Postpartum hemorrhage intelligent monitoring, prevention and control method and system based on multi-parameter integration and AI fusion

Abstract

The invention relates to a multi-parameter integration and AI fusion postpartum hemorrhage intelligent monitoring, prevention and control method and system. The method comprises the following steps: acquiring blood flow data; the blood flow data comprises blood loss volume and blood flow velocity, and the blood flow data is obtained through measurement of an optical fiber sensor; acquiring vital sign information; the vital sign information comprises blood pressure, heart rate and oxyhemoglobin saturation, and the vital sign information is obtained through continuous monitoring of a vital sign patch; according to the blood flow data, the vital sign information and the medical history information, a bleeding risk grade evaluation result is calculated in real time through an AI risk evaluation model, and the AI risk evaluation model is used for dynamically evaluating the bleeding risk; and triggering corresponding grading early warning according to the bleeding risk grade evaluation result, wherein the grading early warning is used for indicating bleeding risk processing. By adopting the method, the early warning accuracy and treatment efficiency of postpartum hemorrhage can be remarkably improved, and complications and death risks of pregnant and lying-in women are effectively reduced.

Description

Technical Field

[0001] The present invention belongs to the field of monitoring and prevention technology, and in particular relates to an intelligent monitoring and prevention method and system for postpartum hemorrhage that integrates multi-parameter integration and AI. Background Art

[0002] Postpartum hemorrhage (PPH) is a leading cause of maternal mortality, particularly in resource-limited settings, where its incidence and mortality remain high. Traditional PPH monitoring relies primarily on medical staff's judgment and manual measurement, which can be subjective, prone to errors, and delayed response. Furthermore, existing technologies lack the ability to integrate and analyze multiple parameters (such as blood loss and vital signs) in real time, leading to inaccurate risk assessments and delayed treatment.

[0003] Most monitoring devices currently on the market are single-function, unable to integrate data and conduct intelligent analysis, and lack seamless integration with electronic medical record systems. Therefore, there is an urgent need for a system that can accurately monitor the risk of postpartum hemorrhage in real time and provide intelligent decision support to reduce the risk of maternal complications and mortality. Summary of the Invention

[0004] Based on this, it is necessary to provide an intelligent monitoring and prevention method and system for postpartum hemorrhage that integrates multiple parameters and AI for accurate monitoring to address the above technical issues.

[0005] In the first aspect, the present application provides a method for intelligent monitoring and prevention of postpartum hemorrhage integrating multi-parameters and AI, including:

[0006] Obtaining blood flow data; blood flow data includes blood loss and blood flow velocity, and blood flow data is obtained by measuring with a fiber optic sensor;

[0007] Obtain vital signs information; vital signs information includes blood pressure, heart rate and blood oxygen saturation, which is continuously monitored through the vital signs patch;

[0008] Based on blood flow data, vital signs information, and medical history information, the AI ​​risk assessment model is used to calculate the bleeding risk level assessment results in real time. The AI ​​risk assessment model is used to dynamically assess bleeding risk;

[0009] The corresponding graded warning is triggered according to the bleeding risk level assessment results, and the graded warning is used to indicate the bleeding risk treatment.

[0010] In one embodiment, obtaining blood flow data includes:

[0011] Obtaining liquid level height data in the collection bag, and calculating blood loss data based on the liquid level height data;

[0012] According to the dynamic change of liquid level height data, blood flow velocity data is obtained through synchronous calculation;

[0013] The data of blood loss volume and blood flow velocity are preliminarily processed to obtain preprocessed blood flow data.

[0014] In one embodiment, an AI risk assessment model is used to calculate the bleeding risk level assessment results in real time based on blood flow data, vital signs information, and medical history information, including:

[0015] Real-time monitoring data on blood loss, blood flow rate, vital signs, and medical history characteristics are sent to a trained AI risk assessment model. The AI ​​risk assessment model then performs weighted analysis and comprehensive evaluation on the real-time monitoring data to obtain a bleeding risk level assessment result.

[0016] Based on the real-time monitoring data that is continuously sent, periodically updated risk assessment results are obtained.

[0017] In one embodiment, the AI ​​risk assessment model performs weighted analysis and comprehensive evaluation on real-time monitoring data to obtain a bleeding risk level assessment result, including:

[0018] The risk probability value is generated according to the real-time monitoring data, and the risk probability value is generated by the calculation formula of the following risk quantification function:

[0019] R(t)=σ(W(t)*X(t)+b)

[0020] Where R(t) is the risk probability value, W(t) is the dynamic weight matrix, X(t) is the time series feature matrix, b is the bias term, and σ is the Sigmoid activation function;

[0021] Real-time risk grading is performed based on the risk probability value to obtain the bleeding risk level assessment result.

[0022] In one embodiment, after obtaining periodically updated risk assessment results based on the continuously transmitted real-time monitoring data, the method further includes:

[0023] Obtain real-time calculated bleeding risk assessment results, which include low, intermediate, and high levels;

[0024] According to the bleeding risk level assessment results, the corresponding treatment plan library is retrieved;

[0025] Based on the patient's medical history, select the appropriate medication list and surgical recommendations from the corresponding treatment plan library;

[0026] Combined with the medication list and surgical recommendations, the recommended treatment options and implementation steps are displayed in a prioritized manner;

[0027] A countdown reminder is triggered when the disposal plan and implementation steps are pushed, and the delay time is calculated in real time to display the delayed time.

[0028] In one embodiment, after obtaining the real-time calculated bleeding risk level assessment result, the method further includes:

[0029] The early warning strategy is matched based on the bleeding risk level assessment results. When the risk level is intermediate, the first level warning is triggered, and a push notification is sent to the nursing mobile terminal based on the first level warning. When the risk level is high, the second level warning is triggered, and the sound and light alarm device is activated according to the second level warning and an emergency push notification is sent to the doctor's mobile terminal, generating an early warning prompt for timely risk handling.

[0030] Monitor the warning response status based on the early warning prompts for timely risk handling and obtain the early warning response results; the early warning response results are obtained through the early warning prompt feedback received by the sound and light alarm device or the designated mobile terminal;

[0031] Obtain the real-time monitoring of the early warning response status. If no confirmation feedback of the early warning response status is received within the preset time, the early warning level will be upgraded;

[0032] The warning trigger time, risk level and response results are synchronously stored in the electronic medical record system.

[0033] In one embodiment, any of the above embodiments further includes:

[0034] Acquiring multivariate data, including blood loss data, blood flow rate data, and vital sign information;

[0035] Perform time alignment, unit unification, and outlier filtering on the collected multi-dimensional data to obtain processed monitoring data;

[0036] Write the processed monitoring data into the hospital electronic medical record system according to the preset format to obtain time series data;

[0037] Draw blood loss curves and vital signs trend charts based on time series data;

[0038] For blood loss change curves and vital signs trend charts, the charts are associated with the original monitoring data and stored, and historical data retrospective analysis is supported.

[0039] Secondly, this application also provides a multi-parameter integrated and AI-integrated intelligent monitoring and prevention system for postpartum hemorrhage, including:

[0040] A blood flow data acquisition module is used to acquire blood flow data; the blood flow data includes blood loss and blood flow velocity, and the blood flow data is obtained by measuring with an optical fiber sensor;

[0041] A vital sign information acquisition module is used to obtain vital sign information; vital sign information includes blood pressure, heart rate and blood oxygen saturation, and vital sign information is continuously monitored by a vital sign patch;

[0042] The AI ​​risk assessment model calculation module is used to calculate the bleeding risk level assessment results in real time based on blood flow data, vital signs information, and medical history information through the AI ​​risk assessment model. The AI ​​risk assessment model is used to dynamically assess bleeding risk;

[0043] The risk classification warning module is used to trigger corresponding classification warnings based on the bleeding risk level assessment results. The classification warnings are used to indicate bleeding risk treatment.

[0044] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above embodiments when executing the computer program.

[0045] In a fourth aspect, the present application further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any of the above embodiments are implemented.

[0046] The above-mentioned intelligent monitoring and prevention method and system for postpartum hemorrhage that integrates multiple parameters and AI fusion collects blood flow data and physiological parameters in real time through fiber optic sensors and vital sign patches, and dynamically analyzes bleeding risks and triggers graded warnings in combination with AI risk assessment models, significantly improving the accuracy and timeliness of postpartum hemorrhage warnings, and reducing the risks of maternal complications and deaths caused by subjective judgment errors or response delays. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 Schematic diagram of the implementation environment of the multi-parameter integration and AI fusion postpartum hemorrhage intelligent monitoring and prevention method of the present invention;

[0049] Figure 2 This is a flow chart of the intelligent monitoring and prevention method for postpartum hemorrhage integrating multi-parameter integration and AI fusion of the present invention;

[0050] Figure 3 This is a structural block diagram of the intelligent monitoring and prevention system for postpartum hemorrhage with multi-parameter integration and AI fusion of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0052] The intelligent monitoring and prevention method for postpartum hemorrhage with multi-parameter integration and AI fusion provided in the embodiments of the present application can be applied to Figure 1 In the implementation environment shown. Among them, the intelligent monitoring terminal 101 communicates with the server 102 through the network. The data storage system can store the data that the server 102 needs to process. The data storage system can be integrated on the server 102, or it can be placed on the cloud or other network servers. Among them, the intelligent monitoring terminal 101 can be, but is not limited to, various personal computers, laptops, smart phones, tablets and portable wearable devices. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server 102 can be implemented as an independent server or a server cluster consisting of multiple servers.

[0053] In combination with the above implementation environment, the application scenarios of the embodiments of the present application are explained.

[0054] The embodiments of this application are applicable to postpartum care wards. In this scenario, medical staff use disposable smart collection bags and vital sign patches on patients to collect relevant data of the mother in real time on the smart monitoring terminal. Combined with the mother's medical history information, the mother's bleeding risk level is determined. Based on the graded warning triggered by the risk level, medical staff promptly intervene according to the recommended treatment plan, such as adjusting the care plan or notifying the doctor to take emergency measures.

[0055] This is only an example and does not limit the specific application scenario.

[0056] In an exemplary embodiment, Figure 2 As shown in the figure, a multi-parameter integration and AI fusion intelligent monitoring and prevention method for postpartum hemorrhage is provided. Figure 1 The intelligent monitoring terminal in FIG. 1 is taken as an example to illustrate the method, which includes the following steps 111 to 114. Among them:

[0057] S111, obtaining blood flow data; blood flow data includes blood loss and blood flow velocity, and the blood flow data is obtained by measuring with a fiber optic sensor.

[0058] Alternatively, in actual applications, a smart collection bag based on the deployment of fiber optic sensors can be used. When a woman experiences postpartum hemorrhage, the blood flowing into the collection bag is converted into blood loss using a preset volume-height curve. The blood flow rate is also calculated in real time by monitoring the rate of change of the liquid level per unit time. This data is transmitted to the smart monitoring terminal via a low-power Bluetooth module at a frequency of once per second. Blood flow data is a physical quantity dataset that quantifies the dynamic characteristics of postpartum hemorrhage using biomedical sensing technology. Fiber optic sensors are miniature optical detection devices constructed using medical-grade polymer fiber bundles. The sensors have an embedded anti-reflux design to prevent contamination. The smart collection bag is sterile, disposable, and made of biocompatible materials.

[0059] S112, obtaining vital sign information; vital sign information includes blood pressure, heart rate and blood oxygen saturation, and vital sign information is obtained through continuous monitoring of the vital sign patch.

[0060] Among them, the vital signs patch can refer to a wireless flexible wearable device made of a biocompatible silicone substrate. The vital signs patch can also be simply referred to as a patch. For example, the patch continuously collects maternal physiological signals through its integrated multimodal sensor array. When working, the patch is attached to the upper sternum of the maternal body. Its blood oxygen module penetrates the subcutaneous tissue by emitting a 660nm / 940nm dual-wavelength light beam, and calculates the blood oxygen saturation based on the difference in hemoglobin absorption of light of different wavelengths. At the same time, the pressure sensor monitors the pulsating pressure waveform of the blood vessel wall at a sampling rate of 50Hz, calculates the real-time heart rate in combination with the R wave peak detection of the ECG lead, and analyzes the pressure wave amplitude change by the oscillation method. All original signals are subjected to baseline drift correction and motion artifact suppression by the built-in ASIC chip of the patch, and finally fused to generate a vital signs data packet.

[0061] In addition, the vital signs patch can be designed as a self-powered flexible patch that uses a friction nanogenerator (TENG) to obtain energy from human movement and can work continuously without batteries; the patch has a built-in edge AI chip that can directly complete vital signs analysis locally (such as heart rate variability to predict the risk of major bleeding), and only upload key data to the cloud, which can reduce network dependence, extend the battery life of the device, and improve real-time performance.

[0062] S113, based on blood flow data, vital signs information and medical history information, the AI ​​risk assessment model is used to calculate the bleeding risk level assessment results in real time. The AI ​​risk assessment model is used to dynamically assess bleeding risk.

[0063] For example, after the intelligent monitoring terminal receives the real-time blood loss data from the fiber optic sensor and the dynamic parameters such as blood pressure, heart rate, and blood oxygen saturation transmitted by the vital sign patch, it first performs spatiotemporal alignment processing, and then forms a multidimensional feature vector together with the maternal medical history characteristics extracted from the electronic medical record system; the vector is input into the pre-trained three-layer LSTM-attention hybrid neural network model. The model first analyzes the temporal characteristics through the LSTM layer, and then calculates the contribution of each parameter through the attention weight layer, and finally outputs a risk probability value between 0 and 1, and dynamically divides the risk level according to the preset threshold.

[0064] S114, triggering a corresponding graded warning based on the bleeding risk level assessment result, and the graded warning is used to indicate bleeding risk treatment.

[0065] Among them, the early warning decision engine is immediately started after the risk level is output. When the risk level is judged to be intermediate, an early warning message is automatically generated and pushed to the responsible nurse's mobile nursing terminal through the hospital's Internet of Things platform. The message content includes the patient's bed number, current blood loss and abnormal key vital signs. At the same time, the corresponding bed logo on the central monitoring screen of the nurse station switches to a flashing yellow state; for high-risk cases, the three-channel alarm protocol is synchronously activated to control the sound and light alarm installed on the top of the delivery room to emit a 105-decibel beep and a red strobe signal, and the highest priority push is sent to the attending physician's emergency PDA (Personal Digital Assistant) through the 5G medical dedicated network. The notification will continue to flash full-screen until manually confirmed, and will automatically link to the electronic medical record to generate a risk warning card containing the treatment plan recommended by the ACOG (American College of Obstetricians and Gynecologists) guidelines; at the same time, a progressive enhancement strategy will be adopted during the early warning execution process.

[0066] In the above-mentioned intelligent monitoring and prevention method for postpartum hemorrhage, multi-parameter integrated assessment of risk levels is used for graded warning, which shortens the clinical response time and ultimately reduces the incidence of serious complications.

[0067] In an exemplary embodiment, obtaining blood flow data includes:

[0068] S211, obtaining the liquid level height data in the collection bag, and calculating the blood loss data based on the liquid level height data.

[0069] Specifically, the fiber optic sensor emits a modulated light signal of a specific wavelength into the smart collection bag. When the light is totally reflected at the blood-air interface, the light intensity attenuation value detected by the receiving end is nonlinearly related to the liquid level height. The intelligent monitoring terminal converts the real-time light intensity signal into an accurate liquid level height value through a pre-calibrated three-dimensional volume-height-light intensity mapping matrix, and then uses an integral algorithm to calculate the current blood loss volume based on the specific geometric structure of the collection bag (such as a trapezoidal cross-section design) (for example, when the liquid level rises from 2 cm to 5 cm, the corresponding blood loss increases from 180 ml to 450 ml). At the same time, the Kalman filter algorithm is used to eliminate the interference of liquid surface fluctuations caused by changes in the mother's body position, and finally outputs blood loss data with an accuracy of ±5 ml.

[0070] S212, synchronously calculate according to the dynamic change of the liquid level height data to obtain the blood flow rate data.

[0071] Optionally, the intelligent monitoring terminal samples the liquid level height data at a frequency of 10 times per second to eliminate random fluctuations caused by respiratory movement or fine-tuning of body position; then a first-order differential operation is performed on the filtered liquid level-time curve. When three consecutive differential values ​​exceeding a threshold (such as Δh>0.3 cm / s) are detected, the flow rate calculation module is triggered to start, and the intelligent monitoring terminal converts the liquid level change rate into blood flow rate in combination with the real-time cross-sectional area of ​​the collection bag.

[0072] S213, preliminarily processing the data on blood loss volume and blood flow velocity to obtain preprocessed blood flow data.

[0073] For example, the original blood loss data is processed by sliding median filtering, and a window with a width of 5 sampling points is used to remove impulse noise. At the same time, linear interpolation is used to compensate for discontinuous data points caused by signal loss. The compensated data is then time-aligned, unit-unified, and outlier-filtered. The final generated standard format preprocessed data packet is encrypted and written into the ring buffer.

[0074] In the above-mentioned intelligent postpartum hemorrhage monitoring method, blood flow data is accurately measured by optical fiber sensors, which can transform the qualitative empirical judgment of traditional postpartum hemorrhage monitoring into quantitative judgment, thereby improving the accuracy of heavy bleeding identification.

[0075] In one embodiment, an AI risk assessment model is used to calculate the bleeding risk level assessment results in real time based on blood flow data, vital signs information, and medical history information, including:

[0076] S311, the real-time monitoring data of blood loss, blood flow rate, vital signs and medical history characteristics are sent to the trained AI risk assessment model, and the real-time monitoring data are weighted analyzed and comprehensively evaluated by the AI ​​risk assessment model to obtain the bleeding risk level assessment result.

[0077] Optionally, through federated learning, AI models can be trained with data from multiple hospitals, protecting patient privacy while improving model generalization. The trained AI models are validated for performance in different populations (e.g., patients with coagulation disorders) and healthcare settings (e.g., home monitoring). The intelligent monitoring terminal first synchronizes the time-series data of blood loss and blood flow rate collected by the fiber optic sensor with the physiological parameters obtained by the vital signs patch to obtain preprocessed dynamic feature data. These dynamic feature data, combined with the static features of the medical history extracted from the electronic medical record, form a multidimensional input vector that is input into a pretrained deep neural network model for processing. The model first analyzes the long-term dependencies of the time-series features through a bidirectional LSTM layer to capture dangerous patterns such as accelerated blood loss. The attention mechanism layer then automatically calculates the weight distribution of each feature, assigning higher importance scores to abnormal signs. After the weighted features undergo nonlinear transformation through the fully connected layer, a risk probability value between 0 and 1 is output. The intelligent monitoring terminal then classifies this continuous value into three risk levels: low, medium, or high, based on the preset threshold range.

[0078] Among them, the trained AI risk assessment model refers to the deep neural network architecture obtained by pre-training on 100,000 clinical data through supervised learning methods; weighted analysis refers to the model assigning dynamically variable weight coefficients to input features; comprehensive evaluation refers to the model mapping the weighted feature vectors to coordinate points in the three-dimensional risk space through the fully connected layer; medical history features refer to the one-hot encoded vectors of static risk factors extracted from the electronic medical record system, including coagulation function indicators, previous bleeding history, etc.

[0079] S312: Based on the continuously sent real-time monitoring data, a periodically updated risk assessment result is obtained.

[0080] For example, a distributed message queue receives medical sensor data in real time, dynamically assessing bleeding risk in 30-second cycles. Using a sliding time window to analyze the last five minutes of data, edge computing nodes complete risk assessments within 300 milliseconds using exponential decay weighting (with new data weighted 50%) and 12-dimensional feature extraction. If an abnormality is detected, such as a sudden increase in blood loss (e.g., from 300ml to 550ml), the assessment cycle is automatically shortened to 10 seconds, triggering an emergency alert. All updated risk assessment results are synchronized to the electronic medical record in real time.

[0081] In the above-mentioned intelligent monitoring method for postpartum hemorrhage, through the real-time fusion of multivariate data and the dynamic risk assessment algorithm, it is possible to transform intermittent manual assessment into continuous intelligent monitoring, thereby improving the timeliness of postpartum hemorrhage risk identification.

[0082] In an exemplary embodiment, the AI ​​risk assessment model performs weighted analysis and comprehensive evaluation on real-time monitoring data to obtain a bleeding risk level assessment result, including:

[0083] S411: Generate the risk probability value based on the real-time monitoring data. The risk probability value is generated by the following risk quantification function calculation formula:

[0084] R(t)=σ(W(t)*X(t)+b)

[0085] Among them, R(t) is the risk probability value, W(t) is the dynamic weight matrix, X(t) is the time series feature matrix, b is the bias term, and σ is the Sigmoid activation function.

[0086] For example, real-time risk assessment is achieved through efficient data processing and intelligent analysis. Monitoring data is extracted and standardized from the circular buffer every 30 seconds to construct a 12-dimensional time series matrix X(t). The feature weight W(t) is dynamically adjusted through the attention mechanism (for example, the weight increases from 0.5 to 0.8 when the bleeding rate exceeds 50ml / min). After the matrix operation is completed in parallel on the edge computing node, the Sigmoid function outputs a risk probability value R(t) of 0-1 (for example, a risk probability value of 0.91 is output when blood pressure drops sharply). The risk change rate is monitored in real time (a slope exceeding 0.1 / minute triggers an early warning). The entire analysis process takes only 200 milliseconds, and the model performance is continuously optimized by persistently storing the weight matrix. The bias term b (initial value -0.2) is used to fine-tune the decision threshold to ensure the accuracy of the assessment.

[0087] S412: Perform real-time risk grading based on the risk probability value to obtain a bleeding risk level assessment result.

[0088] Specifically, the intelligent monitoring terminal performs real-time grading of the probability values ​​output by the risk quantification function. First, a sliding window is used to smooth the data to eliminate fluctuations. Then, a dual-threshold comparator is used to assign risk levels. Probability values ​​below 0.3 are classified as low and displayed in green. Values ​​between 0.3 and 0.7 are intermediate and trigger a yellow warning. Values ​​exceeding 0.7 are upgraded to high and trigger a red emergency alert. A hysteresis interval design is used to prevent frequent jumps in the grading process. For example, a risk value rising from 0.68 to 0.72 requires three consecutive threshold crossings before confirmation of an upgrade. The intelligent monitoring terminal also generates a visual report. If the risk probability rises from 0.25 to 0.82 within 10 minutes, the interface automatically highlights abnormal parameters such as blood loss exceeding 500 ml and systolic blood pressure below 90 mmHg. Grading results are pushed to the delivery room monitor, nurse station dashboard, and physician mobile device via the Medical Internet of Things protocol. A voice announcement is activated for detailed information when a high-level determination is made. All grade changes are recorded in an audit log, including the time of change, duration, and associated indicators, to support quality traceability and analysis.

[0089] In the above-mentioned intelligent early warning method for postpartum hemorrhage, dynamic risk quantification and real-time graded decision-making mechanism are used to transform traditional empirical judgments into data-driven precise graded early warnings, thereby improving the accuracy of clinical intervention and shortening the time for identifying high-risk cases.

[0090] In one embodiment, after obtaining periodically updated risk assessment results based on the continuously transmitted real-time monitoring data, the method further includes:

[0091] S511, obtaining the bleeding risk level assessment result calculated in real time, the bleeding risk level assessment result includes low, medium and high level.

[0092] Specifically, the AI ​​risk assessment model conducts a comprehensive analysis of multivariate data collected in real time (such as blood flow data, vital signs information, and medical history information). The model processes the data according to preset algorithms and weights, and ultimately generates a bleeding risk level assessment result, and grades the risk level according to different preset thresholds.

[0093] S512: Based on the bleeding risk level assessment results, retrieve the corresponding treatment plan library.

[0094] Optionally, based on the real-time calculated bleeding risk assessment, the intelligent monitoring terminal retrieves appropriate treatment measures from a pre-established library of treatment plans based on international guidelines. This library is updated to the system based on real-time changes to international guidelines. For example, if the risk level is intermediate, treatment plans corresponding to intermediate risk are retrieved from the library. These plans typically include detailed medication recommendations, such as using specific doses of uterotonics to promote uterine contractions and reduce bleeding, as well as possible nursing care measures, such as close monitoring of the mother's vital signs and bleeding. If the risk level is high, more urgent and comprehensive treatment plans are retrieved, which may include surgical intervention recommendations, such as preparation for uterine artery ligation, and also outline key pre- and post-operative care points and precautions.

[0095] S513, based on the medical history characteristics, screen the applicable drug list and surgical recommendations from the corresponding treatment plan library.

[0096] For example, the intelligent monitoring terminal will further filter and personalize the treatment plan library based on the mother's medical history. First, the terminal identifies the mother's medical history, such as whether she has coagulation disorders, drug allergies, or a history of uterine surgery. These medical history characteristics will serve as screening criteria to select a medication list and surgical recommendations from the treatment plan library that are suitable for the mother's individual situation.

[0097] S514, combining the medication list and surgical recommendations, displays the recommended treatment options and implementation steps in a prioritized manner;

[0098] Specifically, the intelligent monitoring terminal conducts a comprehensive analysis of the medication list and surgical recommendations based on the mother's bleeding risk level, medical history, and current clinical condition. It then prioritizes and displays the recommended treatment options. Prioritization refers to the order in which recommended treatment measures are arranged based on their importance and urgency. The treatment plan and implementation steps are a comprehensive set of treatment recommendations, derived from the medication list and surgical recommendations, based on the mother's specific condition and risk level. These recommendations include specific operational procedures and steps, providing clear implementation guidance for medical staff.

[0099] S515: trigger a countdown reminder based on the push of the disposal plan and implementation steps, calculate the delay time in real time, and obtain the delayed time display.

[0100] Specifically, while the treatment plan and specific implementation steps are being pushed to medical staff, a countdown reminder function is activated to ensure that medical staff can promptly deal with the risk of postpartum hemorrhage within the specified time.

[0101] For example, when a smart monitoring terminal assesses a pregnant woman's bleeding risk as high and recommends an emergency treatment plan, it sets a countdown based on the urgency of the treatment plan, such as requiring surgery to begin within 30 minutes. This countdown reminder appears on the medical staff's mobile device or medical equipment via a pop-up window, sound, or vibration, prompting them to take immediate action. The system also records the time difference between the time the treatment plan is recommended and the time the medical staff actually starts implementing it, known as the delay time. If the medical staff begins implementing the treatment plan before the countdown expires, the actual response time is recorded and compared with the countdown time to calculate the delay time. If the medical staff fails to start the treatment plan before the countdown expires, the full countdown time is recorded as the delay time. The calculated delay time is intuitively displayed to the medical staff, such as "Delayed for XX minutes" on the mobile device screen, to help them understand the current treatment delay.

[0102] In the above-mentioned intelligent monitoring and prevention method for postpartum hemorrhage, by calling personalized treatment plans and monitoring treatment delays in real time, accurate monitoring, rapid early warning and efficient prevention and control of postpartum hemorrhage are achieved, which significantly improves the prevention and treatment effect of postpartum hemorrhage and the scientific nature of medical decision-making.

[0103] In one embodiment, after obtaining the real-time calculated bleeding risk level assessment result, the method further includes:

[0104] S611, match the early warning strategy based on the bleeding risk level assessment results. When the risk level is intermediate, the first level early warning is triggered, and a push notification is sent to the nursing mobile terminal based on the first level early warning; when the risk level is high, the second level early warning is triggered, and the activation of the sound and light alarm device is controlled according to the second level early warning and an emergency push notification is sent to the doctor's mobile terminal, generating an early warning prompt for timely risk handling.

[0105] Optionally, the risk level changes are monitored in real time. When the risk level is detected to be intermediate, the early warning engine immediately starts the multi-channel alarm protocol. First, a standard alarm message containing the patient's bed number, risk value and key indicator trends is sent to the responsible nurse's mobile nursing terminal through the hospital's Internet of Things platform, and the patient information is highlighted on the electronic whiteboard at the nurse station. For high-risk cases, the sound and light alarm installed on the top of the delivery room is activated synchronously, driving the red LED to flash at a frequency of 2Hz and emit intermittent beeps. At the same time, the highest priority notification is pushed to the attending physician's emergency PDA through the 5G medical dedicated network. The notification will continue to vibrate and cover the current screen display until the physician clicks Confirmation; the warning content is dynamically generated by the clinical decision engine and automatically associated with the patient's electronic medical record data. For example, when it is identified that the mother has a history of thrombocytopenia, the platelet transfusion plan will be recommended as a priority in the recommendation; the first-level warning corresponds to the standardized handling process triggered by the intermediate risk state, and a structured notification message is sent to the nursing staff's mobile terminal through the standard protocol; the second-level warning corresponds to the multi-modal alarm combination triggered by the high-risk state, including controlling the relay to drive the sound and light alarm device to emit a 105-decibel buzzer and a red strobe signal; the sound and light alarm device refers to an industrial-grade alarm controlled by a communication protocol, and its red light wavelength of 630nm ensures clear identification in the delivery room environment.

[0106] S612, monitor the warning response status according to the early warning prompt for timely risk handling and obtain the early warning response result; the early warning response result is obtained through the early warning prompt feedback received by the sound and light alarm device or the designated mobile terminal.

[0107] Specifically, a distributed status tracking engine is activated immediately after an alert is triggered. The IoT middleware continuously polls the reset signals of the audible and visual alarm devices and operational feedback from mobile terminals. When a confirmation message containing the nurse's electronic signature is detected from the nursing terminal, the response timestamp and personnel information are automatically recorded. For advanced risk alerts, additional biometric feedback from the attending physician's PDA is required, and the actual treatment progress is cross-verified with the card swipe records of the operating room access control system. All feedback data is encrypted and written to the blockchain node for storage. The alert response status refers to the management status of the intelligent monitoring terminal's real-time tracking of the treatment process after various alerts are triggered. The alert response result is a structured record generated by parsing the feedback signal, including the response time and the person who handled the situation. The designated mobile terminal refers to a dedicated medical device certified by the medical institution. The alert prompt feedback is the confirmation message returned by the terminal device, containing status tags such as "read," "processing," or "referred."

[0108] In addition, visual operations can be performed through the AR glasses integrated system. When an advanced warning is triggered, a three-dimensional bleeding simulation image (reconstructed based on blood flow data) and treatment instructions are automatically projected. Combined with gesture recognition technology, alienated personnel can use gestures to switch to view recommended drug dosages and surgical anatomical pathways (such as navigation for the implementation of uterine artery ligation), and simultaneously record the operation steps to the monitoring terminal system for storage.

[0109] S613: Obtain the early warning response status of real-time monitoring. If no confirmation feedback of the early warning response status is received within the preset time, the early warning level is upgraded.

[0110] When the initial alert is triggered, the countdown monitoring module is activated simultaneously, comparing the current time with the alert timestamp in real time. If a medium-risk alert persists for 15 minutes or a high-risk alert persists for 5 minutes without receiving a digitally signed response, a three-level escalation process is automatically executed. This process first extends notification to the mobile devices of the attending physician and obstetrics team leader on duty, then activates the ward broadcast system to play a voice prompt, and finally generates a red alert pop-up window in the electronic medical record system. The escalation process utilizes a progressive enhancement strategy. For example, if a woman's postpartum hemorrhage alert is not addressed within 8 minutes, the intelligent monitoring terminal first sends a level 2 alert to the head nurse's terminal. If there is still no response after another 2 minutes, the intelligent monitoring terminal connects to the hospital's emergency response team's dedicated communication terminal and switches the nurses' station's large screen to a full-screen flashing alert mode. All escalation events are recorded as quality control events, and the intelligent monitoring terminal automatically analyzes response delays and generates a report with improvement recommendations. The preset time refers to the tiered response time limit set according to clinical guidelines. Confirmation feedback refers to the electronic disposition certificate with a digital signature and timestamp that the intelligent monitoring terminal must receive.

[0111] S614, the warning trigger time, risk level and response result are synchronously stored in the electronic medical record system.

[0112] For example, after completing the risk level assessment, the intelligent monitoring terminal immediately generates a standard-compliant early warning event resource object. This object encapsulates a timestamp accurate to the millisecond, a standardized risk level code, and preliminary treatment recommendations, and is encrypted and transmitted to the electronic medical record system via the hospital information exchange bus. The intelligent monitoring system terminal regularly generates storage quality reports, analyzes the integrity and timeliness indicators of the early warning data, and provides decision support for the hospital quality management committee. The early warning trigger time refers to the initial occurrence of the risk event accurate to the millisecond level, verified by the server. The risk level is a clinically verified three-level categorical variable output by the AI ​​risk assessment model, using ordered numerical codes 0 / 1 / 2 corresponding to low / medium / high risk respectively. The response result refers to a structured data set containing treatment personnel, measures, and timestamps recorded through the Flag resource in the standard.

[0113] In the above-mentioned intelligent early warning management method for postpartum hemorrhage, through the closed-loop management of multi-level early warning trigger mechanism, real-time response status tracking and automatic archiving of electronic medical records, it is possible to transform the traditional passive emergency response into an intelligent active prevention and control system, so that the timeliness of early warning response is improved and the entire clinical treatment process is traceable, ultimately reducing the incidence of adverse events related to postpartum hemorrhage.

[0114] In an exemplary embodiment, the intelligent early warning method for postpartum hemorrhage provided in the embodiments of the present application may further include the following steps:

[0115] S711, acquiring multivariate data, which includes data on blood loss, blood flow rate, and vital signs information.

[0116] Specifically, the intelligent monitoring terminal uses a variety of sensors and monitoring devices to simultaneously capture a variety of key data related to postpartum hemorrhage. These data sources are diverse and complementary, providing a comprehensive basis for subsequent risk assessment. Fiber optic sensors collect data on blood loss and blood flow rate. For example, after delivery, a fiber optic sensor is installed in a collection bag to monitor blood flow in real time. When bleeding occurs, the sensor accurately measures blood flow rate and calculates blood loss and blood flow rate through integration. This data is recorded in a time series format, providing important information for assessing the dynamic changes in bleeding. Vital signs of the mother, including blood pressure, heart rate, and blood oxygen saturation, are collected using a vital sign patch or monitor.

[0117] By integrating these diverse data sets, a comprehensive understanding of a woman's postpartum hemorrhage can be achieved. Blood loss and blood flow provide specific quantitative information on bleeding, while vital signs reflect the woman's physiological response to bleeding. This combination of data enables a more accurate assessment of the risk of postpartum hemorrhage and provides comprehensive decision-making support for healthcare professionals.

[0118] S712: Time alignment, unit unification, and outlier filtering are performed on the collected multi-dimensional data to obtain processed monitoring data.

[0119] First, time alignment is performed to unify data with different sampling frequencies onto a per-second timeline. For example, interpolation is used to convert per-minute blood loss data to second-level data. IEEE 1588 Precision Time Protocol (PTP) is then used to achieve μs-level synchronization through hardware timestamps. Second, unit unification is performed to convert different units, such as milliliters and milliliters per minute, to a unified standard. Finally, statistical methods are used to filter out outliers caused by sensor failures or transmission errors. After these three preprocessing steps of time alignment, unit unification, and outlier filtering, the intelligent monitoring terminal obtains standardized monitoring data, providing an accurate basis for subsequent risk assessment.

[0120] S713, write the processed monitoring data into the hospital electronic medical record system according to the preset format to obtain time series data.

[0121] Optionally, the preprocessed monitoring data is organized and stored in a specific format required by the hospital's electronic medical record system, allowing medical staff to easily access and analyze the data. The processed monitoring data (such as blood loss, blood flow rate, and vital signs) is formatted. For example, the data is converted to XML, JSON, or CSV formats supported by the electronic medical record system, ensuring that the data fields and structure meet the hospital's regulatory requirements. This formatted data contains not only specific values ​​but also timestamp information to record the specific time of data collection. This formatted data is written to the hospital's electronic medical record system. For example, the data can be transferred to the electronic medical record system through the hospital's medical information exchange interface and stored in the mother's personal medical record. This data is stored in a time series format, arranged in chronological order, making it easier for medical staff to view changes in the mother's monitoring data at different time points. Finally, time series data is generated in the electronic medical record system.

[0122] S714: Draw a blood loss change curve and a vital sign trend chart based on the time series data.

[0123] Among them, the time series data stored in the hospital's electronic medical record system is used to generate intuitive charts through data visualization technology to help medical staff quickly understand the mother's blood loss and changing trends in her vital signs.

[0124] Time series data for parturients was extracted from the electronic medical record system. This data included key indicators such as blood loss, blood flow rate, blood pressure, heart rate, and blood oxygen saturation. Each data point was timestamped. Based on the chronological order of this data, a blood loss curve and vital sign trend chart were plotted.

[0125] For example, a blood loss curve uses time as the horizontal axis and blood loss as the vertical axis, connecting the blood loss data points at each time point to form a continuous curve. This curve can intuitively show the changes in blood loss at different time points after delivery. If the curve shows a rapid upward trend, medical staff can quickly determine that bleeding is increasing and that timely measures are needed.

[0126] For vital sign trend charts, the smart monitoring terminal plots trends for blood pressure, heart rate, and blood oxygen saturation. Taking heart rate as an example, again with time as the horizontal axis and heart rate as the vertical axis, the heart rate data points at each time point are connected to form a trend line. By observing this trend line, medical staff can determine whether the mother's heart rate is stable or experiencing any abnormal fluctuations.

[0127] S715, for blood loss change curves and vital signs trend charts, associates the charts with the original monitoring data and stores them, and supports historical data retrospective analysis.

[0128] Among them, by associating and storing the generated charts with the original monitoring data and providing historical data backtracking analysis functions, a comprehensive and convenient data management and analysis tool is provided for medical staff.

[0129] Blood loss curves and vital sign trend charts are stored as images or data files in the hospital's electronic medical record system. These charts not only contain visual data but also establish a relationship with the original monitoring data. For example, the intelligent monitoring terminal assigns a unique identifier to each chart and associates it with the corresponding original monitoring data record. This allows medical staff to directly access the corresponding original monitoring data and obtain more detailed information by clicking on a data point or time period in the chart.

[0130] Furthermore, the intelligent monitoring terminal supports historical data retrospective analysis. This means that medical staff can access maternal monitoring data at any time, including blood loss, blood flow, blood pressure, heart rate, and other vital signs. For example, medical staff can use the electronic medical record system's time filter function to view blood loss curves and vital sign trend charts for different time points, such as the first and second days after delivery. This retrospective analysis of historical data allows medical staff to better understand the changing trends of a woman's condition, evaluate treatment effectiveness, and even provide reference for similar cases in the future.

[0131] The intelligent monitoring terminal also provides data analysis tools, allowing medical staff to further analyze historical data. For example, medical staff can calculate statistical indicators such as average blood loss, maximum blood loss, and bleeding rate for postpartum hemorrhage, or analyze correlations between vital signs. These analysis results can help medical staff summarize experience and optimize postpartum hemorrhage monitoring and treatment plans.

[0132] By associating and storing charts with original monitoring data and supporting retrospective analysis of historical data, it not only provides medical staff with intuitive visualization tools, but also enhances data traceability and analysis capabilities, thereby improving the scientific nature and efficiency of medical decision-making.

[0133] The above-mentioned intelligent monitoring and prevention method for postpartum hemorrhage that integrates multiple parameters and AI has achieved accurate monitoring, rapid warning and efficient prevention and control of postpartum hemorrhage by collecting multivariate data and performing preprocessing, real-time assessment of bleeding risks, retrieving personalized treatment plans, real-time monitoring of treatment delays, and visual data storage and retrospective analysis, significantly improving the prevention and treatment effects of postpartum hemorrhage and the scientificity and timeliness of medical decision-making.

[0134] In order to further illustrate the solution of the embodiment of the present application, a specific example is given below.

[0135] Example 1, taking the actual application of the obstetric ward of a tertiary hospital as an example, illustrates the specific implementation of the present invention:

[0136] A 32-year-old woman, surnamed Zhang, was a G2P1 (previously a full-term birth and this was her second pregnancy) with a history of gestational hypertension. She underwent a natural birth after admission. Immediately after delivery, a disposable smart blood collection bag was placed under her body, and a fiber optic sensor began monitoring blood loss in real time. A vital sign patch was also applied to her chest. All devices used were designed with low power consumption and a 24-hour battery life, making them suitable for both transport and home postpartum monitoring.

[0137] Data acquisition and processing: A fiber optic sensor, using changes in light refraction to calculate the fluid level, determined cumulative blood loss of 280 ml in the 30 minutes following delivery. Fluid level fluctuations indicated a blood flow rate of 25 ml / min. After applying a Kalman filter to eliminate postural artifacts, the actual blood loss was determined to be 275 ± 5 ml. A vital sign patch revealed a decrease in blood pressure from 120 / 80 mmHg to 105 / 65 mmHg, an increase in heart rate from 85 bpm to 110 bpm, and a maintenance of oxygen saturation of 93%.

[0138] Dynamic risk assessment: The above data and the history of gestational hypertension in the electronic medical record (risk weight coefficient 0.3) are input into the AI ​​risk assessment model. The model recognizes the accelerated increase in blood loss through the LSTM layer (an increase of 150 ml in 10 minutes). Combined with the downward trend in blood pressure, the attention mechanism assigns a blood loss weight of 0.7 and a blood pressure weight of 0.8, and finally outputs a risk probability value of 0.68, which is judged to be a medium risk.

[0139] Grading warning and handling: Intermediate risk triggers the first level warning. A yellow warning box pops up on the monitoring screen at the nurse station, and the head nurse's tablet receives a push notification: "Bed number 5-2, blood loss 275ml (increase rate 15ml / min);

[0140] Recommendations: ① Expand the intravenous access; ② Urgently check the blood routine."

[0141] Failure to respond to escalation promptly and not receiving confirmation from the nurse after 15 minutes;

[0142] Automatic operation: ① Send a strong vibration reminder to the obstetrics and gynecology staff's mobile phone; ② The ward broadcasts "Bed 5-2, please deal with it quickly"; ③ The electronic medical record generates a red pop-up window.

[0143] Clinical intervention and feedback: After receiving the upgraded warning, the medical team immediately administered 10U of oxytocin intravenously and infused 500ml of crystalloid solution; the intelligent monitoring terminal detected that the blood loss rate dropped to 5ml / min and the blood pressure returned to 110 / 70mmHg. After 30 minutes, the risk level was reduced to low risk; all warning events, handling measures and vital signs change curves were automatically archived in the electronic medical record to form a complete timeline record.

[0144] Through data backtracking and optimization, the quality control department found in the "Early Warning Response Analysis Report" generated by the intelligent monitoring terminal that the average response time from the nurse station to the ward in this incident was 8 minutes. Based on this, the manpower allocation for obstetric night shifts was optimized.

[0145] In addition, an implementation method is provided, which uses multi-agent reinforcement learning to simulate the interaction between medical staff, equipment, and patients, and dynamically optimizes the warning trigger threshold and response strategy. For example, according to the current department load (such as insufficient manpower for night shifts), the warning upgrade is automatically adjusted (the high-level risk response time is extended from 5 minutes to 8 minutes), avoiding invalid alarms and reducing "alarm fatigue."

[0146] In this embodiment, objective data replaces subjective judgment, one-stop monitoring and real-time recording are used to support decision-making recommendations, thereby reducing the misdiagnosis rate and improving medical care efficiency, and reducing the incidence of serious complications through early warning.

[0147] Example 2, taking the application of B County Medical Community as an example.

[0148] After pregnant woman Li gave birth at the township health center, the data from the smart collection bag and vital signs patch were uploaded to the county hospital's HIS (Hospital Information System) in real time via the 5G network, automatically linking it to her history of "anemia during pregnancy" in her electronic medical record.

[0149] The county hospital's AI risk assessment model detected a high-risk condition, detecting blood loss of 400 ml within one hour and a drop in hemoglobin to 80 g / L. The model triggered two actions:

[0150] Local handling: The health center's sound and light alarms are activated, and the oxytocin + blood transfusion preparation plan is pushed to the local medical terminal;

[0151] Remote consultation: A consultation request is automatically generated through the HIS interface, and the county hospital expert’s mobile APP receives real-time vital sign flow data (including trend charts), and the video guides the health center to perform uterine compression.

[0152] Post-treatment data (blood transfusion volume, drug usage time) is entered by the health center and simultaneously updated to the HIS electronic medical record and provincial obstetric quality control platform, forming a closed loop of hierarchical diagnosis and treatment.

[0153] In this embodiment, by being compatible with hospital information systems, a county-level medical community is formed, remote consultations are supported, and the efficiency of responding to patient emergencies is improved.

[0154] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0155] Based on the same inventive concept, the embodiment of the present application also provides a system for intelligent monitoring and control of postpartum hemorrhage for realizing the multi-parameter integration and AI fusion involved above. The implementation scheme for solving the problem provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations in the embodiments of one or more multi-parameter integration and AI fusion postpartum hemorrhage intelligent monitoring and control systems provided below can be found in the above limitations of the multi-parameter integration and AI fusion postpartum hemorrhage intelligent monitoring and control method, which will not be repeated here.

[0156] In an exemplary embodiment, Figure 3 As shown, a multi-parameter integrated and AI-integrated intelligent monitoring and prevention system for postpartum hemorrhage 10 is provided, including:

[0157] A blood flow data acquisition module 11 is used to acquire blood flow data; the blood flow data includes blood loss and blood flow velocity, and the blood flow data is measured by an optical fiber sensor;

[0158] A vital sign information acquisition module 12 is used to acquire vital sign information; the vital sign information includes blood pressure, heart rate and blood oxygen saturation, and the vital sign information is continuously monitored by a vital sign patch;

[0159] An AI risk assessment model calculation module 13 is used to calculate the bleeding risk level assessment result in real time based on blood flow data, vital signs information, and medical history information through an AI risk assessment model. The AI ​​risk assessment model is used to dynamically assess bleeding risk;

[0160] The risk classification warning module 14 is used to trigger corresponding classification warnings according to the bleeding risk level assessment results. The classification warnings are used to indicate bleeding risk treatment.

[0161] In one embodiment, the blood flow data acquisition module is further configured to:

[0162] Obtaining liquid level height data in the collection bag, and calculating blood loss data based on the liquid level height data;

[0163] According to the dynamic change of liquid level height data, blood flow velocity data is obtained through synchronous calculation;

[0164] The data of blood loss volume and blood flow velocity are preliminarily processed to obtain preprocessed blood flow data.

[0165] In one embodiment, the AI ​​risk assessment model calculation module is further used to:

[0166] Real-time monitoring data on blood loss, blood flow rate, vital signs, and medical history characteristics are sent to a trained AI risk assessment model. The AI ​​risk assessment model then performs weighted analysis and comprehensive evaluation on the real-time monitoring data to obtain a bleeding risk level assessment result.

[0167] Based on the real-time monitoring data that is continuously sent, periodically updated risk assessment results are obtained.

[0168] In one embodiment, the AI ​​risk assessment model calculation module is further used to:

[0169] The risk probability value is generated according to the real-time monitoring data, and the risk probability value is generated by the calculation formula of the following risk quantification function:

[0170] R(t)=σ(W(t)*X(t)+b)

[0171] Where R(t) is the risk probability value, W(t) is the dynamic weight matrix, X(t) is the time series feature matrix, b is the bias term, and σ is the Sigmoid activation function;

[0172] Real-time risk grading is performed based on the risk probability value to obtain a bleeding risk level assessment result.

[0173] In one embodiment, the AI ​​risk assessment model calculation module is further used to:

[0174] Obtain real-time calculated bleeding risk assessment results, which include low, intermediate, and high levels;

[0175] According to the bleeding risk level assessment results, the corresponding treatment plan library is retrieved;

[0176] Based on the patient's medical history, select the appropriate medication list and surgical recommendations from the corresponding treatment plan library;

[0177] Combined with the medication list and surgical recommendations, the recommended treatment options and implementation steps are displayed in a prioritized manner;

[0178] A countdown reminder is triggered when the disposal plan and implementation steps are pushed, and the delay time is calculated in real time to display the delayed time.

[0179] In one embodiment, the AI ​​risk assessment model calculation module is further used to:

[0180] The early warning strategy is matched based on the bleeding risk level assessment results. When the risk level is intermediate, the first level warning is triggered, and a push notification is sent to the nursing mobile terminal based on the first level warning. When the risk level is high, the second level warning is triggered, and the sound and light alarm device is activated according to the second level warning and an emergency push notification is sent to the doctor's mobile terminal, generating an early warning prompt for timely risk handling.

[0181] Monitor the warning response status based on the early warning prompts for timely risk handling and obtain the early warning response results; the early warning response results are obtained through the early warning prompt feedback received by the sound and light alarm device or the designated mobile terminal;

[0182] Obtain the real-time monitoring of the early warning response status. If no confirmation feedback of the early warning response status is received within the preset time, the early warning level will be upgraded;

[0183] The warning trigger time, risk level and response results are synchronously stored in the electronic medical record system.

[0184] In one embodiment, any of the above modules is further configured to:

[0185] Acquiring multivariate data, including blood loss data, blood flow rate data, and vital sign information;

[0186] Perform time alignment, unit unification, and outlier filtering on the collected multi-dimensional data to obtain processed monitoring data;

[0187] Write the processed monitoring data into the hospital electronic medical record system according to the preset format to obtain time series data;

[0188] Draw blood loss curves and vital signs trend charts based on time series data;

[0189] For blood loss change curves and vital signs trend charts, the charts are associated with the original monitoring data and stored, and historical data retrospective analysis is supported.

[0190] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the power supply safety management method as described above when executing the computer program.

[0191] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0192] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0193] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A multi-parameter integrated and AI-integrated intelligent monitoring and prevention method for postpartum hemorrhage, characterized in that: The method comprises: Acquiring blood flow data; the blood flow data includes blood loss and blood flow velocity, and the blood flow data is obtained by measuring with an optical fiber sensor; Obtaining vital sign information; the vital sign information includes blood pressure, heart rate, and blood oxygen saturation, and the vital sign information is obtained through continuous monitoring of the vital sign patch; Based on the blood flow data, the vital signs information, and the medical history information, an AI risk assessment model is used to calculate a bleeding risk level assessment result in real time. The AI ​​risk assessment model is used to dynamically assess the bleeding risk; A corresponding graded warning is triggered according to the bleeding risk level assessment result, and the graded warning is used to indicate bleeding risk treatment.

2. The method according to claim 1, characterized in that The obtaining of blood flow data comprises: Obtaining liquid level data in the collection bag, and calculating blood loss data based on the liquid level data; Synchronously calculating according to the dynamic changes of the liquid level height data to obtain blood flow velocity data; The data on the amount of blood loss and the data on the blood flow rate are preliminarily processed to obtain the preprocessed blood flow data.

3. The method according to claim 1, characterized in that The method of calculating the bleeding risk level assessment result in real time based on the blood flow data, the vital signs information, and the medical history information through the AI ​​risk assessment model includes: The real-time monitoring data of blood loss, blood flow rate, vital signs and medical history characteristics are sent to the trained AI risk assessment model, and the real-time monitoring data are subjected to weighted analysis and comprehensive evaluation by the AI ​​risk assessment model to obtain the bleeding risk level assessment result; Based on the continuously sent real-time monitoring data, a periodically updated risk assessment result is obtained.

4. The method according to claim 3, characterized in that The AI ​​risk assessment model performs weighted analysis and comprehensive evaluation on the real-time monitoring data to obtain the bleeding risk level assessment result, including: The risk probability value is generated according to the real-time monitoring data, and the risk probability value is generated by the calculation formula of the following risk quantification function: R(t)=σ(W(t)*X(t)+b) Where R(t) is the risk probability value, W(t) is the dynamic weight matrix, X(t) is the time series feature matrix, b is the bias term, and σ is the Sigmoid activation function; Real-time risk grading is performed according to the risk probability value to obtain the bleeding risk level assessment result.

5. The method according to claim 3, characterized in that After obtaining the periodically updated risk assessment results based on the continuously sent real-time monitoring data, the method further includes: Obtaining the bleeding risk level assessment result calculated in real time, wherein the bleeding risk level assessment result includes low, medium and high levels; According to the bleeding risk level assessment result, the corresponding treatment plan library is retrieved; Based on the medical history characteristics, screening the applicable drug list and surgical recommendations from the corresponding treatment plan library; Combined with the drug list and surgical recommendations, the recommended treatment options and implementation steps are displayed in a prioritized manner; When the handling plan and implementation steps are pushed, a countdown reminder is triggered, the delay time is calculated in real time, and the delayed time is displayed.

6. The method according to claim 5, characterized in that After obtaining the bleeding risk level assessment result calculated in real time, the method further includes: An early warning strategy is matched based on the bleeding risk level assessment result. When the risk level is intermediate, a first-level warning is triggered, and a push notification is sent to the nursing mobile terminal according to the first-level warning. When the risk level is high, a second-level warning is triggered, and the activation of the sound and light alarm device is controlled according to the second-level warning, and an emergency push notification is sent to the doctor's mobile terminal to generate an early warning prompt for timely risk handling. Monitoring the early warning response status according to the early warning prompt for timely handling of the risk, and obtaining an early warning response result; the early warning response result is obtained by receiving the early warning prompt feedback through the sound and light alarm device or the designated mobile terminal; Acquire the early warning response status monitored in real time, and upgrade the early warning level if no confirmation feedback of the early warning response status is received within a preset time; The warning trigger time, risk level and response result are synchronously stored in the electronic medical record system.

7. The method according to any one of claims 1 to 6, characterized in that: Also includes: Acquiring multivariate data, the multivariate data including data on blood loss, blood flow rate, and vital sign information; Perform time alignment, unit unification, and outlier filtering on the collected multi-dimensional data to obtain processed monitoring data; Writing the processed monitoring data into the hospital electronic medical record system according to a preset format to obtain time series data; Drawing a blood loss change curve and a vital sign trend graph based on the time series data; For the blood loss change curve and vital signs trend chart, the chart is associated with the original monitoring data and stored, and historical data retrospective analysis is supported.

8. An intelligent monitoring and control system for postpartum hemorrhage integrating multi-parameters and AI, characterized by: The system comprises: A blood flow data acquisition module is used to acquire blood flow data; the blood flow data includes blood loss and blood flow velocity, and the blood flow data is obtained by measuring with an optical fiber sensor; A vital sign information acquisition module is used to acquire vital sign information; the vital sign information includes blood pressure, heart rate and blood oxygen saturation, and the vital sign information is continuously monitored by a vital sign patch; An AI risk assessment model calculation module is used to calculate a bleeding risk level assessment result in real time based on the blood flow data, the vital signs information, and the medical history information through an AI risk assessment model, wherein the AI ​​risk assessment model is used to dynamically assess the bleeding risk; The risk classification warning module is used to trigger a corresponding classification warning according to the bleeding risk level assessment result, and the classification warning is used to indicate the bleeding risk treatment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images