than the 62.1% of the fixed window ordinary routing method, the 71.5% of the risk-aware routing method, and the 81.3% of the semantic packet static scheduling method. This indicates that this invention can prioritize the real-time transmission resources of the core risk packets when the network load fluctuates.

[0287] like Figure 4As shown, when the workload increases to 200 medical semantic units / minute, the average queuing latency of the core risk package in this invention is 176ms, which is lower than other comparative methods. This result indicates that by introducing core risk package marking, waiting time compensation, and a minimum resource ratio for core risk packages, queue blocking and queue starvation under mixed business concurrency conditions can be reduced.

[0288] like Figure 5 As shown, within eight consecutive closed-loop update cycles, the overall transmission efficiency score of this invention improved from 0.78 to 0.94, while the scheme that removed closed-loop updates only improved to 0.81. This result demonstrates that adjusting path parameters, queue parameters, and the proportion of real-time uploads of extended evidence based on the latency, packet loss rate, and transmission success rate of core risk packets can continuously improve the timeliness, reliability, and resource utilization efficiency of medical data transmission.

[0289] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for precise routing and closed-loop transmission of medical data based on patient risk status, characterized in that, Includes the following steps: S1 collects multi-source medical data and network status data from the patient side, forms candidate medical semantic units based on the medical semantic transmission rules configured at the hospital, and determines the transmission control risk level and medical urgency. S2, the candidate medical semantic units are reorganized into mutually bound core risk packages and extended evidence packages, and a dependency relationship is established between the extended evidence packages and the core risk packages. The reorganized core risk packages and extended evidence packages are collectively referred to as medical semantic transmission units. S3. Based on the medical IoT network topology, candidate paths and corresponding transmission plan combinations are generated, and hard constraints are applied to filter them according to security level, encryption capability, minimum carrying capacity and the aforementioned dependency relationship. S4, among the combinations filtered by hard constraints, modulates the delay, queuing, packet loss and congestion costs according to the medical urgency to select the target path; S5 performs queue scheduling and resource allocation based on the unmodulated basic cost, waiting time and core risk packet marking of the target path, so that the core risk packet obtains transmission resources no less than the preset lower limit, and sends the extended evidence packet after the dependency conditions are met. S6 records the transmission delay, queuing delay, packet loss rate, and transmission success rate of core risk packets, and adjusts the proportion of real-time upload of extended evidence and transmission control parameters for the next closed-loop cycle accordingly.

2. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 1, characterized in that, Step S1 includes: The medical semantic transmission rules are pre-configured by the hospital or updated non-diagnostologically based on the patient's historical monitoring data. The medical semantic transmission rules include at least a set of participating indicators, a set of key participating indicators, a risk boundary, a basic observation time, a minimum effective observation time, upper and lower limits of a dynamic time window, a privacy level, and a timeliness level. The multi-source medical data is subjected to timestamp unification, duplicate data filtering, error value removal, short-term missing data handling, data source marking, and data quality evaluation. Based on the distance of key participation indicators (KPIs) from the corresponding risk boundary and their rate of change, the time it takes for the KPIs to approach the risk boundary is estimated. A base time window is determined based on the time it takes for the KPIs to approach the risk boundary. The base time window is then adjusted based on the transmission control risk status of the previous closed-loop cycle, the current network congestion status, and the changing trend of the KPIs to obtain a dynamic time window.

3. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 2, characterized in that, Step S1 involves forming candidate medical semantic units and determining the transmission control risk level and medical urgency, including: Within the dynamic time window, physiological monitoring data, behavioral event data, and equipment status data are grouped into candidate medical semantic units according to time correlation, indicator correlation, and event correlation. The data quality score of the candidate medical semantic unit is determined based on the data quality scores of each participating indicator in the candidate medical semantic unit. The transmission trigger prompt score is determined based on at least two of the following: single indicator threshold prompt, short-term trend prompt, multi-indicator joint prompt, historical state deviation prompt, and preset rule trigger prompt. Based on the transmission trigger prompt score, the trend of changes in participating indicators, the transmission control risk status of the previous closed-loop cycle, and the attention level configured at the hospital, the original transmission control risk level is determined. The original transmission control risk level is then corrected for confidence using the data quality score, the multi-indicator collaborative support, and the historical consistency support to obtain the current transmission control risk level. The medical urgency level is determined based on the current transmission control risk level, transmission trigger prompt score, timeliness level, and the changing trend of participating indicators. The transmission control risk level and medical urgency level are used only for medical data aggregation, packetization, routing, queue scheduling and transmission resource allocation, and are not used to output disease diagnosis conclusions or treatment plans.

4. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 3, characterized in that, Step S2 includes: When the current transmission control risk level or medical urgency reaches the corresponding threshold, the key indicator fields, data generation time, transmission control risk level, medical urgency, data quality score, privacy level, timeliness level and context summary are encapsulated into a core risk package, and at least one of the original monitoring segments, continuous waveforms, historical comparison data and context data of adjacent time periods are encapsulated into an extended evidence package. Configure the same medical semantic unit identifier for the core risk package and the extended evidence package, and write the core risk package identifier to which the extended evidence package belongs, the dependency status, and the maximum allowed completion time into the extended evidence package; Configure the core risk package to upload immediately or in real time, and configure the extended evidence package to be supplemented in real time, trusted caching, asynchronous upload, or batch upload; When the data quality score is lower than the quality threshold, and the multi-indicator collaborative support and historical consistency support are both lower than the corresponding support thresholds, only a summary package or re-collection prompt package with quality markers is generated, without improving the real-time upload level of the complete extended evidence package.

5. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 1, characterized in that, Step S3 includes: Based on the source node, target node, data packet type, upload method, minimum bandwidth, minimum cache space, privacy level, encryption requirements, and maximum allowed completion time of the medical semantic transmission unit, candidate paths and combinations of transmission plans corresponding to each candidate path are generated. The lowest value of the security level of each node and each link in the candidate path is determined as the path security level, and hard constraint screening is performed according to the following conditions: When the path security level is lower than the privacy level of the medical semantic transmission unit, the corresponding combination is eliminated; When the medical semantic transmission unit requires encrypted upload but the candidate path does not support the corresponding encryption method, the corresponding combination is eliminated. When the core risk package is to be uploaded immediately or in real time, and the remaining bandwidth of the bottleneck of the candidate path is lower than the minimum bandwidth, the corresponding combination is eliminated. When the extended evidence package uses trusted caching, asynchronous upload, or batch upload, and there are no trusted cache nodes that meet the privacy level in the candidate path, or the trusted cache remaining space is lower than the minimum cache space, the corresponding combination is removed. When a candidate path and its corresponding transmission plan would cause the extended evidence packet to enter the real-time transmission queue earlier than the core risk packet to which it belongs, the corresponding combination is removed. When all candidate paths and their corresponding transmission plan combinations are eliminated, the core risk packet is written to the local alarm record and retransmission queue, and stored in the trusted cache node without reducing the path security level. The extended evidence packet is written to the delayed upload queue, and the hard constraint screening is re-executed after the network status is restored.

6. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 5, characterized in that, In step S4, the first The medical semantic transmission unit in the first The path cost on each candidate path is calculated using the following formula: ; In the formula, For path cost; For medical emergencies; This is for normalized base transmission delay; To normalize queuing delay; To normalize the risk of packet loss; To normalize the path congestion level; Penalty for mismatches between data volume and remaining path bandwidth or trusted cache capacity; For cache matching degree; to The weights of the corresponding cost items; , , and These are the urgency modulation coefficients for latency, queuing, packet loss, and congestion costs, respectively, and are set or updated based on the transmission feedback of the corresponding network indicators. When the medical semantic transmission unit is the core risk package that is uploaded immediately, Select the candidate path with the lowest path cost as the target path. Among them, the medical urgency level only modulates the network performance cost after passing the hard constraint screening, without changing the path security level, encryption capability, minimum bandwidth and trusted caching conditions.

7. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 6, characterized in that, Step S5 includes: The unmodulated base cost of the target path is calculated using the following formula: ； The scheduling score of the medical semantic transmission unit is calculated according to the following formula: ； In the formula, The base cost of the target path excluding the medical urgency modulation term; The basic transmission priority is determined based on transmission control risk level, timeliness level, privacy level, data quality score, concern level, and data volume. Upload tags immediately; Mark the core risk package; Normalization waiting time; to For queue scheduling weights, Indicates the first Normalized base transmission delay when a medical semantic transmission unit is transmitted via the target path; Indicates the first Normalized queuing delay when a medical semantic transmission unit is transmitted via the target path; Indicates the first Normalized packet loss risk when a medical semantic transmission unit is transmitted via the target path; Indicates the first Normalized congestion level when a medical semantic transmission unit is transmitted via the target path; Indicates the first Penalty for mismatch between the medical semantic transmission unit and the bearer of the target path; Indicates the first The degree of cache matching between each medical semantic transmission unit and the target path; Medical semantic transmission units in the same node queue or link queue are arranged from high to low according to scheduling scores, and data packets with both immediate upload and core risk packet tags are arranged before non-immediate upload data packets. After the scheduling score is nonnegated, transmission resources are allocated according to the proportion of nonnegative scheduling scores for each medical semantic transmission unit. When the transmission resources allocated to the core risk package according to the proportion are lower than the minimum resource proportion of the core risk package, the difference is allocated from the resources of the extended evidence package, asynchronous data upload, or batch data upload to ensure that the transmission resources obtained by the core risk package are not lower than the minimum resource proportion of the core risk package.

8. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 7, characterized in that, The hierarchical transmission in step S5 includes: Core risk packets are sent first according to the queue scheduling results, without interrupting uninterrupted data segments that have already started transmitting; After receiving feedback that the core risk package has been successfully sent, the extended evidence package with the same medical semantic unit identifier as the core risk package is added to the real-time sending queue; or after the receiving end generates a supplementary evidence request, the extended evidence package pointed to by the supplementary evidence request is added to the real-time sending queue. If the core risk package is not successfully sent and no request for supplementary evidence is received, the corresponding extended evidence package will be stored in a trusted cache node that meets the privacy level requirements, or kept in the delayed upload queue. When network congestion decreases, the core risk packet is transmitted, or there are remaining resources for real-time transmission, the extended evidence packet is sent. For data whose transmission control risk level and medical urgency are both below the corresponding thresholds, perform summary upload, compressed upload, or batch upload.

9. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 8, characterized in that, Step S6 includes: Record the successful transmission status, actual end-to-end latency, actual queuing latency, actual packet loss rate, retransmission count, supplementary evidence request flag, and view confirmation flag of the core risk packet according to medical semantic category; Update the following formula: The proportion of extended evidence uploaded in real time corresponding to medical semantics: ； In the formula, and These represent the real-time upload ratios of extended evidence for the current closed-loop cycle and the next closed-loop cycle, respectively. This refers to the amount of attention feedback obtained by aggregating the requests for supplementary evidence and confirmation tags. Normalized actual end-to-end latency for high-emergency medical core risk packages; The normalized actual packet loss rate for the core risk package of high medical urgency; To improve the success rate of sending core risk packages; to This is the adjustment coefficient; and These represent the lower and upper limits of the proportion of extended evidence uploaded in real time, respectively. This indicates that the calculation results will be restricted between the lower and upper limits; When the actual end-to-end latency of the core risk package exceeds the maximum allowed completion time, or when the success rate of sending the core risk package is lower than the success rate threshold, the control priority of reducing the real-time upload ratio of extended evidence is higher than the control priority of increasing the real-time upload ratio of the feedback.

10. The method for precise routing and closed-loop transmission of medical data based on patient risk status according to claim 9, characterized in that, Step S6 also includes: Update the network congestion status for the next closed-loop cycle based on the actual congestion status, actual queue status, and actual packet loss rate of the current closed-loop cycle. Based on the actual end-to-end latency, actual queuing latency, actual packet loss rate, and transmission performance of the high medical urgency core risk package, the urgency modulation coefficient and path cost weight are updated respectively. Based on the average waiting time and transmission success rate of the core risk package, update the waiting time weight, the unmodulated basic cost weight, and the minimum resource ratio of the core risk package; Based on the updated network congestion status, transmission control risk status, and trends in key participation indicators, adjust the dynamic time window for the next closed-loop cycle. Each transmission control parameter is set with a preset initial value, update step size, and upper and lower limits. When the actual end-to-end delay, actual packet loss rate, and transmission success rate of the core risk packet all meet the corresponding target conditions within multiple consecutive closed-loop cycles, the update step size is reduced or the current transmission control parameter is maintained.

Images