A hospital clinical information system based on the Internet of Things

Abstract

This invention relates to the fields of medical Internet of Things (IoT) and hospital information technology, and provides an IoT-based hospital clinical information system, including a wireless ECG terminal node, a ward edge gateway, and a clinical information system. The wireless ECG terminal node is used to acquire raw ECG waveforms and send ECG summaries. The ward edge gateway is used to generate abnormal trend intensity and generate a set of backtracking control parameters, including forward backtracking time windows, backward backtracking time windows, upload priority, and upload quality control parameters. It receives raw ECG waveform segments uploaded by the terminal node and generates a trusted access score based on mapping credibility, event consistency, and waveform consistency. When the trusted access score meets the access threshold, a structured monitoring event object is generated. The clinical information system is used to receive this object. This invention reduces the occupancy of wireless resources in the ward and the power consumption of the terminal, and improves the protection capability of high-value anomalies and the credibility of accessed data.

Description

Technical Field

[0001] This invention relates to the fields of medical Internet of Things (IoT), clinical monitoring data processing, and hospital information technology, specifically to a hospital clinical information system based on the Internet of Things (IoT). Background Technology

[0002] Currently, in clinical settings such as general inpatient wards, postoperative observation wards, and intensive care units, an increasing number of patients are using wireless ECG patches for continuous ECG monitoring. Wireless ECG patches typically continuously acquire the patient's raw ECG waveforms and transmit the monitoring data to a higher-level system via the ward's wireless network for use by clinical information systems, nursing systems, ward workstations, or early warning modules.

[0003] In existing technologies, the following are the main technical approaches for wireless continuous electrocardiogram monitoring:

[0004] The first type is the continuous full-data upload route. Terminal nodes continuously upload raw ECG waveforms, which are then analyzed, displayed, and stored centrally by the server. While this solution ensures data integrity, it has significant drawbacks in actual ward deployment. A ward typically has multiple wireless ECG patch nodes simultaneously connected to the same wireless access point or the same ward edge gateway. When multiple nodes continuously upload raw waveforms, it consumes a large amount of wireless bandwidth and air interface time slots, easily leading to increased wireless link occupancy, increased latency for uploading critical abnormal waveforms, and increased queue length. During peak periods, packet loss, retransmission, or upload failures may even occur, affecting the stability of other wireless medical services in the ward. Furthermore, wireless ECG patches are usually powered by small batteries. If raw waveform uploads are continuous, the RF module will operate at a high duty cycle for extended periods, shortening node battery life and requiring frequent maintenance or equipment replacement, increasing the burden on medical staff.

[0005] The second type involves uploading data after initial feature extraction at the terminal node. The terminal node extracts some ECG features locally and uploads these features along with a portion of the original waveform to the host platform for anomaly identification. While this approach reduces the amount of data transmitted to some extent, it still suffers from the following problems: the terminal node's feature extraction capability is limited by its computing resources and power consumption, making it difficult to execute complex ECG analysis algorithms; the accuracy and completeness of feature extraction rely on preset algorithms and cannot adapt to individual differences among patients; and when an anomaly is detected, it often requires re-requesting the original waveform, increasing system complexity and response latency.

[0006] The third type is the backhaul route after anomaly detection at the edge. Terminal nodes periodically upload feature data. Upon detecting an anomaly, the edge gateway or server triggers the terminal to re-upload the original waveforms before and after the anomaly. This approach reduces the resource pressure of continuous full uploads to some extent, but its anomaly triggering logic is often too simplistic, typically only related to a single feature threshold, anomaly probability threshold, or anomaly classification result. In other words, as long as an anomaly is detected, a backhaul is triggered, treating all anomalies the same. This mechanism cannot differentiate the backhaul priority between high-risk and ordinary patients, cannot determine whether a backhaul is worthwhile based on the current ward resource status, and cannot dynamically adjust the required original waveform context according to different anomaly types. Therefore, although existing solutions introduce conditional uploading, they are essentially still relatively crude binary control.

[0007] Furthermore, in actual clinical use, there are issues such as waveform disturbances caused by bed changes, re-adhesion, re-binding after detachment, and nursing operations like turning over, lead adjustments, and skin cleaning. Short-term motion artifacts may be misidentified as abnormal rhythmic trends, and there may be time misalignments or inconsistencies in interpretation between node summaries and the original waveform. If the transmitted waveform enters the clinical information system directly without further trusted access control, it can easily lead to problems such as incorrectly bound data entering formal records, false anomalies related to nursing disturbances entering medical records or monitoring records, and low-reliability data contaminating the clinical database. Simultaneously, inconsistencies in the time base between the edge and endpoint collaboration can cause the backtracking window and event comparison to fail, affecting system reliability. Summary of the Invention

[0008] To address the technical problems of existing technologies, such as limited wireless resources in wards, link congestion and high node power consumption caused by continuous uploading of raw waveforms, and the simplistic abnormal triggering logic that fails to reflect differences in clinical value, this application proposes an Internet of Things-based hospital clinical information system. This system enables continuous joint control of raw waveform backtracking and its access process to the clinical information system under the condition that multiple nodes in the ward share limited wireless resources.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A hospital clinical information system based on the Internet of Things (IoT) includes interconnected wireless ECG terminal nodes, ward edge gateways, and a clinical information system. The wireless ECG terminal nodes continuously acquire raw ECG waveforms from patients, store these waveforms in a local circular buffer, and periodically send ECG summaries extracted from the waveforms. They generate anomaly trend intensity based on the ECG summaries and perform anomaly value determination based on the anomaly trend intensity, patient priority, and ward resource status. When the anomaly value score meets the backtracking trigger condition, a backtracking control parameter set is generated, including a forward backtracking time window, a backward backtracking time window, upload priority, and upload quality control parameters, and sent to the corresponding wireless ECG terminal node. The system includes a wired ECG terminal node; a wireless ECG terminal node; a receiver that locks and uploads raw ECG waveform segments from the local circular buffer according to the backtracking control parameter set; and upon receiving the raw ECG waveform segments, a trusted access score is generated based on patient-bed-device mapping reliability, consistency of nursing or medical order events, and consistency of uploaded waveforms. When the trusted access score meets the access threshold, a structured monitoring event object is generated and sent to the clinical information system. The structured monitoring event object includes at least a portion of patient identifier, bed identifier, event time, abnormality type, raw waveform location information, and trusted access score. The clinical information system receives the structured monitoring event object.

[0011] The above solution reduces the wireless resource consumption and terminal power consumption caused by continuous full uploads by having the terminal only upload abstracts and only transmit the original waveforms for high-value anomalies. At the same time, through the anomaly value judgment and retrospective parameter adaptive generation mechanism, the system prioritizes the preservation of more complete or higher quality waveform context for high-value anomalies under resource-constrained conditions. Furthermore, through the trusted access judgment mechanism, the risk of low-trust or pseudo-anomaly data contaminating the clinical system is reduced.

[0012] Preferably, the wireless ECG terminal node includes: an ECG acquisition module for continuously acquiring the patient's original ECG waveform; a local circular buffer storage module for continuously storing the most recently preset duration original ECG waveform in a circular buffer manner; a summary extraction module for extracting ECG summaries based on the original ECG waveform; and an upload control module for periodically sending the ECG summaries under normal monitoring conditions, and extracting and uploading the corresponding original ECG waveform segments after receiving the backtracking control parameter set.

[0013] This preferred solution implements local caching and controlled upload functions for terminal nodes through modular design, providing the hardware and logical foundation for the backtracking mechanism.

[0014] Preferably, the ECG summary includes at least one or a combination of the following: mean heart rate; mean RR interval; RR interval dispersion; QRS width variation; waveform quality index.

[0015] This preferred scheme, through multi-dimensional summary features, can more accurately reflect the patient's electrocardiographic status, providing a reliable basis for subsequent abnormal trend calculations.

[0016] Preferably, the ward edge gateway generates an abnormal trend intensity based on the ECG summary, and generates an abnormal value score based on the abnormal trend intensity, the patient priority, and the ward resource status, so as to determine whether to initiate the original ECG waveform segment backtracking based on the comparison result between the abnormal value score and the preset trigger threshold; when the abnormal value score does not reach the preset trigger threshold but meets the additional trigger correction rule, the ward edge gateway still initiates the original ECG waveform segment backtracking.

[0017] This preferred solution achieves accurate determination of whether an anomaly is worth backtracking by quantifying anomaly value scoring, avoiding the blindness of simple threshold triggering.

[0018] Preferably, during the generation of the abnormal trend intensity, at least one summary feature is normalized. The normalization process includes at least one of the following methods: normalization based on a preset reference range; or dynamic normalization based on an individual baseline established during the initial stable monitoring phase after the patient accesses the system.

[0019] This preferred approach reduces the impact of individual patient differences on the calculation of abnormal trends and improves the accuracy of the determination through normalization, especially dynamic normalization based on individual baselines.

[0020] Preferably, the ward edge gateway generates the backtracking control parameter set based on the anomaly category, patient priority, ward resource status, and available cache duration of the local circular buffer. The patient priority is a preset patient risk level, and a correction value is added to the patient priority when the patient is in the postoperative key observation period. The forward backtracking time window and the backward backtracking time window are dynamically determined according to the anomaly category, the upload priority is dynamically determined according to the patient priority, and the upload quality control parameters are dynamically determined according to the ward resource status.

[0021] This preferred approach enables adaptive generation of backtracking parameters, ensuring that backtracking behavior matches the type of abnormality, patient status, and resource availability, thereby optimizing resource utilization efficiency.

[0022] Preferably, the wireless ECG terminal node determines the starting position of reading in the local circular buffer according to the forward backtracking time window, and continues to collect and upload subsequent raw ECG waveforms according to the backward backtracking time window. When entering the backtracking upload state, the corresponding buffer segment is temporarily locked to prevent the raw ECG waveforms that have not been uploaded from being overwritten by the newly sampled data.

[0023] This preferred scheme ensures the integrity and accuracy of the original waveform backtracking through a circular buffer pointer control and locking mechanism.

[0024] Preferably, a time synchronization and delay compensation module is also included. The time synchronization and delay compensation module is used to enable bidirectional time synchronization between the wireless ECG terminal node and the ward edge gateway to determine the clock deviation and link round-trip delay. When the determined clock deviation exceeds a preset threshold, the wireless ECG terminal node performs local time correction.

[0025] This preferred solution ensures the feasibility and reliability of original waveform backtracking and event comparison through end-edge time synchronization and delay compensation.

[0026] Preferably, the wireless ECG terminal node further includes a link priority execution module, which is used to set the access priority parameters of the underlying wireless link according to the upload priority, so as to improve the access priority of high-priority raw ECG waveform segments in wireless channel contention.

[0027] This preferred approach maps upload priority to the underlying wireless link, enhancing the access priority of high-value abnormal waveforms in the ward wireless network.

[0028] Preferably, when the wireless ECG terminal node communicates with the ward edge gateway via a wireless local area network based on IEEE 802.11, the link priority execution module is used to map the upload priority to the WMM access category of the media access control layer, wherein a high-priority original ECG waveform segment corresponds to a higher-priority access category, and periodically sent ECG summaries correspond to a best-effort access category.

[0029] This preferred scheme specifies the priority mapping method, utilizes the characteristics of standard protocols to realize differentiated services, and further ensures the transmission quality of critical data.

[0030] Beneficial effects:

[0031] This invention provides a hospital clinical information system based on the Internet of Things, which has the following advantages compared with the prior art:

[0032] (1) Reduce wireless resource usage and terminal node communication power consumption in the ward. This invention avoids the waste of wireless resources caused by continuous full upload by using a mechanism where terminals only upload summaries and only transmit the original waveform for high-value anomalies. Under normal monitoring conditions, terminal nodes only periodically send ECG summaries with small data volumes, which significantly reduces the wireless channel occupancy time and RF module working time, thereby extending the terminal node's battery life and reducing the load pressure on the ward's wireless network.

[0033] (2) Achieving accurate determination of anomaly value and optimizing resource allocation efficiency. This invention generates a quantitative anomaly value score by comprehensively considering the intensity of the anomaly trend, patient priority, and ward resource status through an anomaly value determination mechanism, thereby screening out anomalies that are truly worth occupying high data transmission resources. This mechanism avoids the blindness of simple threshold triggering, can distinguish the transmission priority of high-risk patients and ordinary patients, and determines whether it is worthwhile to transmit based on the current ward resource status, thus achieving the ability to protect high-value anomalies under limited wireless resources.

[0034] (3) Dynamically and adaptively generating backtracking parameters, prioritizing the retention of high-value waveform contexts. This invention utilizes a backtracking parameter adaptive generation mechanism to dynamically determine the forward and backward backtracking time windows based on the abnormality category, dynamically determine the upload priority based on patient priority, and dynamically determine the upload quality control parameters based on the ward resource status. This mechanism ensures that backtracking behavior matches the abnormality type, patient status, and resource availability, prioritizing the retention of waveform contexts with higher review value and optimizing resource utilization efficiency.

[0035] (4) Improve the credibility of accessed data and reduce the risk of low-credibility data contamination. This invention uses a trusted access determination mechanism, combining the credibility of patient-bed-equipment mapping, consistency of nursing or medical order events, and consistency of uploaded waveforms, to control the eligibility of transmitted waveforms to enter the formal clinical data link. This mechanism can effectively identify and filter low-credibility data caused by equipment bed changes, nursing operations, motion artifacts, etc., reducing the probability of misbinding, nursing disturbances, or inconsistent waveform data directly entering the clinical information system, and improving the credibility of clinical accessed data.

[0036] (5) Ensuring the feasibility and reliability of original waveform backtracking and event comparison. This invention maintains time base consistency between the wireless ECG terminal node and the ward edge gateway through end-edge time synchronization, link delay compensation, and time tolerance rules, ensuring the accuracy of the original waveform backtracking window and the effectiveness of event comparison. This mechanism is particularly important in actual ward environments, as it can overcome the influence of factors such as wireless transmission delay and clock drift, ensuring the reliable operation of the system.

[0037] (6) Enhancing the access priority of high-value abnormal waveforms in the ward wireless network. This invention maps the upload priority to the access priority parameters of the underlying wireless link through the link priority execution module. In particular, in IEEE 802.11-based wireless LANs, the upload priority is mapped to the WMM access class of the Media Access Control layer. This mechanism utilizes the characteristics of standard protocols to achieve differentiated services, further ensuring the transmission quality of critical data and reducing the upload latency and packet loss rate of high-value abnormal waveforms. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the overall architecture of a hospital clinical information system based on the Internet of Things provided in an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of the module structure of the wireless ECG terminal node provided in an embodiment of the present invention;

[0040] Figure 3 This is a schematic diagram of the process for determining abnormal value and generating backtracking control parameters in the ward edge gateway provided in an embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of the trusted access determination process provided in an embodiment of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0044] Example 1:

[0045] like Figure 1As shown, this embodiment provides a hospital clinical information system based on the Internet of Things (IoT). The system includes a wireless ECG terminal node, a ward edge gateway, and a clinical information system. The wireless ECG terminal node, ward edge gateway, and clinical information system establish physical connections and logical communication relationships through wired or wireless networks. Specifically, the wireless ECG terminal node is worn on the patient's body surface and interacts with the ward edge gateway deployed within the ward via the ward's wireless local area network. The ward edge gateway connects to the clinical information system through the hospital's internal wired network. The clinical information system can be a hospital information system (HIS), an electronic medical record system (EMR), or a dedicated monitoring data platform.

[0046] The wireless ECG terminal node is used to continuously acquire the patient's original ECG waveform, store the original ECG waveform in a local circular buffer, and periodically send ECG summaries extracted based on the original ECG waveform.

[0047] Specifically, under normal monitoring conditions, the wireless ECG terminal node does not continuously upload high-volume raw ECG waveforms. Instead, it caches the waveforms locally and sends low-volume ECG summaries to the ward edge gateway at preset intervals (e.g., every 5 seconds). This design significantly reduces the radio frequency transmission duty cycle of the wireless ECG terminal node, thereby extending the node's battery life and reducing the bandwidth consumption of the ward's wireless network. The ECG summary is low-dimensional data obtained after feature extraction from the raw ECG waveform. It can reflect the patient's current ECG status trend, but the data volume is much smaller than the raw waveform.

[0048] The ward edge gateway is used to receive the ECG summary, perform anomaly value determination based on anomaly trends, patient priorities, and ward resource status; when the anomaly value score meets the backtracking trigger condition, it generates a backtracking control parameter set including a forward backtracking time window, a backward backtracking time window, upload priority, and upload quality control parameters, and sends it to the corresponding wireless ECG terminal node; it receives the original ECG waveform segment locked and uploaded by the wireless ECG terminal node from the local circular buffer according to the backtracking control parameter set; and after receiving the original ECG waveform segment, it generates a trusted access score based on patient-bed-equipment mapping credibility, nursing or medical order event consistency, and uploaded waveform consistency, and generates a structured monitoring event object when the trusted access score meets the access threshold and sends it to the clinical information system.

[0049] Specifically, the ward edge gateway is the core control node in this embodiment, executing two core mechanisms: "on-demand backtracking" and "trusted access." In the "on-demand backtracking" mechanism, the ward edge gateway no longer judges anomalies solely based on a single threshold. Instead, it comprehensively considers the intensity of the abnormal trend, patient priority (e.g., high-risk or ordinary patients), and the current ward's wireless resource status (e.g., bandwidth occupancy) to calculate an anomaly value score. Only when this score indicates that the anomaly event has sufficient backtracking value and backhaul resources allow it will the original waveform backtracking be triggered. The backtracking control parameter set consists of fine-grained control instructions issued by the ward edge gateway to the terminal nodes. The forward and backward backtracking time windows define the waveform time range to be backtracked, the upload priority determines the data's competitiveness in the wireless channel, and the upload quality control parameters dynamically adjust the data volume based on resource status. In the "trusted access" mechanism, after receiving the backtracked original waveform, the ward edge gateway does not directly write it into the clinical information system but further verifies its trustworthiness. For example, it checks whether the binding relationship between the device and the patient has been confirmed by the nurse, whether there were any nursing disturbances such as turning over or lead adjustments near the time of the abnormal event, and whether the uploaded waveform is consistent with the characteristics of the previous summary. Only after passing these verifications will structured clinical event objects be generated, ensuring that the data entering the clinical information system has high clinical credibility.

[0050] The clinical information system is used to receive the structured monitoring event objects.

[0051] Specifically, the clinical information system is the final destination of data, used to store, display and manage high-quality ECG events that have been screened and verified by the ward edge gateway, providing medical staff with monitoring data display, review and processing references.

[0052] Through the above scheme, this embodiment constructs a complete closed-loop process of "summary upload - abnormal value determination - backtracking control - waveform upload - trusted access". Compared with the continuous full upload or simple threshold-triggered backtracking in the prior art, this embodiment realizes on-demand allocation of resources through abnormal value determination and ensures the access reliability of the data to the database through trusted access determination. Thus, under the condition of limited wireless resources in the ward, it realizes continuous joint control of the backtracking of the original ECG waveform and the access admission process.

[0053] Example 2:

[0054] This embodiment, based on Embodiment 1, provides a detailed explanation of the internal structure of the wireless ECG terminal node and the specific implementation method for summary extraction. For example... Figure 2As shown, the wireless ECG terminal node, worn on the patient's body surface, typically takes the form of a wireless ECG patch. To achieve continuous acquisition, local caching, and controlled uploading, the node integrates an ECG acquisition module, a local circular buffer storage module, a summary extraction module, and an upload control module.

[0055] The ECG acquisition module is used to continuously acquire the patient's raw ECG waveforms. This module typically includes an ECG analog front-end chip and an analog-to-digital converter, capable of converting the ECG signal from the patient's body surface into digital waveform data at a preset sampling rate. In a preferred embodiment, the sampling rate is set to 250Hz to meet clinical needs for observing ECG waveform details; in other embodiments, depending on different clinical scenarios or power budgets, the sampling rate can also be set to 125Hz, 500Hz, or other values. The acquired raw ECG waveform data is transmitted in real-time to a local ring buffer storage module.

[0056] The local circular buffer storage module is used to continuously store the most recent raw ECG waveforms of a preset duration in a circular buffering manner. This module typically consists of a contiguous storage space composed of volatile memory (such as SRAM or DRAM) or non-volatile memory (such as Flash). The core logic of the circular buffering method lies in the "write pointer increment logic": Assuming the total capacity of the circular buffer is N sampling points, and the current write pointer position is... Each time a new sampling point is written, the write pointer will move according to the formula. The buffer increments sequentially. When the write pointer reaches the end of the buffer, it automatically wraps back to the beginning of the buffer, overwriting the oldest historical data. This mechanism ensures that the buffer always contains the waveform data most recent to the current moment. For example, when the buffer capacity is set to 60 seconds, the system always retains the original waveform of the most recent 60 seconds, providing a data basis for subsequent backtracking.

[0057] To support the backtracking function, the local ring buffer storage module must also have a "temporary locking mechanism." Under normal monitoring conditions, the buffer allows new data to overwrite old data. However, when the upload control module receives the backtracking control parameter set from the ward edge gateway and enters the backtracking upload state, the system temporarily locks the segment in the buffer that is about to be read and uploaded. Specifically, the system marks the start and end addresses of the segment. During the locking period, although the write pointer continues to increment, once the write pointer moves to the locked segment, the write operation will be temporarily suspended, or the locked segment will be skipped and written to a subsequent location (if storage space allows), to prevent the incomplete upload of the original ECG waveform from being overwritten by new sampled data. This mechanism is a key defense point for ensuring the integrity of the backtracking waveform, ensuring that even if new ECG data is generated during the backtracking upload process, the historical waveform segment being uploaded will not be corrupted.

[0058] The summary extraction module is used to extract ECG summaries based on the original ECG waveform. This module is typically implemented by a microcontroller (MCU) running specific signal processing algorithms. The ECG summary is low-dimensional data obtained by feature compression of the original waveform; its data volume is much smaller than the original waveform, making it suitable for periodic transmission under normal conditions to reduce wireless resource consumption. The ECG summary includes at least one or a combination of the following: average heart rate, average RR interval, RR interval dispersion, QRS width variation, and waveform quality indicators.

[0059] Mean heart rate reflects the average frequency of a patient's heartbeat and is a fundamental indicator for diagnosing tachycardia or bradycardia. Mean RR interval reflects the average time interval between two adjacent R waves and is the reciprocal of mean heart rate. RR interval dispersion reflects the regularity of the heart rhythm; this indicator is highly sensitive in identifying arrhythmias such as atrial fibrillation, and a significant increase in RR interval dispersion often suggests a tendency towards arrhythmia. QRS width variation reflects changes in the ventricular depolarization process and is valuable for identifying ventricular arrhythmias or conduction blocks. Waveform quality indicators are used to assess signal purity, such as the presence of baseline drift, power line interference, or lead dropout, providing a confidence reference for subsequent anomaly assessment. Through the combination of these multidimensional features, the abstract extraction module can generate low-volume abstract packages containing rich clinical information, providing reliable data support for anomaly trend analysis at the ward edge gateway.

[0060] The upload control module periodically sends ECG summaries during normal monitoring and extracts and uploads the corresponding original ECG waveform segments upon receiving the backtracking control parameter set. During normal monitoring, the upload control module packages and sends the ECG summaries generated by the summary extraction module according to a preset upload cycle (e.g., every 5 seconds). At this time, the radio frequency module operates at a low duty cycle, effectively reducing node power consumption. When the backtracking control parameter set is received from the ward edge gateway, the upload control module parses the forward and backward backtracking time windows in the parameter set, calculates the starting position of the circular buffer based on the forward time window, and activates a temporary locking mechanism, controlling the radio frequency module to upload the locked original waveform segments with higher priority. Through the coordinated work of these modules, the wireless ECG terminal node achieves a complete functional closed loop from continuous acquisition and local caching to on-demand backtracking, reducing resource consumption during normal operation while ensuring the complete acquisition of critical waveform data in case of anomalies.

[0061] Example 3:

[0062] This embodiment, based on Embodiment 1, provides a detailed explanation of the specific mechanism for anomaly value determination performed by the ward edge gateway. For example... Figure 3As shown, after receiving the ECG summaries periodically sent by the wireless ECG terminal nodes, the ward edge gateway does not directly determine whether there is an abnormality based on a single threshold. Instead, it first generates an abnormal trend intensity based on the ECG summaries. .

[0063] In generating abnormal trend intensity Normalizing the abstract features is a crucial step in the process. This embodiment provides two normalization methods to adapt to different clinical application scenarios.

[0064] The first approach is normalization based on a preset reference range. This is a general and standardized approach. Specifically, for a given feature X, the min-max normalization method can be used: .in, and These are the preset lower and upper limits of the reference range for this feature in the statistical distribution of the normal population. For example, for average heart rate, It can be set to 60 times / minute. The heart rate can be set to 100 beats per minute. This method is simple to calculate and suitable for initial consultations where patient history data is lacking. However, its drawback is that it ignores individual physiological differences among patients. For example, athletes who engage in endurance sports for a long time may maintain a resting heart rate of around 50 beats per minute year-round. If the above global reference range is used, the system may continuously classify this as abnormal, leading to false alarms.

[0065] To address the aforementioned issues, this embodiment provides a second preferred normalization method: dynamic normalization based on the individual baseline established during the initial stable monitoring phase after the patient connects to the system. This method effectively eliminates the influence of individual patient differences and improves the accuracy of abnormality detection. Specifically, when a patient first connects to the system wearing the wireless ECG terminal node, the system automatically identifies a waveform whose quality meets preset conditions (e.g., An initial stable monitoring phase (above a threshold), such as the first 10 minutes after access. During this phase, the system calculates individual baseline values ​​for various patient characteristics. and the allowed range of changes The normalization formula is adjusted as follows: .in, The value can be the difference between the maximum and minimum values ​​of the feature within the baseline window, or a certain percentage range of the baseline average. Taking the aforementioned athlete as an example, their individual baseline... It could be 50 times per minute. The normalized value is likely 10 beats per minute. When the heart rate is maintained between 45 and 55 beats per minute, the normalized value will be at a low level, and the system will judge it as normal. Only when the heart rate suddenly rises above 70 beats per minute, significantly deviating from the individual's baseline, will the normalized value increase, thus triggering an anomaly judgment. This dynamic normalization based on the individual's baseline makes the calculation of the intensity of abnormal trends more closely reflect the patient's actual physiological state, significantly reducing the false alarm rate and demonstrating the system's level of intelligence.

[0066] After calculating the abnormal trend intensity Vabn, the ward edge gateway further incorporates patient priority. and ward resource status Generate anomaly value scores This step is the core logic behind the "on-demand backtracking" achieved in this invention. Specifically, it involves anomaly value scoring. The following weighted model can be used: .in, The strength of the abnormal trend after smoothing; The patient priority score can be determined by at least one of the following: nursing grade, medical order grade, postoperative observation status, and early warning level. For example, high priority patients can be set to 1.0, medium priority patients to 0.7, and ordinary patients to 0.4. If the patient is in the postoperative critical observation period, a correction value can be added. The score represents the resource usage in the ward area, reflecting the current level of congestion on the wireless network. For example, low usage is 0.2 and high usage is 0.8. , , These are weighting coefficients, which can be set to, for example, 0.50, 0.35, and 0.15. The physical meaning of this formula is: the stronger the abnormal trend and the higher the patient priority, the greater the value of retrospective analysis; conversely, the higher the ward resource consumption, the greater the cost of retrospective analysis. Only when... Exceeding the preset trigger threshold The system only determines that the preset conditions are met when the value is 0.55 (e.g., 0.55), and then initiates the original ECG waveform segment retrospective. This determination mechanism is no longer a simple "abnormality triggers retrospective", but a "value-driven" decision that comprehensively considers clinical value and resource costs, ensuring that valuable transmission bandwidth is allocated to abnormal events with high clinical value in the context of limited wireless resources in the ward.

[0067] Example 4:

[0068] Based on Examples 1 to 3, this embodiment provides a detailed explanation of the specific logic for the ward edge gateway to generate the backtracking control parameter set and the specific process for the wireless ECG terminal node to perform backtracking.

[0069] Specifically, when the anomaly value score Sval exceeds a preset trigger threshold, and the ward edge gateway determines that it needs to initiate a retrospective analysis of the original ECG waveform segment, it does not simply send an "upload" command to the terminal. Instead, it generates a retrospective control parameter set containing refined control parameters. The generation of this parameter set is a dynamic decision-making process coupled with multiple factors. The ward edge gateway generates the retrospective control parameter set based on the anomaly category, patient priority, ward resource status, and the available buffer duration of the local circular buffer.

[0070] The forward and backward lookback time windows are dynamically determined based on the abnormality category. Different abnormality categories require different waveform context durations for clinical interpretation. For example, for suspected abnormal rhythm trends, due to the need to observe long-term irregularities in the RR interval, both the forward and backward lookback time windows can be dynamically set to 30 seconds; for suspected short bursts of rapid rhythm trends, due to their sudden onset, both the forward and backward lookback time windows can be set to 20 seconds; while for trends of poor lead contact, a shorter time window is often sufficient for assessment, and both the forward and backward lookback time windows can be set to 10 seconds. Furthermore, if the patient has a high priority (e.g., Pclin ≥ 0.8), the system can appropriately increase the lookback time window length to retain more complete clinical evidence; if ward resources are strained (e.g., Rnet ≥ 0.8), the system can appropriately shorten the time window length to reduce data transmission volume. Upload priority is dynamically determined based on patient priority. Waveform feedback from high-risk patients will be given the highest priority to ensure priority transmission opportunities in wireless channel contention. Upload quality control parameters are dynamically determined based on the ward's resource status. When ward resources are plentiful, full-precision uploads are used to preserve the original waveform details; when ward resources are scarce, reduced-precision uploads (e.g., downsampling) or high compression levels are used to reduce data volume while preserving key morphological features, preventing channel congestion. This dynamic generation mechanism ensures a precise match between retrospective actions and clinical needs and resource availability, achieving intelligent control of "on-demand retrospectives."

[0071] After receiving the set of retrospective control parameters from the ward edge gateway, the wireless ECG terminal node will parse the parameter content and perform the corresponding waveform locking and uploading operations.

[0072] The most crucial step is determining the start position for reading within the local circular buffer based on the forward backtracking time window. Let the terminal's current sampling rate be... The circular buffer has a total capacity of N sampling points, and the current write pointer position is... The forward retrospective time window is The number of forward sample points that need to be backtracked is... Read the start pointer The calculation formula is: The "+N" operation in this formula is a key defensive design feature used to handle write pointers. Less than The pointer wraps around the scene. For example, suppose F_s = 250Hz, =30s, then =7500. If the current write pointer... =3000, calculate directly This will result in a negative value of -4500, which is meaningless in memory addressing. By adding the total capacity N (assuming N = 15000), the calculation becomes... This ensures that the correct data segment is pointed to at the end of the buffer. The terminal node then... Initially, waveform data within the forward window is read sequentially. Simultaneously, subsequent raw ECG waveforms are collected and uploaded according to the backward time window. This means that the terminal not only uploads historical cached data but also continues to collect and upload real-time waveforms for a period of time (e.g., 30 seconds) after the current moment, thus forming a complete context fragment of the abnormal event.

[0073] To ensure data integrity during the backtracking upload process, the wireless ECG terminal node temporarily locks the corresponding buffer segment when entering the backtracking upload state to prevent the original ECG waveform that has not been uploaded from being overwritten by newly sampled data.

[0074] Specifically, under normal monitoring conditions, the circular buffer allows new data to overwrite the oldest data. However, during backtracking uploads, if the backward time window is long, the write pointer may catch up with and overwrite the forward window data that has not yet been uploaded. A temporary locking mechanism temporarily freezes write access to the buffer segment being uploaded by marking that segment. When the write pointer increments to the boundary of the locked area, the write operation is temporarily suspended, or the new data is redirected to a backup buffer (if hardware resources allow), until the waveform data segment is uploaded and unlocked. This mechanism effectively prevents critical waveform segments from being corrupted during transmission, ensuring that the uploaded waveform segments are continuous and complete. Through the dynamic generation of the aforementioned backtracking control parameter set and the precise execution by the terminal, this embodiment achieves differentiated waveform backtracking for abnormal events of different value levels under limited wireless resource conditions, avoiding the waste of resources from full uploads while ensuring the complete acquisition of high-value abnormal data.

[0075] Example 5:

[0076] This embodiment, based on the above embodiments, provides a detailed description of the system's time synchronization and latency compensation mechanism. In wireless continuous ECG monitoring scenarios, the terminal node summary formation time, raw waveform buffering time, edge gateway processing time, and nursing or medical order event recording time must be comparable. If there is significant clock drift between the terminal node and the ward edge gateway, or if wireless link transmission latency is not handled, it will cause the abnormal triggering point to be misaligned with the local waveform window, resulting in inaccurate forward or backward time window truncation of the uploaded waveform, severely affecting waveform verification and event comparison. Therefore, the system in this embodiment also includes a time synchronization and latency compensation module.

[0077] Specifically, the time synchronization and latency compensation module enables bidirectional time synchronization between the wireless ECG terminal node and the ward edge gateway to determine clock skew and link round-trip latency. When the determined clock skew exceeds a preset threshold, the wireless ECG terminal node performs local time correction. The bidirectional time synchronization process is achieved through… This is achieved using four time markers, from T_4 to T_4. Step S501: The wireless ECG terminal node, at its local time... A synchronization request is sent to the ward edge gateway. In step S502, upon receiving the synchronization request, the ward edge gateway records the gateway time. Step S503: The ward edge gateway records the gateway time when sending a synchronization response. Step S504: The wireless ECG terminal node records the local time upon receiving the synchronization response. Using these four timestamps, the terminal node can independently calculate the link round-trip time. and clock deviation The derivation process is as follows: Link round-trip delay It equals the total time difference minus the gateway processing time, i.e. Clock skew Then take the average of the request path deviation and the response path deviation, that is... This two-way mechanism can effectively eliminate the errors caused by the unknown delay in one-way transmission, and is more accurate than one-way broadcast synchronization.

[0078] After calculating the clock deviation Then, the system will determine whether its absolute value exceeds a preset threshold. In a preferred embodiment, a preset threshold is used. Set to 100 milliseconds. When At this time, the wireless ECG terminal node performs local time correction. The correction strategy can be to directly adjust the local time forward or backward. Alternatively, the local clock's running rate can be adjusted to gradually correct the deviation. To prevent frequent time jumps at the terminal due to wireless link jitter, which could affect the continuity of data recording, this embodiment further employs a smoothing strategy. Specifically, the terminal node adjusts the local clock's running rate within several consecutive synchronization cycles. Perform a moving average process, such as calculating the average of the most recent three periods. Only when Time adjustments are only performed when the timing is right. This design enhances the system's robustness while ensuring time synchronization accuracy.

[0079] This time synchronization and delay compensation mechanism plays a crucial supporting role in ensuring the accuracy of the backtracking window time. As described in the aforementioned embodiments, the wireless ECG terminal node needs to adjust the backtracking time window accordingly. Calculate the start position of the read operation in the circular buffer. If there is a significant clock discrepancy between the terminal and the gateway, such as the terminal clock lagging behind the gateway clock, the moment the gateway determines an anomaly will correspond to an earlier time point on the terminal's timeline. Without correction, the terminal will capture incorrect waveform segments, causing a mismatch between the retrospective waveform and the actual time of the anomaly. This embodiment's bidirectional time synchronization mechanism ensures the consistency of the time base between the terminal node and the ward edge gateway, thereby guaranteeing the accurate implementation of the time window parameters in the retrospective control parameters set, enabling the original waveform segments from the retrospective to truly reflect the ECG state before and after the anomaly occurred.

[0080] It should be understood that the aforementioned two-way time synchronization process is implemented through the software protocol layer. In other implementations, time synchronization between the terminal node and the ward edge gateway can also be achieved using the hardware time base, connection event timestamp, or beacon frame time information provided by the underlying wireless communication protocol, as long as the consistency of the time bases at both ends can be guaranteed.

[0081] Example 6:

[0082] This embodiment, based on the above embodiments, provides a detailed explanation of how a wireless ECG terminal node ensures the priority of high-value abnormal waveform data transmission. In wireless communication networks, especially in shared-medium wireless LAN environments, when multiple nodes or data streams compete for the same channel resources, the access mechanism of the underlying link has a decisive impact on transmission latency and success rate. If only application-layer logical priority is relied upon without underlying link support, high-priority data packets may still compete fairly with low-priority data packets in channel contention, leading to transmission congestion or uncontrollable latency. Therefore, the wireless ECG terminal node also includes a link priority execution module, used to set the access priority parameters of the underlying wireless link according to the upload priority, to improve the access priority of high-priority raw ECG waveform segments in wireless channel contention.

[0083] Specifically, the link priority execution module maps the upload priority parameters generated by the application layer to access control parameters that the underlying wireless communication protocol stack can recognize and execute. Under normal conditions, the amount of ECG summary data periodically transmitted by the wireless ECG terminal node is small and the frequency is fixed, with relatively relaxed latency requirements. However, under abnormal backtracking conditions, the amount of original ECG waveform fragment data is large, highly bursty, and has high clinical timeliness requirements. Without priority differentiation, when the ward's wireless environment is congested, a large number of low-value summary packets may occupy the channel, preventing high-value original waveform packets from being transmitted in a timely manner. The link priority execution module modifies the wireless network card's transmission queue configuration, enabling high-priority data packets to obtain transmission opportunities earlier in channel contention, thereby ensuring the transmission efficiency of critical data at the physical level.

[0084] In a preferred embodiment of this invention, when the wireless ECG terminal node and the ward edge gateway communicate via a wireless LAN based on IEEE 802.11, the link priority execution module maps the upload priority to the WMM access class of the Media Access Control layer. High-priority raw ECG waveform segments correspond to higher-priority access classes, while periodically transmitted ECG summaries correspond to the best-effort access class. WMM is a wireless multimedia service quality assurance mechanism developed by the Wi-Fi Alliance based on the IEEE 802.11e standard. It defines four access classes: AC_VO, AC_VI, AC_BE, and AC_BK, corresponding to voice, video, best-effort, and background traffic, respectively. Different access classes correspond to different channel contention parameters, such as the number of arbitration inter-frame intervals and the contention window size.

[0085] In this specific implementation, the link priority execution module maintains a mapping table. When the upload control module determines that a high-priority raw ECG waveform segment needs to be uploaded, the link priority execution module maps the data stream to the AC_VO access category. AC_VO has the shortest AIFSN and the smallest contention window, meaning that during channel idle detection, the AC_VO queue only needs to wait a short time before attempting to send, and the backoff time for retransmission after a collision is also short. In contrast, regularly periodically sent ECG summary data is mapped to the AC_BE access category, which has a larger AIFSN and contention window parameters, resulting in a lower channel access priority. It should be understood that the above mapping relationship is only a preferred example of the present invention. In other embodiments, high-priority waveforms can also be mapped to AC_VI, or summary data can be mapped to AC_BK, depending on the actual network load, as long as the access parameters of high-priority data are better than those of low-priority data.

[0086] Through this mapping mechanism, this embodiment achieves end-to-end priority flow from clinical value determination to wireless channel access. When the ward edge gateway determines that a patient has a high-risk abnormality and issues a high upload priority command, the wireless ECG terminal node not only processes the data preferentially at the application layer but also gains "fast track" status in the underlying wireless channel contention. This ensures that even in extreme cases where ward wireless resources are scarce and multiple nodes compete for channels simultaneously, high-clinical-value abnormal waveform data can more likely preempt channel resources, significantly reducing the transmission latency of critical data and avoiding the loss or delay of critical data due to channel congestion. This provides more timely data support for medical staff to review and handle the data promptly.

[0087] Example 7:

[0088] This embodiment uses a scenario of monitoring suspected abnormal rhythm trends in a general ward as an example to provide a detailed explanation of the specific operation process of the system described in the aforementioned embodiments. This scenario aims to verify how the system can achieve accurate backtracking and reliable access to high-value abnormalities through an edge-end collaboration mechanism in a real clinical environment.

[0089] The scenario is as follows: A general inpatient ward has 20 beds, of which 8 patients are continuously monitored using wireless ECG terminal nodes. Patient A is a postoperative observation patient, classified as medium-to-high risk. The wireless ECG terminal node worn by Patient A is configured as follows: sampling rate Fs is 250Hz, and the local circular buffer length is... The abstract time window is 60 seconds. The upload interval is 5 seconds, and the summary upload cycle Tu is 5 seconds. A ward edge gateway is deployed within the ward, which connects to the clinical information system via the hospital's internal network.

[0090] Step S701, Individual Baseline Establishment. After patient A connects to the system, the system automatically establishes an individual baseline within the first 10 minutes when the waveform quality meets preset conditions. Within this baseline window, the system calculates and records the baseline value of patient A's RR interval dispersion. , Wavewidth baseline value and waveform quality baseline value This step, which determines the permissible range of variation for each feature, provides a reference standard for subsequent dynamic normalization based on individual baselines, effectively eliminating the influence of individual physiological differences on abnormality assessment.

[0091] Step S702, Routine Monitoring and Summary Upload. Under routine monitoring conditions, the wireless ECG terminal node continuously acquires the raw ECG waveform of patient A and stores it in a local circular buffer. Simultaneously, the summary extraction module extracts an ECG summary based on the raw waveform, including the mean heart rate and RR interval dispersion. QRS pulse width variation and waveform quality indicators The terminal node sends a summary packet to the ward edge gateway every 5 seconds. At this time, the terminal node only uploads a low-data-volume summary, and the radio frequency module operates in a low duty cycle state, effectively reducing node power consumption and wireless resource usage.

[0092] Step S703, Abnormal Trend Calculation. The ward edge gateway continuously receives three summary packets from patient A and detects the RR interval dispersion. The trend is showing a continuous increase. The gateway uses a dynamic normalization method based on individual baselines to calculate the normalized values ​​of each feature. Specifically, the calculated values ​​are... , , Subsequently, the intensity of the abnormal trend was calculated using a weighted scoring model: For three consecutive cycles Perform smoothing to obtain the strength of the smoothed abnormal trend. This value indicates a significant abnormal trend in patient A's electrocardiographic status.

[0093] Step S704, Abnormal Value Determination. The ward edge gateway further determines abnormal value by combining patient priority and ward resource status. Patient A is a medium-to-high priority patient and is in the postoperative observation period; their patient priority score is... The current resource status of the ward is medium occupancy, with a resource occupancy score of 0.8. The score is 0.5. The anomaly value score is calculated based on the weighted model: Preset trigger threshold It is 0.55. Although The anomaly was slightly below the threshold, but considering Patient A's specific clinical background, the system applied an additional trigger correction rule, determining that the anomaly warranted backtracking. This determination process reflects the core logic of the system's "value-driven" decision-making, namely, prioritizing the allocation of transmission resources to high-value anomalies under resource-constrained conditions.

[0094] Step S705: Generation of Retrospective Control Parameter Set. The ward edge gateway identifies the current anomaly category as a suspected abnormal rhythm trend and dynamically generates a retrospective control parameter set. Based on the anomaly category, a forward retrospective time window is established. Set to 30 seconds, backward rollback time window Set to 30 seconds. Considering patient A's high priority, the upload priority is adjusted accordingly. Set to "High". Meanwhile, given the current adequate ward resources, upload quality control parameters. Set to "Full Precision Upload". This parameter set is sent to the wireless ECG terminal node worn by patient A.

[0095] Step S706: The terminal performs backtracking and waveform locking. After receiving the backtracking control parameter set, the wireless ECG terminal node parses the parameters and performs the backtracking operation. First, it calculates the start position of the circular buffer read based on the forward backtracking time window. It is assumed that the pointer is written when an abnormality is triggered. Number of forward samples Then read the start pointer. The terminal node from Starting at the designated location, a temporary lock is applied to the buffer segment to prevent new data from overwriting it, and waveform data within the forward window is read sequentially. Simultaneously, the terminal node continues to acquire and upload real-time waveforms for the next 30 seconds, forming a complete context segment of the abnormal event.

[0096] Step S707, Link Priority Execution. To ensure the transmission efficiency of high-value abnormal waveforms, the link priority execution module maps the application layer's "high" upload priority to the access priority of the underlying wireless link. Specifically, in an IEEE 802.11 WLAN environment, the raw waveform return data is mapped to the AC_VO access category to obtain the shortest channel latency and the smallest contention window; while regular ECG summary data is mapped to the AC_BE access category. This mapping mechanism ensures that critical waveform data can obtain priority transmission opportunities in wireless channel contention, significantly reducing transmission latency.

[0097] Step S708, Trusted Access Determination. For example... Figure 4 As shown, after receiving a 60-second raw waveform clip from patient A, the ward edge gateway performs a trusted access determination. First, it queries the patient-bed-device mapping relationship. If a nurse has confirmed a bed assignment within the last 10 minutes and there is no record of bed changes, the mapping trustworthiness is determined. Secondly, a review of the nursing event records revealed no instances of patient turning or lead adjustments occurring near the time of the abnormal event, indicating consistency in the nursing events. Finally, analysis of the uploaded waveform revealed a significant irregularity in rhythm, consistent with the trend in the abstract, and good waveform consistency. Calculate the trusted access score: The score was significantly higher than the automatic access threshold of 0.80.

[0098] Step S709: Generate a structured monitoring event object. Based on the above judgment results, the ward edge gateway generates a structured monitoring event object for the "suspected abnormal rhythm retrospective event". This object includes fields such as patient identifier, bed identifier, event time, abnormality type, original waveform location information, and trusted access score. This object is automatically sent to the clinical information system for storage and display.

[0099] Step S710: System operation effect verification. Through the above process, this embodiment fully demonstrates the entire process from individual baseline establishment, abnormal trend detection, value judgment, parameter generation, waveform backtracking to trusted access. The system only transmits the original waveform when the abnormal value is high and resources permit, and after transmission, multi-dimensional verification is used to ensure the access credibility of the data. Compared with the prior art, this embodiment verifies that the system can effectively reduce wireless resource consumption and terminal power consumption. At the same time, under the condition of limited resources in the ward, it prioritizes the transmission of key waveform data of high-risk patients and successfully blocks the path of low-credibility data into the clinical information system, realizing continuous joint control of original ECG waveform backtracking and access qualification.

[0100] It should be understood that the specific parameter values ​​described in the above embodiments, such as the sampling rate of 250Hz, the buffer duration of 60 seconds, the weighting coefficient, and the threshold, are merely illustrative. Those skilled in the art can make adaptive adjustments according to actual application scenarios and device performance, and these adjustments do not depart from the protection scope of this invention.

[0101] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. For example, without departing from the spirit of the present invention, the weighted model for abnormal value scoring can be adaptively adjusted, or other wireless communication protocol standards can be used to implement link priority mapping; these transformations all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An Internet of Things based hospital clinical information system characterized in that, The system includes interconnected wireless ECG terminal nodes, a ward edge gateway, and a clinical information system. The wireless ECG terminal nodes continuously acquire raw ECG waveforms from patients, store these waveforms in a local circular buffer, and periodically send ECG summaries extracted from the raw waveforms. They generate anomaly trend intensity based on the ECG summaries and perform anomaly value determination based on the anomaly trend intensity, patient priority, and ward resource status. When the anomaly value score meets the backtracking trigger condition, a backtracking control parameter set is generated, including a forward backtracking time window, a backward backtracking time window, upload priority, and upload quality control parameters, and sent to the corresponding wireless ECG terminal node. The wireless ECG terminal node receives the original ECG waveform segment locked and uploaded from the local circular buffer according to the backtracking control parameter set; and after receiving the original ECG waveform segment, it generates a trusted access score based on the patient-bed-device mapping trustworthiness, nursing or medical order event consistency, and uploaded waveform consistency. When the trusted access score meets the access threshold, it generates a structured monitoring event object and sends it to the clinical information system. The structured monitoring event object includes at least a portion of patient identifier, bed identifier, event time, abnormality type, original waveform location information, and trusted access score. The clinical information system is used to receive the structured monitoring event object.

2. The hospital clinical information system based on Internet of Things according to claim 1, wherein, The wireless ECG terminal node includes: an ECG acquisition module for continuously acquiring the patient's original ECG waveform; a local circular buffer storage module for continuously storing the most recently preset duration original ECG waveform in a circular buffer manner; a summary extraction module for extracting ECG summaries based on the original ECG waveforms; and an upload control module for periodically sending the ECG summaries under normal monitoring conditions, and extracting and uploading the corresponding original ECG waveform segments after receiving the backtracking control parameter set.

3. The hospital clinical information system based on Internet of Things according to claim 2, characterized in that, The ECG summary includes at least one or a combination of the following: mean heart rate; mean RR interval; RR interval dispersion; QRS width variation; waveform quality index.

4. The hospital clinical information system based on Internet of Things according to claim 1, characterized in that, The ward edge gateway generates an abnormal trend intensity based on the ECG summary, and generates an abnormal value score based on the abnormal trend intensity, the patient priority, and the ward resource status. It then determines whether to initiate a retrospective analysis of the original ECG waveform segment based on a comparison between the abnormal value score and a preset trigger threshold. If the abnormal value score does not reach the preset trigger threshold but meets the additional trigger correction rule, the ward edge gateway still initiates a retrospective analysis of the original ECG waveform segment.

5. The hospital clinical information system based on Internet of Things according to claim 4, characterized in that, During the generation of the abnormal trend intensity, at least one summary feature is normalized. The normalization process includes at least one of the following methods: normalization based on a preset reference range; or dynamic normalization based on the individual baseline established during the initial stable monitoring phase after the patient accesses the system.

6. The hospital clinical information system based on Internet of Things according to claim 1, wherein, The ward edge gateway generates the backtracking control parameter set based on the anomaly category, patient priority, ward resource status, and available cache duration of the local circular buffer. The patient priority is a preset patient risk level, and a correction value is added to the patient priority when the patient is in the postoperative critical observation period. The forward backtracking time window and the backward backtracking time window are dynamically determined according to the anomaly category, the upload priority is dynamically determined according to the patient priority, and the upload quality control parameters are dynamically determined according to the ward resource status.

7. The hospital clinical information system based on Internet of Things according to claim 1, characterized in that, The wireless ECG terminal node determines the starting position of reading in the local circular buffer according to the forward backtracking time window, and continues to collect and upload subsequent raw ECG waveforms according to the backward backtracking time window. When entering the backtracking upload state, the corresponding buffer segment is temporarily locked to prevent the raw ECG waveforms that have not been uploaded from being overwritten by the newly sampled data.

8. The hospital clinical information system based on Internet of Things according to claim 1, characterized in that, It also includes a time synchronization and delay compensation module, which is used to enable bidirectional time synchronization between the wireless ECG terminal node and the ward edge gateway to determine clock deviation and link round-trip delay, and to enable the wireless ECG terminal node to perform local time correction when the determined clock deviation exceeds a preset threshold.

9. The hospital clinical information system based on Internet of Things according to claim 1, wherein, The wireless ECG terminal node also includes a link priority execution module, which is used to set the access priority parameters of the underlying wireless link according to the upload priority, so as to improve the access priority of high-priority raw ECG waveform segments in wireless channel contention.

10. A hospital clinical information system based on the Internet of Things according to claim 9, characterized in that, When the wireless ECG terminal node communicates with the ward edge gateway via a wireless local area network based on IEEE 802.11, the link priority execution module is used to map the upload priority to the WMM access category of the media access control layer, wherein a high-priority original ECG waveform segment corresponds to a higher-priority access category, and periodically sent ECG summaries correspond to a best-effort access category.

Images