Intensive care unit patient early warning method based on clinical data

Abstract

The invention discloses an intensive care unit patient early warning method based on clinical data, and relates to the technical field of medical data processing. Comprising the following steps: collecting patient data in real time, and performing multi-modal data fusion; performing real-time analysis on the breathing waveform, extracting morphological characteristics in the waveform, and identifying subtle abnormal changes in the respiratory system; based on historical data and real-time data of the patient, constructing a time sequence model, and predicting whether the patient has a potential risk; and dynamically adjusting an early warning threshold value and a strategy based on the change trend of the historical data and the real-time data. According to the invention, the level of risk management of the patient in the intensive care unit is greatly improved, precise identification of subtle abnormity of the patient is ensured, and effective intervention can be carried out before a problem occurs.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, in particular to an intensive care unit patient early warning method based on clinical data. Background Art

[0002] In recent years, early warning systems for intensive care unit (ICU) patients in the medical field have developed rapidly. Thanks to the deep integration of medical and artificial intelligence technologies, the monitoring of ICU patients has gradually evolved from passive monitoring to active early warning. Currently, a large amount of clinical data can be collected in real time by monitoring equipment, including vital signs such as heart rate, blood pressure, and blood oxygen saturation, as well as more complex biosignals such as respiratory waveforms and electrocardiograms (ECGs). Using this data, modern medicine has achieved preliminary health status monitoring and disease trend analysis. However, despite continuous technological advancements, existing early warning systems still have limitations. In particular, in complex and volatile clinical environments, a single parameter early warning method cannot effectively respond to the dynamic changes in a patient's condition.

[0003] Most existing early warning systems rely on the monitoring of a single or a few parameters and are unable to fully utilize historical patient data and multimodal real-time data for comprehensive analysis. This limitation limits the sensitivity and specificity of existing early warning systems, making it difficult to capture subtle changes in a patient's condition, especially in the early stages of respiratory system abnormalities. The respiratory waveform, as an important physiological signal reflecting respiratory function, has morphological characteristics that can reflect many subtle respiratory abnormalities. However, traditional systems are generally unable to analyze complex respiratory waveform morphology in real time. In addition, fixed early warning thresholds and strategies cannot cope with changes in individual patient conditions, especially when a patient's physiological condition fluctuates dramatically in a short period of time. Traditional early warning mechanisms appear sluggish and cannot achieve dynamic adjustment and flexible response. Summary of the Invention

[0004] In view of the problems existing in the existing early warning system, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to timely detect potential risks of patients through morphological analysis of respiratory waveforms and dynamic threshold adjustment.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, an embodiment of the present invention provides an intensive care unit patient early warning method based on clinical data, which includes real-time collection of patient data and multimodal data fusion; real-time analysis of the patient's respiratory waveform, extraction of morphological features in the waveform, and identification of subtle abnormal changes in the respiratory system; construction of a time series model based on the patient's historical data and real-time data to predict whether the patient has potential risks; and dynamic adjustment of early warning thresholds and strategies based on the changing trends of historical data and real-time data.

[0007] As a preferred solution of the ICU patient early warning method based on clinical data of the present invention, the multimodal data fusion includes: extracting key features from the data of each modality; fusing under a unified framework based on the features of different modalities: there are n different physiological indicators, which are respectively , the weight corresponding to each indicator is , set heart rate and blood oxygen saturation The weights are and , then the calculation formula of the comprehensive index X after weighted fusion is: ;in, Adjust according to the importance of each physiological indicator.

[0008] As a preferred solution of the intensive care unit patient early warning method based on clinical data described in the present invention, the method includes: performing real-time analysis on the patient's respiratory waveform, extracting morphological features in the waveform, and identifying subtle abnormal changes in the respiratory system, including the following steps: real-time monitoring of the patient's respiratory waveform data, extracting basic features including peaks, troughs, amplitudes, and periods; extracting advanced features including amplitude standard deviation and respiratory rhythm change rate; identifying abnormal phenomena in the respiratory system by analyzing the amplitude standard deviation and the respiratory rhythm change rate; identifying whether there is a systemic respiratory deterioration trend by accumulating small changes in multiple respiratory waveforms; when a small change in the respiratory waveform is detected, the electrocardiogram, blood pressure, and blood oxygen data are simultaneously used as verification items to determine whether there is a synchronous abnormality; and comprehensively evaluating whether the respiratory waveform abnormality is valid based on the verification results.

[0009] As a preferred solution of the intensive care unit patient warning method based on clinical data of the present invention, wherein: the identification of abnormal phenomena in the respiratory system includes: the amplitude standard deviation The calculation formula is: ;in, is the amplitude, is the average value of the amplitude; The respiratory rhythm change rate for: ;in, The duration of each breathing cycle, Indicates the k The length of a cycle is the time recorded in the current cycle. Indicates the k +1 period of time; Identifying whether there is a systemic respiratory deterioration trend includes: Set the detection window length to , accumulate the changes of multiple cycles, and set a threshold according to the magnitude of the change. When the cumulative change exceeds the threshold, an early warning is triggered. The cumulative change formula is: ;in, Based on the detection window length The amplitude mean; when the cumulative change Exceeding the set threshold When the alarm is triggered.

[0010] As a preferred embodiment of the clinical data-based early warning method for intensive care unit patients described in the present invention, the method comprises the following steps: using electrocardiogram, blood pressure and blood oxygen data as verification items to determine whether there is a synchronization abnormality, including the following steps: extracting the RR interval in the electrocardiogram, calculating the instantaneous heart rate, and comparing it with the normal heart rate range. If the heart rate fluctuation exceeds the fluctuation threshold, it is considered abnormal; collecting systolic and diastolic blood pressure, and comparing them with the normal blood pressure range. If the systolic or diastolic blood pressure exceeds the normal range, it is determined to be abnormal; if the blood pressure is within the normal range, it is determined to be normal; collecting blood oxygen data to determine whether it is lower than the normal value. If it is lower than the blood oxygen concentration threshold, it is considered abnormal; if it is within the normal range, it is considered to be normal.

[0011] As a preferred solution of the intensive care unit patient warning method based on clinical data of the present invention, the comprehensive evaluation of whether the abnormal respiratory waveform is effective according to the verification results includes: summarizing the verification results of all physiological parameters to form an abnormal state vector ,in, is the change in heart rate, is the change in blood pressure, is the change in blood oxygen saturation; if all check items are 0, the priority of this abnormal breathing warning is reduced, it is considered to be a false alarm and no alarm is triggered; if at least one item is 1, that is If there is one or more abnormalities in the respiratory waveform, the respiratory abnormality is confirmed to be valid, an early warning signal is issued, and relevant data is recorded; under the false alarm reduction mechanism, if the abnormality detected by the respiratory waveform fails to pass the verification of other physiological parameters, the abnormality is recorded as a false alarm event, and the frequency of the same type of false alarms is reduced.

[0012] As a preferred embodiment of the clinical data-based early warning method for intensive care unit patients described in the present invention, the construction of a time series model to predict whether a patient has potential risks includes the following steps: constructing a time series model based on the patient's historical respiratory data and real-time collected data, while considering short-term, medium-term and long-term data changes; using a weighted algorithm to combine abnormal respiratory waveform characteristics with physiological parameters to generate a risk score for each patient's time series data, and comprehensively evaluate the patient's respiratory status; when it is judged that there is a high risk of respiratory failure in the future, the system will issue a high-level warning, prompting medical staff to intervene immediately.

[0013] As a preferred embodiment of the ICU patient early warning method based on clinical data of the present invention, the dynamic adjustment of the early warning threshold and strategy includes: using a machine learning algorithm to automatically and dynamically adjust the threshold of the physiological indicator. The adaptive threshold formula is as follows: ;in, For the i Dynamic adjustment threshold of each indicator, is the basic threshold, To learn the parameters, The change in the indicator at the current time point; different early warning strategies are implemented according to the development of the patient's condition.

[0014] In a second aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the intensive care unit patient early warning method based on clinical data as described in the first aspect of the present invention are implemented.

[0015] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the intensive care unit patient early warning method based on clinical data as described in the first aspect of the present invention are implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention can detect the trend of respiratory deterioration by detecting small changes in amplitude and rhythm, and judge whether there are potential problems in the respiratory system through multiple cumulative analyses. It effectively reduces the incidence of false alarms by combining electrocardiogram, blood pressure and blood oxygen data as verification items, and achieves the effect of accurate early warning. In addition, compared with the traditional single data analysis method, the present invention accurately predicts future risks through time series models, realizes the transition from passive monitoring to active early warning, improves the risk prediction ability of patients in the intensive care unit, and can adaptively adjust the early warning strategy according to the specific development of the patient's condition, avoids the false alarm problem caused by fixed thresholds, and ensures the flexibility of the early warning mechanism. The present invention greatly improves the level of risk management of patients in the intensive care unit, ensures the accurate identification of subtle abnormalities of patients, and can effectively intervene before problems occur. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 Flowchart of the early warning method for intensive care unit patients based on clinical data.

[0019] Figure 2 FIG. 1 is a schematic diagram of the structure of a terminal in the first embodiment of the present invention.

[0020] Figure 3 Schematic diagram of the structure of the computer device in the first embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0022] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0024] Example 1: Reference Figures 1 to 3 , which is the first embodiment of the present invention, provides an intensive care unit patient early warning method based on clinical data, such as Figure 1 Shown, including, S1: Collect the patient's physiological data in real time and perform multimodal data fusion.

[0025] Specifically, various sensors installed in the respiratory intensive care unit (including ventilators, electrocardiogram monitors, blood pressure monitors, oximeters, etc.) collect a variety of physiological data of patients in real time. The collected data includes but is not limited to respiratory rate, tidal volume, blood oxygen saturation, heart rate, and blood pressure.

[0026] It should be noted that in order to ensure the accuracy and stability of the data, the sensor layout should take into account the patient's bed position, the connection of the respiratory equipment, and the interference of the patient's movement. For example, it should ensure that the sensor cable does not affect the daily operations of medical staff and that the equipment can automatically recalibrate when the patient moves.

[0027] The collected physiological data of different types are integrated into multimodal data to provide multi-dimensional information support for subsequent early warning steps. The specific operations are as follows: First, since different devices have different sampling frequencies (for example, heart rate data may be collected once per second, while blood pressure data may be collected once per minute), the data needs to be aligned through timestamps to ensure synchronization of data from different modalities.

[0028] Before data fusion, key features are first extracted from the data of each modality. For example, the heart rate change features are extracted from the ECG monitor, and the change trends of respiratory rate, respiratory waveform and tidal volume are extracted from the ventilator. Based on the features of different modalities, these data are fused under a unified framework. There are n different physiological indicators, namely , the weight corresponding to each indicator is Since heart rate and blood oxygen saturation are more important indicators, they need to be given greater weights. and blood oxygen saturation The weights are and , then the calculation formula of the comprehensive index X after weighted fusion is: ; Among them, the weight The value can be adjusted according to the importance of each physiological indicator. By reasonably allocating weights, more important physiological indicators will occupy a larger proportion in the fusion results, ensuring that they have a greater impact on the results.

[0029] The fused data set consists of timestamps and various physiological data features, including respiratory rate, tidal volume, blood oxygen saturation, heart rate, and blood pressure. The data set is arranged in a time series, ensuring continuity and consistency. The fused data can be sliced ​​into specific time windows (e.g., every 5 minutes), facilitating subsequent early warning analysis.

[0030] S2: Analyze the respiratory waveform in real time, extract the morphological features in the waveform, and identify subtle abnormal changes in the respiratory system.

[0031] Preferably, the patient's respiratory waveform data, especially key parameters such as the tidal volume waveform and respiratory rate waveform, are monitored in real time, and the respiratory waveform is analyzed in real time using a signal processing algorithm. In addition to the traditional peak, amplitude, and period, the detection of subtle changes in the respiratory waveform is added, including the standard deviation of the waveform amplitude, the rate of change of the respiratory rhythm, etc. By comparing the slight changes in the respiratory waveform, the patient's respiratory deterioration trend can be discovered in advance.

[0032] Specifically, the basic features extracted include peaks, troughs, amplitudes, and periods; the advanced features extracted include amplitude standard deviation and respiratory rhythm change rate.

[0033] Among them, the standard deviation of the amplitude is the degree of fluctuation of the waveform amplitude, which indicates the amplitude change of each breath. For N cycles, the standard deviation of the amplitude is The calculation formula is: ;in, is the amplitude, is the average value of the amplitude.

[0034] The respiratory rhythm change rate is the rate of change of the respiratory cycle, which detects whether the breathing becomes faster or slower. for: ;in, The duration of each breathing cycle, Indicates the k The length of a cycle is the time recorded in the current cycle. Indicates the k +1 cycle length; this indicator is used to monitor fluctuations in respiratory rhythm, especially when detecting respiratory abnormalities, such as breathing becoming erratic or irregular.

[0035] By analyzing the standard deviation and rate of change in the waveform, abnormalities in the respiratory system can be identified. Common subtle abnormalities include: Decreased respiratory amplitude: The tidal volume of each breath is significantly reduced, which may mean that the breathing becomes shallow or the breathing capacity is limited. The downward trend of breathing depth can be detected early.

[0036] Unstable respiratory rhythm: by respiratory rhythm change rate , detect whether the breathing cycle becomes irregular or the fluctuation amplitude increases.

[0037] By accumulating the slight changes in multiple respiratory waveforms, we can identify whether there is a systematic trend of respiratory deterioration and set a detection window length. To accumulate changes over multiple cycles, and set a threshold based on the magnitude of the change, trigger an alert when the cumulative change exceeds the threshold. The cumulative change formula is: ;in, Based on the detection window length The amplitude mean; when the cumulative change Exceeding the set threshold When the alarm is triggered.

[0038] When a slight change in the respiratory waveform is detected (such as decreased respiratory depth or unstable respiratory rhythm), the ECG, blood pressure, and blood oxygen data are simultaneously used as verification items to determine whether there is synchronization abnormality. The verification process is as follows: Extract the RR interval in the electrocardiogram, calculate the instantaneous heart rate (HR), and compare it with the normal heart rate range. If the heart rate fluctuates greatly and exceeds the fluctuation threshold (such as tachycardia, bradycardia), it is considered abnormal; if the heart rate is stable within the normal range, it is considered normal; collect systolic blood pressure (SBP) and diastolic blood pressure (DBP) and compare them with the normal blood pressure range. If the systolic blood pressure or diastolic blood pressure exceeds the normal range, it is considered abnormal; if the blood pressure is within the normal range, it is considered normal; collect blood oxygen data and monitor Is it lower than the normal value (usually lower than 90% is abnormal), if If the blood oxygen concentration is lower than the threshold, it is considered abnormal; if If it is within the normal range, it is considered normal.

[0039] After performing ECG, blood pressure, and blood oxygen tests, comprehensively evaluate whether the respiratory waveform abnormality is valid based on the test results to reduce false alarms. The specific steps for verification and judgment are as follows: Summarize the verification results of all physiological parameters to form an abnormal state vector ,in, is the change in heart rate, is the change in blood pressure, The change in blood oxygen saturation. If all check items are 0, the priority of the abnormal breathing warning is reduced, and it is considered to be a false alarm and no alarm is triggered. If at least one item is 1, that is, If there is one or more abnormalities in the breathing, the abnormal breathing is confirmed to be valid, an early warning signal is issued, and relevant data is recorded.

[0040] Under the false alarm reduction mechanism, if the abnormality detected in the respiratory waveform fails to pass the verification of other physiological parameters, the abnormality will be recorded as a "false alarm" event, and the false alarm weight will be updated to reduce the frequency of the same type of false alarms.

[0041] S3: Based on the patient's historical respiratory data and real-time collected data, a time series model is constructed to predict whether the patient has potential risks.

[0042] S3.1: Based on the patient's historical respiratory data and real-time data, a time series model is constructed that simultaneously considers short-term (seconds), medium-term (minutes), and long-term (hours) data changes. Through multi-scale fusion, the instantaneous changes and long-term trends of the patient's condition can be more comprehensively captured.

[0043] Time series analysis usually uses ARIMA models or LSTM models. Here, based on multimodal data and different time scales, we use a generalized multivariate autoregressive moving average model (VARMA) or a long short-term memory network (LSTM) model to predict future respiratory failure risk.

[0044] Data at each time scale are integrated through a multi-scale fusion algorithm. The fused data can more comprehensively capture the short-term changes and long-term trends of the patient's respiratory status.

[0045] Among them, the time series prediction formula is: ;in, is the predicted future value, is the past time series data, f is a time series model function (such as an LSTM model), is the number of valid historical data points used for the time series forecasting model.

[0046] S3.2: In time series analysis, abnormal respiratory waveform features (such as amplitude standard deviation and respiratory rhythm change rate) are integrated with physiological parameters such as heart rate, blood pressure, and blood oxygen saturation to generate a risk score for each patient's time series data. A weighted algorithm is used to weight the frequency of respiratory waveform abnormalities, the amplitude of heart rate fluctuations, and the changing trends of blood pressure and blood oxygen to comprehensively assess the patient's respiratory status. When the risk score exceeds the preset risk value, it is determined that there is a high risk of respiratory failure in the future.

[0047] S3.3: When it is judged that there is a high risk of respiratory failure in the future, the system will issue a high-level warning, prompting medical staff to intervene immediately.

[0048] When a high risk score is obtained through time series analysis and multimodal data integration, a high-level warning will be automatically issued. The warning content includes: Specific abnormalities in the respiratory waveform (such as a gradual decrease in respiratory depth, significant instability in respiratory rhythm, etc.); Abnormal synchronization of heart rate, blood pressure, and blood oxygen (such as tachycardia, sudden drop in blood pressure, or persistent decrease in blood oxygen saturation); Predicted time frame for respiratory failure (e.g., there is a 90% probability of respiratory failure occurring within the next 5 minutes).

[0049] Furthermore, not only will an alarm be issued, but corresponding intervention suggestions will also be provided to prompt medical staff to take necessary measures. The intervention suggestions include: Immediately increase oxygen supply and adjust ventilator parameters; For patients with abnormal heart rhythm, use antiarrhythmic drugs or perform electric cardioversion; For patients with abnormal blood pressure, inject pressor drugs or take emergency infusion measures.

[0050] S4: Dynamically adjust warning thresholds and strategies based on changing trends of historical data and real-time data.

[0051] In actual clinical scenarios, a patient's condition may worsen suddenly or progress slowly, requiring dynamic adjustment of the alert level based on real-time data. For example, if a patient's breathing becomes rapid or their blood oxygen level continues to drop, the system will automatically upgrade the low-level alert to a high-level alert and notify the doctor to take immediate intervention measures.

[0052] Specifically, machine learning algorithms (such as Bayesian optimization or reinforcement learning) are used to automatically and dynamically adjust the thresholds of physiological indicators such as respiratory waveforms, heart rate, and blood pressure. For example, the system will continuously learn based on real-time fluctuation data and automatically narrow or expand the threshold range to adapt to different pathological conditions. The adaptive threshold formula is as follows: ;in, For the i Dynamic adjustment threshold of each indicator, is the basic threshold, To learn the parameters, The change in the indicator at the current time point.

[0053] Different early warning strategies are implemented according to the progression of the patient's condition. For example, in the case of mild abnormalities, a low-level automatic warning reminder can be issued first, prompting nurses to increase the frequency of rounds. In the case of high risk of respiratory failure, a high-level warning is directly triggered, calling a doctor for immediate intervention.

[0054] This embodiment also provides a computer device, which is suitable for the intensive care unit patient early warning method based on clinical data, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the intensive care unit patient early warning method based on clinical data proposed in the above embodiment.

[0055] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0056] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the method for realizing an early warning method for intensive care unit patients based on clinical data as proposed in the above embodiment is implemented.

[0057] In summary, the beneficial effects of the present invention are as follows: the present invention can detect the trend of respiratory deterioration by detecting slight changes in amplitude and rhythm, and judge whether there are potential problems in the respiratory system through multiple cumulative analyses. It effectively reduces the incidence of false alarms by combining electrocardiogram, blood pressure and blood oxygen data as verification items, and achieves the effect of accurate early warning. In addition, compared with the traditional single data analysis method, the present invention accurately predicts future risks through a time series model, realizes the transition from passive monitoring to active early warning, improves the risk prediction ability of patients in the intensive care unit, and can adaptively adjust the early warning strategy according to the specific development of the patient's condition, avoids the false alarm problem caused by fixed thresholds, and ensures the flexibility of the early warning mechanism. The present invention greatly improves the level of risk management of patients in the intensive care unit, ensures the accurate identification of subtle abnormalities of patients, and can effectively intervene before problems occur.

[0058] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An early warning method for patients in an intensive care unit based on clinical data, characterized in that: include, Collect patient data in real time and perform multimodal data fusion; Analyze the patient's respiratory waveform in real time, extract the morphological features in the waveform, and identify subtle abnormal changes in the respiratory system; Based on the patient's historical data and real-time data, a time series model is constructed to predict whether the patient has potential risks; Dynamically adjust warning thresholds and strategies based on changing trends of historical data and real-time data.

2. The intensive care unit patient early warning method based on clinical data as claimed in claim 1, characterized in that: The multimodal data fusion includes: Extract key features from the data of each modality; Based on the features of different modalities, they are integrated under a unified framework: Suppose there are n different physiological indicators, namely , the weight corresponding to each indicator is , set heart rate and blood oxygen saturation The weights are and , then the calculation formula of the comprehensive index X after weighted fusion is: ;in, Adjust according to the importance of each physiological indicator.

3. The intensive care unit patient early warning method based on clinical data as claimed in claim 2, characterized in that: Real-time analysis of the patient's respiratory waveform, extraction of morphological features in the waveform, and identification of subtle abnormal changes in the respiratory system include the following steps: Monitor the patient's respiratory waveform data in real time, extract basic features including peaks, troughs, amplitudes, and periods; extract advanced features including amplitude standard deviation and respiratory rhythm change rate; Identifying abnormalities in the respiratory system by analyzing the amplitude standard deviation and the respiratory rhythm change rate; By accumulating and detecting small changes in multiple respiratory waveforms, it can identify whether there is a systematic trend of respiratory deterioration; When a slight change in the respiratory waveform is detected, the ECG, blood pressure and blood oxygen data are used as verification items to determine whether there is a synchronization abnormality; Comprehensively evaluate whether the abnormal respiratory waveform is valid based on the verification results.

4. The intensive care unit patient early warning method based on clinical data as claimed in claim 3, characterized in that: The identifying of abnormalities in the respiratory system includes: For N cycles, the amplitude standard deviation The calculation formula is: ;in, is the amplitude, is the average value of the amplitude; The respiratory rhythm change rate for: ;in, The duration of each breathing cycle is Indicates k The length of a cycle is the time recorded in the current cycle. Indicates k +1 period of time; Identifying whether there is a systemic respiratory deterioration trend includes: Set the detection window length to , accumulate changes in multiple cycles, and set a threshold based on the magnitude of the change. When the cumulative change exceeds the threshold, an early warning is triggered. The cumulative change formula is: ;in, Based on the detection window length The amplitude mean; when the cumulative change Exceeding the set threshold When the alarm is triggered.

5. The intensive care unit patient early warning method based on clinical data as claimed in claim 4, characterized in that: The electrocardiogram, blood pressure and blood oxygen data are used as verification items to determine whether there is synchronization abnormality The steps include: Extract the RR interval in the electrocardiogram, calculate the instantaneous heart rate, and compare it with the normal heart rate range. If the heart rate fluctuation exceeds the fluctuation threshold, it is considered abnormal; The systolic and diastolic blood pressures are collected and compared with the normal blood pressure range. If the systolic or diastolic blood pressure is outside the normal range, it is considered abnormal. If the blood pressure is within the normal range, it is considered normal; Collect blood oxygen data to determine whether it is lower than the normal value. If it is lower than the blood oxygen concentration threshold, it is considered abnormal; if it is within the normal range, it is considered normal.

6. The intensive care unit patient early warning method based on clinical data according to claim 5, characterized in that: The comprehensive evaluation of whether the abnormal respiratory waveform is effective according to the verification result includes: Summarize the verification results of all physiological parameters to form an abnormal state vector ,in, is the change in heart rate, is the change in blood pressure, is the change in blood oxygen saturation; If all check items are 0, the priority of this abnormal breathing warning is lowered, and it is considered to be a false alarm, and the alarm is not triggered; If at least one item is 1, If one or more of the above are abnormal, the abnormal breathing is confirmed to be effective, an early warning signal is issued, and relevant data is recorded; Under the false alarm reduction mechanism, if the abnormality detected by the respiratory waveform fails to pass the verification of other physiological parameters, the abnormality will be recorded as a false alarm event, while reducing the frequency of the same type of false alarms.

7. The intensive care unit patient early warning method based on clinical data according to claim 6, characterized in that: The construction of the time series model to predict whether a patient has potential risk includes the following steps: Based on the patient's historical respiratory data and real-time collected data, a time series model is constructed, taking into account short-term, medium-term and long-term data changes; Abnormal respiratory waveform characteristics and physiological parameters are combined through a weighted algorithm to generate a risk score for each patient's time series data, and comprehensively evaluate the patient's respiratory status; When it is judged that there is a high risk of respiratory failure in the future, the system will issue a high-level warning, prompting medical staff to intervene immediately.

8. The intensive care unit patient early warning method based on clinical data according to claim 7, characterized in that: The dynamic adjustment of warning thresholds and strategies includes: The thresholds of physiological indicators are automatically and dynamically adjusted using machine learning algorithms. The adaptive threshold formula is as follows: ;in, For the i Dynamic adjustment thresholds for indicators, is the basic threshold, To learn the parameters, is the change of the indicator at the current time point; Different early warning strategies are implemented according to the development of the patient's condition.

Images