A method and system for prevention, intervention, and closed-loop risk feedback control of cerebral hemorrhage

Abstract

This invention discloses a closed-loop control method and system for prevention, intervention, and risk feedback in cerebral hemorrhage, relating to the field of cerebral hemorrhage risk intervention technology. It includes the following steps: S1, real-time acquisition of multi-source cerebrovascular monitoring data, data preprocessing, and identification of high-pressure peak events; S2, quantification of key dynamic characteristics of each peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak events, and assessment of high-pressure peak abnormalities; S3, fusion of multi-source cerebrovascular monitoring data and high-pressure peak abnormality assessment results, quantification of the risk probability of each peak event, identification of physiological fluctuations and risk events, and risk warning and response; S4, real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, assessment of warning accuracy, identification of missed and false alarms, and threshold optimization. This solves the technical problems of existing technologies, such as difficulty in distinguishing between physiological and pathological peaks, susceptibility to misjudgments, and impact on the timeliness and individualized management of risk intervention.

Description

Technical Field

[0001] This invention relates to the field of cerebral hemorrhage risk intervention technology, and in particular to a method and system for cerebral hemorrhage prevention intervention and risk feedback closed-loop control. Background Technology

[0002] With the aging population, rising incidence of chronic diseases, and accelerated pace of life, brain diseases, including cerebrovascular diseases, have become a major public health problem seriously threatening human health and safety. Cerebral hemorrhage, a common acute and critical condition following stroke, has a rapid onset, extremely high mortality and disability rates, and imposes a heavy burden on individuals, families, and society. In the treatment of brain diseases, timely and scientific risk identification and intervention are of great significance in reducing the incidence of cerebral hemorrhage and improving prognosis.

[0003] For example, invention patent CN116869503A discloses a microwave brain hemorrhage detection device, system, and method for monitoring brain hemorrhage status. The microwave brain hemorrhage detection device includes a wearable unit, a transmitting unit, several receiving units, and a control unit. The wearable unit is removably mounted on the user's head. The transmitting unit is located inside the wearable unit for transmitting microwave signals, and the several receiving units are located inside the wearable unit for receiving microwave signals. The control unit is located inside the wearable unit and connected to the transmitting unit and the several receiving units respectively. The transmitting unit and the several receiving units are arranged symmetrically around the user. Its advantages are: combining the transmitting unit, receiving unit, and control unit into the wearable unit significantly reduces the size of the brain hemorrhage detection device, making it easier for patients to wear; it is portable and miniaturized, making it easy for medical personnel to carry; and it is non-invasive, with continuous monitoring being harmless to humans.

[0004] For example, the invention patent with announcement number CN118436333B announces a non-invasive radiofrequency microwave acute brain hemorrhage detection system, belonging to the field of non-invasive brain imaging technology. It includes: a microwave transmitting module that sends microwave signals to the brain of the patient through multiple microwave transmitters; a microwave receiving module that receives echo signals reflected from the brain of the patient through multiple microwave receivers; and a signal processing module that performs power attenuation and frequency shift preprocessing on the echo signals, detects bleeding points in the brain based on the preprocessed echo signals, further analyzes the distance data between the intracranial bleeding points and the microwave receivers, performs mixing and spectrum analysis, fuses multiple acquired two-dimensional images to obtain a three-dimensional image of the patient's brain. Medical personnel can then determine the status of the brain hemorrhage based on the three-dimensional image and take appropriate medical measures. This solves the problems of complex operation and long detection time associated with traditional detection methods such as computed tomography (CT) scans, and improves the timeliness of patient treatment.

[0005] However, occasional blood pressure spikes may simply be temporary physiological stress responses or transient fluctuations caused by external factors, or they may conceal serious warning signs of fragile blood vessels and a sudden increase in the risk of cerebral hemorrhage. Traditional early warning methods are unable to effectively distinguish between physiological and pathological spikes, easily leading to misjudgments and missed diagnoses, thus affecting timely intervention and individualized management.

[0006] Therefore, in response to the above problems, there is an urgent need for a method and system for the prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage. Summary of the Invention

[0007] To address the technical problems of existing technologies, such as the difficulty in distinguishing between physiological and pathological peaks, the potential for misjudgment, and the impact on the timeliness of risk intervention and individualized management, this invention provides a method and system for closed-loop control of cerebral hemorrhage prevention, intervention, and risk feedback. The technical solution is as follows:

[0008] On the one hand, a closed-loop control method for prevention, intervention, and risk feedback in cerebral hemorrhage is provided. This method includes: S1, real-time acquisition of multi-source cerebrovascular monitoring data, noise filtering, outlier correction, missing value completion, and standard normalization preprocessing of the multi-source cerebrovascular monitoring data, and identification of high-pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data; S2, quantification of key dynamic characteristics of the peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak events, and assessment of high-pressure peak abnormalities based on the dynamic characteristics; S3, fusion of multi-source cerebrovascular monitoring data and high-pressure peak abnormality assessment results, quantification of the risk probability of each peak event, identification of physiological fluctuations and risk events based on the risk probability, and risk warning and response for different events; S4, real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, quantification of the warning effect, assessment of the warning accuracy, identification of missed and false alarms based on the warning effect, and threshold optimization.

[0009] Furthermore, the specific process of real-time acquisition of multi-source cerebrovascular monitoring data and preprocessing of the multi-source cerebrovascular monitoring data for noise filtering, outlier correction, missing data completion, and standardization is as follows: Real-time acquisition of multi-source cerebrovascular monitoring data, including blood pressure, heart rate, and blood oxygen saturation; removal of high-frequency noise and short-term spikes in the original multi-source cerebrovascular monitoring data using moving average, median filtering, and bandpass filtering; identification of outliers in the multi-source cerebrovascular monitoring data through physical threshold comparison, and replacement of outliers with the mean of the preceding and following sliding windows; completion of missing data using linear interpolation; standardization and dimensionless normalization of all multi-source cerebrovascular monitoring data; and establishment of a cerebral hemorrhage risk monitoring database, storing the original and preprocessed multi-source cerebrovascular monitoring data with timestamps in the cerebral hemorrhage risk monitoring database.

[0010] Furthermore, the specific process for identifying hypertension peak events based on preprocessed multi-source cerebrovascular monitoring data is as follows: statistically analyze blood pressure data and identify peak events: detect whether the blood pressure at each point is greater than the blood pressure values ​​of the n points before and after it; if so, record it as a peak; within the window before the peak, calculate the longest continuous sub-interval with a positive blood pressure slope as the rising segment; identify the interval after the peak where the blood pressure falls back to the blood pressure baseline threshold as the recovery segment; and write each peak event and the corresponding multi-source cerebrovascular monitoring data into the cerebral hemorrhage risk monitoring database.

[0011] Furthermore, based on the multi-source cerebrovascular monitoring data corresponding to the systolic peak event, the key dynamic characteristics of each peak are quantified, and the specific process of assessing systolic peak abnormalities based on these dynamic characteristics is as follows: Within the rising segment of the peak event, the slope is calculated for each pair of adjacent blood pressures, and the largest positive value is selected as the maximum rate value of the rising segment; within the recovery segment of the peak event, the slope is calculated for each pair of adjacent blood pressures, and the largest negative value is selected, with the absolute value taken as the maximum rate value of the recovery segment; within the sliding time window before the peak, the peak point is removed, and the standard deviation of the remaining blood pressure is calculated to obtain the baseline fluctuation value; the timestamps of the recovery segment endpoint and the rising segment start point are obtained, and the time interval is calculated to obtain the peak duration; the maximum rate value of the rising segment is multiplied by the peak duration to obtain the peak rise intensity factor; the product of the baseline fluctuation value and the maximum rate value of the recovery segment is added to the minimum constant value to obtain the fluctuation recovery adjustment factor; the peak rise intensity factor is divided by the fluctuation recovery adjustment factor to obtain the peak abnormality comprehensive value; the peak abnormality comprehensive value is calculated in real time after each peak event and then entered into the cerebral hemorrhage risk assessment process.

[0012] Further, the specific process of integrating multi-source cerebrovascular monitoring data with the evaluation results of high-pressure peak anomalies and quantifying the risk probability of each peak event is as follows: Real-time obtain multi-source cerebrovascular monitoring data and the comprehensive peak anomaly value, and execute the intracerebral hemorrhage risk assessment process: Based on a sliding time window, calculate the mean and standard deviation of the historical comprehensive peak anomaly value, and calculate the sum of the mean of the historical comprehensive peak anomaly value and twice the standard deviation as the peak anomaly quantile value; At the same time, obtain the historical heart rate and historical blood oxygen saturation based on the sliding time window, and calculate the mean of the historical heart rate and the mean of the historical blood oxygen saturation; For each peak event, obtain the maximum heart rate within a fixed sliding time window before and after the peak to get the maximum heart rate related to the peak, and obtain the blood oxygen saturation corresponding to the peak; Divide the current comprehensive peak anomaly value by the peak anomaly quantile value to get the peak anomaly normalization value; Divide the maximum heart rate related to the peak by the mean of the historical heart rate to get the heart rate normalization value; Divide the blood oxygen saturation corresponding to the peak by the mean of the historical blood oxygen saturation to get the normalized blood oxygen value, subtract the normalized blood oxygen value from the constant one to get the blood oxygen rapid decline activation term, if the result is less than 0, assign the blood oxygen rapid decline activation term as 0, otherwise assign the original calculation result; Add the blood oxygen rapid decline activation term to the constant one to get the blood oxygen modulation enhancement value; Multiply the peak anomaly normalization value, the heart rate normalization value, and the blood oxygen modulation enhancement value, and take the negative value as the exponential power for natural exponential operation to get the risk index fusion value; Subtract the risk index fusion value from the constant one to get the intracerebral hemorrhage risk probability value.

[0013] Further, the specific process of identifying physiological fluctuations and risk events based on the risk probability is as follows: Write the intracerebral hemorrhage risk probability value into the intracerebral hemorrhage risk monitoring database, and compare the intracerebral hemorrhage risk probability value with the multi-level risk thresholds P1 and P2: When < P1, it is determined as transient physiological fluctuations; when P1 ≤ < P2, it is determined as an intracerebral hemorrhage warning event; when ≥ P2, it is determined as an intracerebral hemorrhage dangerous event.

[0014] Further, the specific process of risk warning and response for different events is as follows: Conduct risk warnings for different situations. For transient physiological fluctuations, only make internal records and continuously monitor the intracerebral hemorrhage risk probability value; For intracerebral hemorrhage warning events, push warning information to the doctor side, and issue attention and reexamination reminders, increasing the monitoring frequency of multi-source cerebrovascular monitoring data; For intracerebral hemorrhage dangerous events, conduct multi-channel emergency warnings.

[0015] Furthermore, the specific process of real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, quantifying the warning effect, and evaluating the accuracy of the warning is as follows: After risk warning, continuously monitor multi-source cerebrovascular monitoring data and obtain blood pressure, heart rate, and blood oxygen saturation at the warning time; based on a sliding time window after risk warning, the difference between the maximum and minimum blood pressure values ​​is used to obtain the blood pressure variation amplitude, the difference between the maximum and minimum heart rate values ​​is used to obtain the heart rate variation amplitude, and the difference between the maximum and minimum blood oxygen saturation values ​​is used to obtain the blood oxygen saturation variation amplitude; the absolute values ​​of blood pressure, heart rate, and blood oxygen saturation at the warning time are added together to obtain the warning signal baseline value; the absolute values ​​of blood pressure variation amplitude, heart rate variation amplitude, and blood oxygen saturation variation amplitude are added together to obtain the signal variation value after the warning; the warning signal baseline value is divided by the signal variation value after the warning to obtain the warning effect evaluation value.

[0016] Furthermore, the specific process for identifying missed and false alarms based on the early warning effect and optimizing the threshold is as follows: After a risk warning, the early warning effect assessment value is calculated in real time and written into the cerebral hemorrhage risk monitoring database; if the early warning effect assessment value is less than the effect threshold, the multi-level risk threshold is reduced and the sampling frequency of multi-source cerebrovascular monitoring data is increased, and the multi-source cerebrovascular monitoring data before and after the risk warning is reviewed to identify whether there are any missed events; if the early warning effect assessment value is greater than or equal to the effect threshold, the existing multi-level risk threshold and sampling frequency are maintained, and the absolute values ​​of blood pressure change, heart rate change, and blood oxygen saturation change are analyzed to identify false alarms; all missed and false alarm events are statistically analyzed to form a source tracing report, including the time of the event, multi-source cerebrovascular monitoring data, and the identified cause, and pushed to the doctor's terminal for manual review and feedback, while Bayesian optimization and reinforcement learning algorithms are used to optimize the multi-level risk threshold and effect threshold.

[0017] On the other hand, a closed-loop control system for the prevention, intervention, and risk feedback of cerebral hemorrhage is provided. This system is applied to a closed-loop control method for the prevention, intervention, and risk feedback of cerebral hemorrhage, including: a data acquisition and high-pressure peak extraction module, used to acquire multi-source cerebrovascular monitoring data in real time, perform noise filtering, outlier correction, missing value completion, and standard normalization preprocessing on the multi-source cerebrovascular monitoring data, and identify high-pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data; a high-pressure peak monitoring quantification module, used to quantify the key dynamic characteristics of the peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak event, and evaluate the high-pressure peak abnormality based on the dynamic characteristics; a risk identification and classification module, used to fuse multi-source cerebrovascular monitoring data and high-pressure peak abnormality assessment results, quantify the risk probability of each peak event, identify physiological fluctuations and risk events based on the risk probability, and perform risk warnings and responses for different events; and a risk feedback and optimization module, used to monitor multi-source cerebrovascular monitoring data in real time after risk warning, quantify the warning effect, evaluate the warning accuracy, identify missed and false alarms based on the warning effect, and optimize the threshold.

[0018] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following:

[0019] (1) This invention integrates multi-source cerebrovascular monitoring data of blood pressure, heart rate, and blood oxygen saturation, and achieves high-quality raw data acquisition through multi-level filtering, anomaly correction, and normalization preprocessing. Furthermore, it quantifies each peak event based on the multi-dimensional characteristics of dynamic blood pressure peak rise rate, duration, recovery rate, and individual variability. Compared to traditional single-indicator early warning methods, this significantly improves the accuracy of identifying abnormal high-risk events and the reliability of the data foundation.

[0020] (2) This invention constructs a risk probability calculation model based on the fusion of peak anomaly comprehensive value and multimodal signs, which realizes the quantitative output of the risk probability of each blood pressure peak event and automatically matches the corresponding early warning response strategy according to different risk levels. It can accurately distinguish between transient physiological fluctuations, cerebral hemorrhage warnings and dangerous events, effectively improving the individualization, grading and clinical operability of early warning.

[0021] (3) This invention constructs an early warning effect assessment value, regularly statistically analyzes and dynamically examines the actual response intensity of multi-source vital signs after an early warning, and identifies and archives false alarms and missed alarms. Based on the assessment results, it can intelligently adjust multi-level risk thresholds and sampling frequencies, continuously optimize early warning parameters and strategies, realize a closed loop of risk monitoring, feedback and optimization, and effectively improve self-learning ability and early warning reliability.

[0022] (4) This invention, by combining machine learning algorithms such as Bayesian optimization and reinforcement learning, continuously adaptively optimizes the threshold based on large-sample historical data and real-time closed-loop feedback, thereby improving the sensitivity and specificity of risk warning. It generates traceability reports to support clinical review and quality control, providing data support and decision-making tools for intelligent prevention, early identification, and scientific management of high-risk groups for cerebral hemorrhage. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a closed-loop control method for prevention, intervention, and risk feedback of cerebral hemorrhage provided in an embodiment of the present invention;

[0025] Figure 2 This is a structural diagram of a closed-loop control system for prevention, intervention, and risk feedback of cerebral hemorrhage provided in an embodiment of the present invention;

[0026] Figure 3 This is a blood pressure peak time-series curve provided in an embodiment of the present invention;

[0027] Figure 4 This is a comparison chart of peak anomaly comprehensive values ​​provided in the embodiments of the present invention. Detailed Implementation

[0028] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0029] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0030] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0031] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0032] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0033] This invention provides a method for prevention, intervention, and closed-loop risk feedback control of cerebral hemorrhage, such as... Figure 1 The flowchart shown is a closed-loop control method for prevention, intervention, and risk feedback in cerebral hemorrhage. The processing flow of this method may include the following steps: S1, real-time acquisition of multi-source cerebrovascular monitoring data, noise filtering, outlier correction, missing value completion, and standard normalization preprocessing of the multi-source cerebrovascular monitoring data, and identification of high-pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data; S2, quantification of key dynamic characteristics of the peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak events, and assessment of high-pressure peak abnormalities based on the dynamic characteristics; S3, fusion of multi-source cerebrovascular monitoring data and high-pressure peak abnormality assessment results, quantification of the risk probability of each peak event, identification of physiological fluctuations and risk events based on the risk probability, and risk warning and response for different events; S4, real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, quantification of the warning effect, assessment of the warning accuracy, identification of missed and false alarms based on the warning effect, and threshold optimization.

[0034] Optionally, the real-time acquisition of multi-source cerebrovascular monitoring data, and the specific process of preprocessing the multi-source cerebrovascular monitoring data for noise filtering, outlier correction, missing value completion, and standard normalization are as follows: Real-time acquisition of multi-source cerebrovascular monitoring data, including blood pressure, heart rate, and blood oxygen saturation; removal of high-frequency noise and short-term spikes from the original multi-source cerebrovascular monitoring data using moving average, median filtering, and bandpass filtering; Moving average filtering refers to smoothing the average of data within a certain time window centered on each moment of continuous data acquisition, effectively reducing the influence of high-frequency noise; median filtering involves taking the median of all sampled values ​​within the same time window to suppress isolated outliers; bandpass filtering, by setting a reasonable frequency range, retains only data components in the signal that correspond to normal physiological fluctuation frequencies, filtering out interference exceeding the normal range. Short-term spikes typically manifest as single-point extreme anomalies, which are eliminated through the above filtering steps. Abnormal points in multi-source cerebrovascular monitoring data are identified by comparing physical thresholds. This refers to combining clinically recognized normal physiological ranges, such as normal systolic blood pressure of 90 to 140 mmHg, heart rate of 60 to 100 bpm, and blood oxygen saturation of 95% to 100% for adults. Sampling points exceeding the physical thresholds are automatically identified as abnormal values, and the average of the previous and subsequent sliding windows is used to replace the abnormal points, thereby ensuring the continuity and rationality of the signal and avoiding the impact of abnormal values ​​on subsequent analysis. To address potential data loss and sampling omissions in the acquisition process, a maximum acceptable missing rate setting is supported, preferably no more than 5% within a single sliding window. This means that a maximum of 5% of data points are allowed to be missing within a single window. When missing sampling data is detected at a certain moment, linear interpolation is used to fill in the missing data using adjacent data points before and after that point. All multi-source cerebrovascular monitoring data undergoes standardization and dimensionless normalization to eliminate the influence of different units and dimensions on various features, ensuring the accuracy and stability of subsequent modeling and analysis. A cerebral hemorrhage risk monitoring database is established, with timestamps attached to the raw and preprocessed multi-source cerebrovascular monitoring data. The timestamp accuracy is preferably no less than 1 second. This database is a structured electronic data storage system that can be implemented using local servers and cloud platforms, supporting batch data reading, retrieval, and export.

[0035] This implementation plan effectively ensures the high quality and completeness of critical cerebrovascular vital sign data by integrating real-time acquisition of multi-source cerebrovascular monitoring data, signal and noise filtering, outlier correction, missing data completion, and standardized normalization preprocessing. Systematic data cleaning and completion significantly improve the accuracy and stability of subsequent risk analysis, providing a solid data foundation for the early identification and dynamic monitoring of high-risk events such as cerebral hemorrhage. Simultaneously, by establishing a structured cerebral hemorrhage risk monitoring database, it supports efficient storage, retrieval, and retrospection of multi-source data, providing strong support for individualized risk assessment, model training, and intelligent intervention, significantly enhancing the scientific rigor and practicality of cerebral hemorrhage early warning and risk closed-loop management.

[0036] Optionally, the specific process for identifying hypertension peak events based on preprocessed multi-source cerebrovascular monitoring data is as follows: Statistical analysis of blood pressure data and identification of peak events: Detect whether the blood pressure at each point is greater than the blood pressure values ​​of the preceding and following n points; if so, record it as a peak. Here, n is a positive integer representing the number of comparison sampling points required for peak determination. The specific value of n can be flexibly set according to the sampling frequency of the monitoring equipment, the period of blood pressure fluctuation, and the detection sensitivity. Preferably, the value of n ranges from 2 to 10, with commonly used values ​​from 3 to 5. When the sampling frequency is high (≥1 time / second) and blood pressure fluctuations are slow, a larger n value can be selected to enhance robustness and reduce false detections; when the sampling frequency is low or a brief peak needs to be captured, the n value can be appropriately reduced to improve sensitivity. Within the window before the peak, the longest continuous sub-interval with a positive blood pressure slope is calculated as the rising segment; the slope is calculated by dividing the difference between the blood pressure values ​​of two adjacent sampling points by the sampling time interval. If the slope of multiple consecutive points is positive, it is considered a continuous rise. The starting point of the longest continuous sub-interval is the initial rising point of this peak, and the ending point is the peak occurrence point. The recovery period is defined as the interval during which blood pressure falls back to the baseline threshold after the peak. The recovery period is the time interval during which blood pressure continues to decline after the peak and eventually falls to or below the baseline threshold. Identifying the recovery period helps measure the self-recovery capacity and ongoing risk following blood pressure abnormalities. Each peak event and its corresponding multi-source cerebrovascular monitoring data are entered into a cerebral hemorrhage risk monitoring database.

[0037] In this implementation plan, based on preprocessed high-quality multi-source cerebrovascular monitoring data, a method combining sliding window extreme value detection and dynamic feature extraction can accurately and efficiently identify each blood pressure peak event, automatically distinguish between the rising and recovery phases, and quantitatively reflect the entire process of blood pressure abnormality occurrence, development, and recovery. By recording peak events and their associated multi-source vital signs in detail, precise capture and dynamic data retention of high-risk blood pressure fluctuations are achieved, providing a scientific basis for early identification, risk trend assessment, and individualized intelligent intervention for high-risk events such as cerebral hemorrhage.

[0038] Optionally, based on multi-source cerebrovascular monitoring data corresponding to the systolic peak event, the specific process of quantifying key dynamic characteristics of the peak and assessing systolic peak abnormalities based on these dynamic characteristics is as follows: Within the rising segment of the peak event, the slope is calculated for each pair of adjacent blood pressures, and the maximum positive value is selected. The maximum positive value represents the fastest rate of blood pressure rise during the peak rise, reflecting the abruptness of the abnormal event, and is used as the maximum rate value of the rising segment. Within the recovery segment of the peak event, the slope is calculated for each pair of adjacent blood pressures, and the maximum negative value, i.e., the fastest rate of blood pressure decrease during the recovery phase, is selected. The absolute value is taken as the maximum rate value of the recovery segment. The absolute value is used to characterize the blood pressure self-regulation and recovery ability. The larger the value, the faster the decline, and the relatively controllable risk; the smaller the value, the slower the recovery, and the need to be wary of persistent risks. Within the sliding time window before the peak, the peak point is removed to avoid interference from abnormal values ​​on the baseline assessment, and the standard deviation of the remaining blood pressure is calculated. The process involves obtaining baseline fluctuation values; acquiring the timestamps of the recovery phase endpoint and the ascent phase start point, calculating the peak duration (the total time from the start of the ascent phase to the end of the recovery phase), reflecting the sustained risk of abnormal blood pressure; multiplying the peak duration by the peak rate value to obtain the peak rise intensity factor, which comprehensively reflects the suddenness and cumulative effect of blood pressure abnormalities; a higher value indicates a higher risk; adding the product of the baseline fluctuation value and the maximum recovery rate value to a minimum constant value yields the fluctuation recovery adjustment factor; a higher value indicates greater individual fluctuation and faster recovery, helping to dilute the risk; a lower value indicates a potentially amplified risk; the minimum constant value is set to 0.1; dividing the peak rise intensity factor by the fluctuation recovery adjustment factor yields the peak anomaly composite value; this is a dimensionless risk quantification indicator, with a higher value representing a higher degree of abnormality and risk level in the systolic blood pressure peak. The peak anomaly composite value is calculated in real-time after each peak event, synchronously written into the cerebral hemorrhage risk monitoring database, and entered into the cerebral hemorrhage risk assessment process.

[0039] The specific formula for the comprehensive value of peak anomalies is as follows:

[0040] ;

[0041] In the formula, The peak abnormality composite value is used to quantify the degree of abnormality of each blood pressure peak. It comprehensively considers multiple factors such as the rise rate, duration, individual blood pressure variability, and recovery speed of the blood pressure peak. The larger the peak abnormality composite value, the higher the risk of peak abnormality, which requires special attention and early warning. It reflects the dynamic balance between the danger intensity of the peak event and the individual's basic variability and self-recovery ability. This indicates the maximum rate of increase during the rising phase, reflecting the fastest rate of blood pressure rise during this phase. It measures whether the blood pressure abnormality is sudden and rapid; the larger the value, the faster the blood pressure rises and the higher the potential danger. This indicates the duration of the peak. The longer the duration, the longer the abnormal state persists and the greater the accumulated risk. This represents the baseline fluctuation value, reflecting the degree of blood pressure fluctuation in an individual under normal conditions before the peak. Large fluctuations mean that the individual's blood pressure is unstable and the risk of occasional peaks is relatively small, while small fluctuations mean that occasional peaks are more unusual. This represents the maximum rate of recovery, indicating the fastest speed at which blood pressure returns to normal after the peak. The faster the recovery, the stronger the individual's self-regulation ability; a slow recovery indicates that the danger persists. This represents a very small constant value, ensuring that the denominator is not zero, and its value is 0.1; The peak rise intensity factor is used to comprehensively measure the intensity of abnormal blood pressure events: they are both sudden and prolonged, and the higher the value, the more dangerous they are. This represents the volatility recovery adjustment factor, which comprehensively reflects an individual's basic volatility and post-abnormal recovery ability: the greater the volatility, the faster the recovery, and a large denominator means the risk is diluted; the smaller the volatility, the slower the recovery, and a small denominator means the risk is amplified.

[0042] In this embodiment, Table 1 is a comprehensive value data table of peak anomalies. The table details the maximum rise rate, peak duration, baseline fluctuation, maximum recovery rate, and comprehensive peak anomaly value for each of the five peak events. Specifically, peak event 1 has a maximum rise rate of 7.8, peak duration of 2.2, baseline fluctuation of 2.1, maximum recovery rate of 6.4, and comprehensive peak anomaly value of 1.27; peak event 2 has a maximum rise rate of 12.5, peak duration of 3.3, baseline fluctuation of 2.9, maximum recovery rate of 7.7, and comprehensive peak anomaly value of 1.84; peak event 3... The corresponding maximum rate of the rising segment is 5.6, the peak duration is 1.5, the baseline fluctuation is 1.7, the maximum rate of the recovery segment is 5.3, and the peak anomaly composite value is 0.92; the maximum rate of the rising segment corresponding to peak event 4 is 14.2, the peak duration is 4.1, the baseline fluctuation is 2.3, the maximum rate of the recovery segment is 10.8, and the peak anomaly composite value is 2.33; the maximum rate of the rising segment corresponding to peak event 5 is 7.3, the peak duration is 2.6, the baseline fluctuation is 2.4, the maximum rate of the recovery segment is 6.9, and the peak anomaly composite value is 1.14.

[0043] Table 1. Comprehensive Value Data of Peak Anomalies

[0044]

[0045] like Figure 3The figure shows a time-series curve of blood pressure peaks. It displays the blood pressure changes over time for five different peak events. The horizontal axis represents time, and the vertical axis represents blood pressure. Each curve represents the blood pressure peak process under a specific set of parameters, and the colors distinguish different events. The overall curves show that: Peak event 4 exhibits the highest and widest peak, with the largest maximum rate of ascent, peak duration, and peak abnormality composite value, indicating the most rapid and prolonged increase in blood pressure, with a significantly higher risk than the other events; Peak event 2 is the second highest, with a distinct peak, a relatively rapid increase in blood pressure, and a relatively long duration; Peak events 1 and 5 have moderate increases in blood pressure amplitude and duration, with relatively mild peaks; Peak event 3 exhibits the lowest and narrowest peak, with a slower rise and faster recovery, the smallest overall blood pressure fluctuation amplitude, and a lower risk.

[0046] like Figure 4 The image shows a comparison chart of peak anomaly comprehensive values. The horizontal axis represents the peak event number, and the vertical axis represents the peak anomaly comprehensive value. Each bar height corresponds to a peak anomaly comprehensive value. (Based on Table 1 and...) Figure 4 It can be seen that the peak abnormality composite value of peak event 4 is the highest, indicating that the composite rise rate is fast, the duration is long and the recovery is slow, and the degree of blood pressure abnormality is the highest; peak event 2, peak event 1, and peak event 5 are next; peak abnormality composite value of peak event 3 is the lowest, the rise rate and duration are both small, the peak shape is low and the recovery is fast, which is the lowest risk peak and is clearly distinguishable from high-risk pathological events.

[0047] This implementation plan achieves real-time, quantitative, and comprehensive assessment of blood pressure abnormality risk by finely quantifying key dynamic characteristics of systolic blood pressure peak events and combining multidimensional parameters such as rise rate, duration, baseline fluctuation, and recovery rate. This effectively enhances the early identification and tiered management capabilities for high-risk events of cerebral hemorrhage. The fully automated extraction and comprehensive calculation of dynamic characteristics not only improves the sensitivity and intelligence of monitoring but also provides a solid data foundation and theoretical support for individualized intervention, risk-based response, and subsequent big data analysis.

[0048] Optionally, the specific process of integrating multi-source cerebrovascular monitoring data and high-pressure peak anomaly assessment results to quantify the risk probability of each peak event is as follows: Real-time acquisition of multi-source cerebrovascular monitoring data and peak anomaly composite values; execution of the cerebral hemorrhage risk assessment process: Based on a sliding time window, calculation of the mean and standard deviation of historical peak anomaly composite values, which helps reflect an individual's normal fluctuation range; calculation of the sum of the mean and twice the standard deviation of historical peak anomaly composite values ​​as the peak anomaly quantile value, with twice the standard deviation serving as a common high-risk event boundary reference; simultaneous acquisition of historical heart rate and historical blood oxygen saturation based on a sliding time window, and calculation of the historical mean heart rate and historical mean blood oxygen saturation; for each peak event, the maximum heart rate within a fixed sliding time window before and after the peak is taken to obtain the peak-related maximum heart rate. The value measures the intensity of the body's stress response at the time of the event and obtains the blood oxygen saturation corresponding to the peak. The peak abnormality standardized value is obtained by dividing the current peak abnormality composite value by the peak abnormality percentile; a larger value indicates a more severe deviation of the current peak from the normal baseline. The peak-related maximum heart rate is divided by the historical average heart rate to obtain the standardized heart rate value, highlighting the contribution of a sharp increase in heart rate to risk amplification. The peak-corresponding blood oxygen saturation is divided by the historical average blood oxygen saturation to obtain the standardized blood oxygen value. Subtracting the standardized blood oxygen value from a constant yields the blood oxygen drop activation term. If the result is less than 0, the blood oxygen drop activation term is set to 0; otherwise, it is set to the original calculation result. The blood oxygen drop activation term reflects the decrease in blood oxygen saturation relative to the baseline at the peak and is activated only when actual blood oxygen levels decrease, avoiding reverse misjudgment. If there is no decrease, it is 0, ensuring that subsequent risk amplification is only effective for actual hypoxic conditions. The activation term for a rapid drop in blood oxygenation is added to a constant to obtain the blood oxygen modulation enhancement value. A blood oxygen modulation enhancement value greater than 1 indicates a significant drop in blood oxygenation, which ultimately amplifies the output of the risk probability, reflecting a multi-source synergistic amplification mechanism. The risk index fusion value is obtained by multiplying the peak abnormality value, the heart rate standardized value, and the blood oxygen modulation enhancement value, and taking the negative value as the exponent for natural exponentiation. The risk index fusion value is obtained by subtracting the constant from the risk index fusion value; the larger the value, the higher the risk probability of cerebral hemorrhage.

[0049] The specific formula for the probability value of cerebral hemorrhage is as follows:

[0050] ;

[0051] In the formula, It represents the probability value of cerebral hemorrhage risk, which is used to fuse multiple physiological signals and perform risk amplification processing to quantitatively identify and output the probability of each cerebral hemorrhage risk event in real time; This indicates the overall abnormality value of the current blood pressure peak, reflecting the overall intensity of the current abnormality. The larger the value, the more severe the abnormality. Indicates the peak anomaly quantile value; Represents the maximum heart rate related to the wave peak, reflecting the stress response. The higher it is, the more intense the body's reaction; Represents the historical average heart rate, a standardized heart rate feature that eliminates individual differences; Represents the blood oxygen saturation corresponding to the wave peak. The lower it is, the higher the risk; Represents the historical average blood oxygen saturation, serving as a reference point to highlight the relative decline effect; Represents the abnormal standardized value of the wave peak, measuring the deviation of the current wave peak abnormality from the individual's normal range. The larger the value, the rarer and more worthy of vigilance this abnormality is; Represents the heart rate standardized value, which amplifies the additional risk brought by the heart rate stress response during blood pressure abnormalities, eliminates the basic differences, and makes the risk event assessment more accurate; Represents the blood oxygen modulation enhancement value. Only when the blood oxygen saturation is significantly lower than the historical average blood oxygen saturation, the blood oxygen modulation enhancement value is greater than 1, thereby amplifying the final probability output.

[0052] In this implementation plan, through the integration of multi-source cerebrovascular monitoring data and comprehensive assessment of high-pressure wave peak abnormalities, the multi-dimensional feature quantification of high-risk events of cerebral hemorrhage and real-time discrimination of risk probability are achieved. By dynamically updating the individual historical baseline through a sliding window, combined with the dynamic standardization and collaborative amplification mechanism of blood pressure, heart rate, and blood oxygen saturation, not only is the ability to distinguish sudden pathological wave peaks and physiological fluctuations greatly improved, but the sensitivity and accuracy of risk assessment are also significantly enhanced. The risk probability quantification results provide a scientific basis for clinical grading early warning, intelligent intervention decision-making, and personalized management, greatly improving the practicality and prospectiveness of cerebral hemorrhage risk monitoring, and laying a solid foundation for achieving early precise warning and closed-loop health management.

[0053] Optionally, the specific process of identifying physiological fluctuations and risk events based on the risk probability is as follows: Write the cerebral hemorrhage risk probability value into the cerebral hemorrhage risk monitoring database, and Compare it with the multi-level risk thresholds P1 and P2. P1 is the dividing line between physiological fluctuations and pathological events, and P2 is the dividing line between general risks and high-risk risks: When < P1, it is determined as transient physiological fluctuations; when P1 ≤ < P2, it is determined as a cerebral hemorrhage warning event; when ≥ P2, it is determined as a cerebral hemorrhage dangerous event.

[0054] This implementation plan, by introducing a multi-level risk probability threshold comparison and automatic classification mechanism, can accurately distinguish complex cerebral hemorrhage risk events into transient physiological fluctuations, general risk warnings, and high-risk events, achieving dynamic identification and scientific hierarchical management of different risk levels. This not only improves the rapid response capability to high-risk events and reduces misjudgments and omissions, but also greatly enhances the systematicness and traceability of event records, providing strong data support for subsequent individualized risk optimization and intervention.

[0055] Optionally, the specific process for risk warning and response for different events is as follows: For different situations, risk warnings are issued. For transient physiological fluctuations, only internal recording is performed, and the probability value of cerebral hemorrhage risk is continuously monitored to ensure real-time tracking of physiological fluctuation trends. For cerebral hemorrhage warning events, warning information is pushed to the doctor's end, along with attention and follow-up reminders, ensuring that relevant medical personnel are aware of the risk signal immediately. The attention and follow-up reminders are automatically generated operation instructions, prompting clinicians to conduct key follow-ups on patients. The monitoring frequency of multi-source cerebrovascular monitoring data is increased. For dangerous cerebral hemorrhage events, multi-channel emergency warnings are implemented. These multi-channel emergency warnings include: automatically triggered warning information push, highlighted pop-ups at nurses' and doctors' workstations, and simultaneous alarms to the emergency green channel command platform, ensuring that the medical team at multiple levels and with multiple roles can receive high-risk alerts in the shortest possible time, facilitating rapid activation of the emergency response process and minimizing delays and risks. To ensure timely and reliable alarm response, the emergency warning response time should not exceed 10 seconds, and it should at least support in-hospital app push notifications, SMS notifications, and linkage with audible and visual alarms. Each channel must have high reliability and real-time performance to prevent single-point failure from causing alarm loss. All alarm information should include the patient's unique identifier, event level, trigger time, probability value of cerebral hemorrhage risk, and recommended intervention measures.

[0056] This implementation plan significantly improves the efficiency of early identification and emergency response to cerebral hemorrhage risk events by setting multi-level risk grading responses, strict response time limits, and multi-channel early warning capabilities. For events of different levels, it can automatically complete tiered recording, tiered push notifications, and dynamically adjust sampling density, achieving closed-loop management of the entire process from routine monitoring to emergency treatment. Especially in high-risk events, it supports simultaneous early warning through multiple channels within the hospital, greatly enhancing the timeliness, accuracy, and clinical usability of the warnings. Overall, it greatly improves the security, scalability, and practical application value of intelligent early warning for cerebral hemorrhage risk.

[0057] Optionally, the specific process of real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, quantifying the warning effect, and evaluating the accuracy of the warning is as follows: After risk warning, continuously monitor multi-source cerebrovascular monitoring data and obtain blood pressure, heart rate, and blood oxygen saturation at the warning time; based on a sliding time window after the risk warning (the sliding time window can be flexibly adjusted according to actual clinical needs), the difference between the maximum and minimum blood pressure values ​​is used to obtain the blood pressure variation, the difference between the maximum and minimum heart rate values ​​is used to obtain the heart rate variation, and the difference between the maximum and minimum blood oxygen saturation values ​​is used to obtain the blood oxygen saturation variation; the data at the warning time is then... The baseline value of the warning signal is obtained by adding the absolute values ​​of blood pressure, heart rate, and blood oxygen saturation at the warning time, which reflects the overall intensity of the individual's physiological state at the moment the warning is triggered. The absolute values ​​of blood pressure, heart rate, and blood oxygen saturation changes are added together to obtain the signal change value after the warning, which represents the overall fluctuation range of various physiological parameters after the warning. The larger the value, the more obvious the physiological response of the individual after the warning. The warning effect assessment value is obtained by dividing the baseline value of the warning signal by the signal change value after the warning, which is used to quantify the actual physiological impact of the risk warning.

[0058] The specific formula for evaluating the early warning effect is as follows:

[0059] ;

[0060] In the formula, The warning effect assessment value is used to quantify the dynamic changes of core physiological signals within the follow-up window after each risk warning and compare them with the signal baseline at the time of the warning. In other words, it evaluates whether the risk warning captured the subsequent high-risk physiological fluctuations. It is used for risk feedback, adaptive adjustment and warning effectiveness assessment. The larger the warning effect assessment value, the more effective the warning is. The smaller the warning effect assessment value, the less timely the subsequent physiological abnormalities are captured, and there is a risk of lag and underreporting. Blood pressure at the warning point; Heart rate at the time of warning; Indicates blood oxygen saturation at the time of the warning; Indicates the range of blood pressure fluctuations; Indicates the range of heart rate variation; This indicates the range of change in blood oxygen saturation; It represents the baseline value of the warning signal, reflects the overall physiological load level at the time of the warning, and is the total amount of immediate vital signs of the risk signal; This indicates the change in signal value after the warning, reflecting the total fluctuation range of physiological signals over a period of time after the warning, i.e., the actual dynamic response intensity of the risk signal.

[0061] This implementation plan, through continuous high-frequency collection and real-time analysis of multi-source cerebrovascular monitoring data after risk warning, constructs a warning effect assessment value, thereby quantifying and validating the effectiveness of the warning response. By meticulously statistically analyzing and comparing the dynamic fluctuations of key physiological indicators such as blood pressure, heart rate, and blood oxygen saturation, it can accurately distinguish between high-risk events and general fluctuations, effectively identify false alarms and missed alarms, and improve the accuracy and intelligence of the warning system. This significantly enhances the practicality and reliability of closed-loop risk management for cerebral hemorrhage.

[0062] Optionally, the specific process for identifying missed and false alarms based on the early warning effect and optimizing the threshold is as follows: After a risk warning, the early warning effect assessment value is calculated in real time and written into the cerebral hemorrhage risk monitoring database; if the early warning effect assessment value is less than the effect threshold, the multi-level risk threshold is reduced, and the sampling frequency of multi-source cerebrovascular monitoring data is increased to improve sensitivity and reduce subsequent missed detections; multi-source cerebrovascular monitoring data before and after the risk warning is reviewed to identify whether there are missed events: based on sliding time window analysis of multi-source cerebrovascular monitoring data, automatic detection is performed to identify events where signals have undergone abrupt changes without warning, and such changes exceed twice the standard deviation of the historical distribution; if the early warning effect assessment value is greater than or equal to the effect threshold, the existing multi-level risk threshold and sampling frequency are maintained, and the absolute values ​​of blood pressure fluctuations, heart rate fluctuations, and blood oxygen saturation fluctuations are analyzed to identify false alarm events: if the blood pressure fluctuation after the risk warning is greater than or equal to the effect threshold, the false alarm events are identified. False alarms occur when the absolute values ​​of temperature, heart rate fluctuations, and blood oxygen saturation fluctuations are all below their respective minimum fluctuation thresholds, and no external intervention is required. A source tracing report is generated from all missed and false alarm events, including the time of occurrence, multi-source cerebrovascular monitoring data, and identified causes. This report is then sent to the doctor's terminal for manual review and feedback. Simultaneously, Bayesian optimization and reinforcement learning algorithms are used to optimize multi-level risk and effect thresholds. The optimization objective function can be set to minimize the weighted sum of the risk missed and false alarm rates, while maximizing the warning recall and accuracy. Several safety constraints are introduced during the optimization process, including: the overall false alarm rate must not exceed a pre-set maximum safety threshold, such as 5%; the adjustment range of multi-level risk and effect thresholds must not exceed 5% each time to prevent drastic threshold fluctuations leading to instability or frequent false alarms; threshold optimization is only performed in batches at set minimum intervals to prevent overfitting and data drift. All automatic update history and optimization results are fully recorded, supporting manual review and manual intervention rollback at any time.

[0063] This implementation scheme achieves intelligent adjustment and self-evolution of risk and effect thresholds through a dynamic closed-loop adaptive mechanism based on the early warning effect, combined with automatic identification of missed and false alarms, dynamic threshold optimization, and strict safety constraints. It can automatically balance sensitivity and specificity based on real-time assessment and historical traceability, continuously improving the accuracy of identifying missed and false alarms. Simultaneously, batch optimization and safety margin limits ensure monitoring stability and robustness. All parameter updates and optimization histories are traceable and support manual intervention, enabling the risk warning system to possess both intelligent adaptability and clinical controllability. Overall, it significantly improves the scientific rigor, sensitivity, and sustainable optimization capabilities of closed-loop risk management for cerebral hemorrhage.

[0064] like Figure 2 As shown, another aspect of the present invention provides a closed-loop control system for the prevention, intervention, and risk feedback of cerebral hemorrhage. This system is applied to a closed-loop control method for the prevention, intervention, and risk feedback of cerebral hemorrhage, comprising: a data acquisition and high-pressure peak extraction module, used to acquire multi-source cerebrovascular monitoring data in real time, perform noise filtering, outlier correction, missing value completion, and standard normalization preprocessing on the multi-source cerebrovascular monitoring data, and identify high-pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data; a high-pressure peak monitoring quantification module, used to quantify the key dynamic characteristics of the peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak event, and evaluate the high-pressure peak abnormality based on the dynamic characteristics; a risk identification and classification module, used to fuse multi-source cerebrovascular monitoring data and high-pressure peak abnormality assessment results, quantify the risk probability of each peak event, identify physiological fluctuations and risk events based on the risk probability, and perform risk warning and response for different events; and a risk feedback and optimization module, used to monitor multi-source cerebrovascular monitoring data in real time after risk warning, quantify the warning effect, evaluate the warning accuracy, identify missed and false alarms based on the warning effect, and optimize the threshold.

[0065] This implementation plan achieves fully automated processing, dynamic risk quantification, and intelligent closed-loop feedback of multi-source cerebrovascular monitoring data. Through high-precision data acquisition and high-pressure peak identification, multi-feature quantification of abnormal events, risk probability classification, and threshold adaptive optimization based on early warning effects, a complete closed loop is constructed, from data monitoring to intelligent early warning and continuous self-learning optimization. This not only significantly improves the early detection, dynamic tracking, and individualized risk response capabilities for high-risk events such as cerebral hemorrhage, but also effectively reduces the rates of missed and false alarms, significantly improving the scientific rigor, sensitivity, and practicality of risk management, providing solid technical support for intelligent health management.

[0066] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0067] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0068] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0069] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0070] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0071] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0072] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0073] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0075] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] 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. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for prevention, intervention, and closed-loop risk feedback control of cerebral hemorrhage, characterized in that, The method includes: S1 collects multi-source cerebrovascular monitoring data in real time, performs noise filtering, outlier correction, missing value completion and standard normalization preprocessing on the multi-source cerebrovascular monitoring data, and identifies high pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data. S2, based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak event, quantify the key dynamic characteristics of each peak, and assess the abnormality of the high-pressure peak based on the dynamic characteristics. S3 integrates multi-source cerebrovascular monitoring data with high-pressure peak abnormality assessment results, quantifies the risk probability of each peak event, identifies physiological fluctuations and risk events based on the risk probability, and provides risk warnings and responses for different events; S4. After risk warning, monitor multi-source cerebrovascular monitoring data in real time, quantify the warning effect, evaluate the accuracy of the warning, identify missed and false alarms based on the warning effect, and optimize the threshold. The specific process of quantifying the risk probability of each peak event by integrating multi-source cerebrovascular monitoring data and high-pressure peak anomaly assessment results is as follows: Real-time acquisition of multi-source cerebrovascular monitoring data and peak abnormality comprehensive values, and execution of cerebral hemorrhage risk assessment process: Based on the sliding time window, the mean and standard deviation of historical peak abnormality comprehensive values ​​are calculated, and the sum of the mean and twice the standard deviation of historical peak abnormality comprehensive values ​​is calculated as the peak abnormality quantile value; at the same time, historical heart rate and historical blood oxygen saturation are acquired based on the sliding time window, and the historical mean heart rate and historical mean blood oxygen saturation are calculated; For each peak event, the maximum heart rate value related to the peak is obtained by taking the maximum heart rate within a fixed sliding time window before and after the peak, and the blood oxygen saturation corresponding to the peak is obtained. The standardized value of the peak abnormality is obtained by dividing the current peak abnormality composite value by the peak abnormality percentile; the standardized value of the heart rate is obtained by dividing the peak-related maximum heart rate by the historical mean heart rate; the standardized blood oxygen value is obtained by dividing the blood oxygen saturation corresponding to the peak by the historical mean blood oxygen saturation; the blood oxygen drop activation term is obtained by subtracting the standardized blood oxygen value from the constant; if the result is less than 0, the blood oxygen drop activation term is assigned a value of 0, otherwise it is assigned the original calculation result; the blood oxygen drop activation term is added to the constant to obtain the blood oxygen modulation enhancement value. The risk index fusion value is obtained by multiplying the peak abnormality value, the heart rate fusion value, and the blood oxygen modulation enhancement value, and taking the negative value as the exponent for natural exponentiation; the risk index fusion value is obtained by subtracting the risk index fusion value from the constant.

2. The method for prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage according to claim 1, characterized in that, The specific process of real-time acquisition of multi-source cerebrovascular monitoring data, and the preprocessing of the multi-source cerebrovascular monitoring data including noise filtering, outlier correction, missing value completion, and standard normalization is as follows: Real-time acquisition of multi-source cerebrovascular monitoring data, including blood pressure, heart rate, and blood oxygen saturation; High-frequency noise and short-term spikes in the original multi-source cerebrovascular monitoring data were removed using moving average, median filtering, and bandpass filtering. Outliers in the multi-source cerebrovascular monitoring data were identified by physical threshold comparison and replaced with the mean of the preceding and following sliding windows. Missing data were completed using linear interpolation. All multi-source cerebrovascular monitoring data were standardized and dimensionless normalized. A cerebral hemorrhage risk monitoring database was established, and the original and preprocessed multi-source cerebrovascular monitoring data were stored in the cerebral hemorrhage risk monitoring database with timestamps.

3. The method for prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage according to claim 1, characterized in that, The specific process for identifying high-pressure peak events based on preprocessed multi-source cerebrovascular monitoring data is as follows: Statistical analysis of blood pressure data and identification of peak events: Detect whether the blood pressure at each point is greater than the blood pressure values ​​of the n points before and after it. If so, record it as a peak. Within the window before the peak, calculate the longest continuous sub-interval with a positive blood pressure slope as the rising segment. Identify the interval after the peak where the blood pressure falls back to the blood pressure baseline threshold as the recovery segment. Write each peak event and the corresponding multi-source cerebrovascular monitoring data into the cerebral hemorrhage risk monitoring database.

4. The method for prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage according to claim 1, characterized in that, The specific process of quantifying key dynamic characteristics of the high-pressure peak based on multi-source cerebrovascular monitoring data corresponding to the high-pressure peak event, and assessing the abnormality of the high-pressure peak based on the dynamic characteristics, is as follows: Within the rising segment of the peak event, the slope is calculated for each pair of adjacent blood pressure values, and the largest positive value is selected as the maximum rate value of the rising segment. Within the recovery segment of the peak event, the slope is calculated for each pair of adjacent blood pressure values, and the largest negative value is selected, with the absolute value taken as the maximum rate value of the recovery segment. Within the sliding time window before the peak, the peak point is removed, and the standard deviation of the remaining blood pressure is calculated to obtain the baseline fluctuation value. The timestamps of the recovery segment end and the rising segment start are obtained, and the time interval is calculated to obtain the peak duration. Multiplying the maximum rate value of the rising segment by the peak duration yields the peak rise intensity factor; adding the product of the baseline fluctuation value and the maximum rate value of the recovery segment to the minimum constant value yields the fluctuation recovery adjustment factor. The peak anomaly composite value is obtained by dividing the peak rise intensity factor by the fluctuation recovery adjustment factor. After each peak event, the peak abnormality composite value is calculated in real time and then incorporated into the cerebral hemorrhage risk assessment process.

5. The method for prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage according to claim 1, characterized in that, The specific process for identifying physiological fluctuations and risk events based on risk probability is as follows: Write the intracerebral hemorrhage risk probability value into the intracerebral hemorrhage risk monitoring database, and compare the intracerebral hemorrhage risk probability value with multi-level risk thresholds P1 and P2: When < P1, it is determined as transient physiological fluctuation; when P1 ≤ < P2, it is determined as an intracerebral hemorrhage early warning event; when ≥ P2, it is determined as an intracerebral hemorrhage dangerous event.

6. The method for prevention, intervention, and closed-loop risk feedback control of cerebral hemorrhage according to claim 1, characterized in that, The specific process for risk warning and response to different events is as follows: Risk warnings are issued for different situations. For transient physiological fluctuations, only internal records are made, and the probability value of cerebral hemorrhage risk is continuously monitored. For cerebral hemorrhage warning events, warning information is pushed to the doctor's terminal, and reminders for attention and follow-up examination are issued, increasing the monitoring frequency of multi-source cerebrovascular monitoring data. For dangerous cerebral hemorrhage events, multi-channel emergency warnings are issued.

7. The method for prevention, intervention, and closed-loop risk feedback control of cerebral hemorrhage according to claim 1, characterized in that, The specific process for real-time monitoring of multi-source cerebrovascular monitoring data after risk warning, quantifying the warning effect, and evaluating the accuracy of the warning is as follows: After a risk warning is issued, multi-source cerebrovascular monitoring data is continuously monitored, and blood pressure, heart rate, and blood oxygen saturation at the warning time are obtained. Based on a sliding time window after the risk warning, the difference between the maximum and minimum blood pressure values ​​is used to obtain the blood pressure variation range, the difference between the maximum and minimum heart rate values ​​is used to obtain the heart rate variation range, and the difference between the maximum and minimum blood oxygen saturation values ​​is used to obtain the blood oxygen saturation variation range. The baseline value of the warning signal is obtained by adding the absolute values ​​of blood pressure, heart rate, and blood oxygen saturation at the warning time. The absolute values ​​of blood pressure variation, heart rate variation, and blood oxygen saturation variation are added together to obtain the signal change value after the warning; the warning effect assessment value is obtained by dividing the warning signal baseline value by the signal change value after the warning.

8. The method for prevention, intervention, and risk feedback closed-loop control of cerebral hemorrhage according to claim 1, characterized in that, The specific process of identifying missed and false alarms based on the early warning effect and optimizing the threshold is as follows: After a risk warning is issued, the warning effect assessment value is calculated in real time and written into the cerebral hemorrhage risk monitoring database. If the warning effect assessment value is less than the effect threshold, the multi-level risk threshold is reduced and the sampling frequency of multi-source cerebrovascular monitoring data is increased. Multi-source cerebrovascular monitoring data before and after the risk warning are reviewed to identify whether there are any missed events. If the warning effect assessment value is greater than or equal to the effect threshold, the existing multi-level risk threshold and sampling frequency are maintained, and the absolute values ​​of blood pressure change, heart rate change, and blood oxygen saturation change are analyzed to identify false alarm events. All missed and false alarm events are statistically analyzed to form a source tracing report, which includes the time of the event, multi-source cerebrovascular monitoring data, and the identified cause. This report is then pushed to the doctor's terminal for manual review and feedback. At the same time, Bayesian optimization and reinforcement learning algorithms are used to optimize multi-level risk thresholds and effect thresholds.

9. A closed-loop control system for prevention, intervention, and risk feedback in cerebral hemorrhage, characterized in that, The system includes: The data acquisition and high-pressure peak extraction module is used to acquire multi-source cerebrovascular monitoring data in real time, perform noise filtering, outlier correction, missing value completion and standard normalization preprocessing on the multi-source cerebrovascular monitoring data, and identify high-pressure peak events based on the preprocessed multi-source cerebrovascular monitoring data. The high-pressure peak monitoring and quantification module is used to quantify the key dynamic characteristics of the peak based on the multi-source cerebrovascular monitoring data corresponding to the high-pressure peak event, and to assess the abnormality of the high-pressure peak based on the dynamic characteristics. The risk identification and classification module is used to acquire multi-source cerebrovascular monitoring data and peak abnormality composite values ​​in real time, and execute the cerebral hemorrhage risk assessment process: Based on a sliding time window, it calculates the mean and standard deviation of historical peak abnormality composite values, and calculates the sum of the mean and twice the standard deviation of historical peak abnormality composite values ​​as the peak abnormality quantile; simultaneously, it acquires historical heart rate and historical blood oxygen saturation based on a sliding time window, and calculates the historical mean heart rate and historical mean blood oxygen saturation; for each peak event, it takes the maximum heart rate within a fixed sliding time window before and after the peak to obtain the peak-related maximum heart rate, and obtains the blood oxygen saturation corresponding to the peak; it divides the current peak abnormality composite value by the peak abnormality quantile to obtain the peak abnormality standardized value; and it uses the peak-related heart rate... The maximum value is divided by the historical mean heart rate to obtain the standardized heart rate value; the blood oxygen saturation corresponding to the peak is divided by the historical mean blood oxygen saturation to obtain the standardized blood oxygen value; the blood oxygen drop activation term is obtained by subtracting the standardized blood oxygen value from the constant; if the result is less than 0, the blood oxygen drop activation term is assigned a value of 0, otherwise it is assigned the original calculation result; the blood oxygen drop activation term is added to the constant to obtain the blood oxygen modulation enhancement value; the peak abnormality standardized value, the heart rate standardized value and the blood oxygen modulation enhancement value are multiplied, and the negative value is used as the exponent for natural exponentiation to obtain the risk index fusion value; the risk index fusion value is subtracted from the constant to obtain the cerebral hemorrhage risk probability value; physiological fluctuations and risk events are identified based on the risk probability, and risk warnings and responses are provided for different events. The risk feedback and optimization module is used to monitor multi-source cerebrovascular monitoring data in real time after risk warning, quantify the warning effect, evaluate the accuracy of the warning, identify missed and false alarms based on the warning effect, and optimize the threshold.

Images