High-dependence ward patient vital sign abnormity early warning nursing intervention method and system

Abstract

The invention relates to the technical field of vital sign monitoring, in particular to a high-dependence ward patient vital sign abnormity early warning nursing intervention method and system.The method comprises the following steps that heart rate systolic pressure and diastolic pressure blood oxygen data are collected, abnormity or recovery is marked according to intervals, abnormity and recovery time spans are accumulated to form a time structure, and the time structure is calculated; the method comprises the following steps: acquiring heart rate, blood pressure and blood oxygen data, calculating an abnormal occupancy ratio, analyzing cross-cycle progressive increase, extracting a sequence of first abnormal time points, counting continuous anomalies and successive evolution, and outputting early warning intervention information when continuous progressive increase is met and unsynchronized accumulation and sequential counting rules are recovered. The method comprises the following steps: forming a nursing period state sequence, accumulating abnormity and recovering a time span to construct a time structure, combining an abnormity occupation ratio and a preorder period change, extracting a first abnormity time point, sorting and identifying continuous abnormity and evolution, and outputting early warning according to a continuous increasing and unsynchronized accumulation and sequence counting recovering rule. Invalid alarm is reduced; and intervention consistency is improved.

Description

Technical Field

[0001] This invention relates to the field of vital sign monitoring technology, and in particular to a method and system for early warning and nursing intervention for patients in high-dependency wards with abnormal vital signs. Background Technology

[0002] The field of vital sign monitoring technology refers to a collection of technologies related to the continuous or periodic collection, recording, analysis, and interpretation of basic physiological parameters such as heart rate, respiratory rate, blood pressure, blood oxygen saturation, and body temperature of patients. It typically involves physiological signal acquisition devices, data acquisition channels, parameter threshold settings, time series recording, abnormal judgment rules, and information presentation methods that are integrated with clinical nursing processes. Overall, it is used to support medical personnel in understanding and assessing changes in patients' physiological status, and is especially suitable for daily monitoring and management scenarios of intensive care or high-risk patients.

[0003] Among them, the traditional nursing intervention method and system for early warning of abnormal vital signs in patients in high-dependence wards refers to the collection of vital sign data such as heart rate, blood pressure, respiration and blood oxygen by bedside monitoring equipment in high-dependence wards. The collected values ​​are compared with the pre-set normal range or warning threshold. When one or more indicators exceed the set range, an alarm message is generated. Nursing staff then manually check the monitoring screen to record the abnormal time point according to the alarm prompt, manually fill in the nursing record sheet, and implement the corresponding intervention measures according to the established nursing procedure.

[0004] Traditional methods primarily rely on triggering alarms based on single threshold exceedances. However, they lack a unified characterization of the correlation between indicators and the evolution of states. Abnormalities and recovery only present instantaneous location information, making it difficult to reflect the persistence and recurrence of abnormalities within the same nursing cycle. Alarms are triggered frequently and lack trend evidence. Nursing staff need to repeatedly check the screen and manually record time points, increasing their workload and affecting the consistency of records. Intervention judgments rely on individual experience, which can easily lead to delayed responses or over-intervention, affecting the accuracy of nursing resource allocation and risk management. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a method for early warning and nursing intervention for patients in highly dependent wards with abnormal vital signs, comprising the following steps: S1: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compare them with the preset normal nursing range, mark abnormal or recovery status, and construct vital sign status; S2: Based on the vital signs status, analyze the time span of abnormal and recovery states of all data in the nursing monitoring cycle, accumulate the time spans of abnormal and recovery states, and generate an abnormal time structure. S3: Based on the abnormal time structure, calculate the ratio of the sum of abnormal state time accumulation to the sum of abnormal and recovery state time accumulation, analyze the increasing relationship between the current and previous nursing cycle abnormal time occupancy ratios, and generate abnormal accumulation results; S4: Based on the vital signs status, extract the first abnormal sampling time point of all monitoring data in the current nursing monitoring cycle and sort them. Combine the abnormal accumulation results to determine and count the continuous abnormal state of the preceding vital signs when the subsequent abnormality occurs, and generate an abnormal evolution relationship. S5: Based on the abnormal evolution relationship, when the abnormal time occupancy ratio increases and the recovery state time is not accumulated synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule, abnormal early warning nursing intervention information is generated.

[0006] As a further aspect of the present invention, the vital signs status includes heart rate abnormality markers, systolic blood pressure abnormality markers, diastolic blood pressure abnormality markers, and blood oxygen saturation abnormality markers; the abnormal time structure includes abnormal state time span segments, recovery state time span segments, abnormal state cumulative duration, and recovery state cumulative duration; the abnormal accumulation result includes abnormal state time cumulative value, total nursing monitoring cycle duration, and abnormal time occupancy ratio; the abnormal evolution relationship includes vital sign abnormality sequence index, preceding abnormality persistence markers, and cross-index abnormality count values; and the abnormal early warning nursing intervention information includes early warning trigger markers, corresponding vital sign combination markers, and nursing intervention level markers.

[0007] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation monitoring data of patients in high-dependency wards during the nursing cycle, perform timestamp alignment and unit unification processing on each monitoring data, and arrange and aggregate fields in sequence according to the nursing sampling cycle to generate a vital signs sampling sequence; S102: Based on the vital signs sampling sequence, according to the normal ranges of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation in the nursing management guidelines, perform interval boundary comparisons on each monitoring data in the sequence, map the comparison results to abnormal, normal, or recovery markers, and combine them to generate a vital signs interval determination sequence. S103: Based on the vital sign interval determination sequence, perform state bit integration and logical merging on the labeling results of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation for each nursing sampling cycle, encode normal, abnormal, and recovery states into unified cycle state items and arrange them in chronological order to establish vital sign status.

[0008] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the vital signs status, the normal, abnormal and recovery status markers corresponding to heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation within the same nursing monitoring cycle are aligned on the time axis, and the start and end sampling points of multiple markers are mapped to continuous time intervals to generate a set of status time intervals. S202: Call the set of state time intervals and the nursing sampling cycle duration, calculate the interval time difference for the abnormal state interval and the recovery state interval respectively, and perform a step-by-step accumulation operation on the time difference for the same state to obtain the state cumulative duration vector; S203: For the cumulative duration vector of the state, the cumulative duration of the abnormal state and the cumulative duration of the recovery state are structurally combined according to a unified time dimension. The combination result is identified by the state type and the duration field is fixed. The abnormal state duration is filtered in combination with the preset abnormal duration threshold to generate an abnormal time structure.

[0009] As a further aspect of the present invention, the abnormal duration threshold is determined by statistically analyzing the cumulative duration sample set of abnormal states obtained within the same nursing monitoring cycle, collecting the cumulative duration values ​​of abnormal states corresponding to multiple nursing sampling cycles, constructing a duration sequence and sorting it, and calculating the median value of the cumulative duration sequence of abnormal states.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the abnormal time structure, for the same vital sign within the current nursing monitoring cycle, retrieve the cumulative value of abnormal state time, obtain the total duration of the nursing monitoring cycle, perform a ratio calculation between the cumulative value of abnormal state time and the total duration of the nursing monitoring cycle, and generate the abnormal time occupancy ratio. S302: Based on the abnormal time occupancy ratio of the current nursing monitoring cycle, the abnormal time occupancy ratio of the previous nursing monitoring cycle is collected at the same time. The ratio values ​​of the two cycles are arranged into a ratio sequence according to the time order. An incremental relationship judgment is performed on adjacent ratio values ​​to generate a ratio increment judgment sequence. S303: Based on the increasing ratio judgment sequence, the proportional items marked as increasing within the continuous nursing monitoring period are sequence aggregated, the number of increases is counted, and the abnormal time occupancy ratio, the cumulative value of abnormal state time and the number of increases are associated and combined to generate an abnormal accumulation result.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the vital signs status, collect the corresponding sampling sequences of heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation in the current nursing monitoring cycle, perform state switching detection for multiple sampling sequences, mark the sampling time point corresponding to the first transition from normal state to abnormal state, and generate an abnormal initial time point sequence. S402: Based on the abnormal initial time point sequence, sort and analyze adjacent abnormal event pairs of multiple vital signs abnormal time points, determine whether the status identifier of the preceding vital sign is abnormal in the sampling interval corresponding to the abnormal time point of the subsequent vital sign, and write it into the sequence relationship field to generate an abnormal sequence relationship set. S403: For the set of abnormal sequence relationships, perform a counting operation on the relationship items that satisfy the condition of the continuation of the preceding abnormal state, filter the relationship items whose count value exceeds the abnormal sequence counting threshold, and index and associate the filtered count value with the corresponding vital sign combination identifier to generate abnormal evolution relationship.

[0012] As a further aspect of the present invention, the preset abnormal sequence counting threshold is determined by collecting the abnormal sequence relationship count values ​​of all vital signs within the current nursing monitoring cycle and calculating the median statistic of the distribution of abnormal sequence relationship count values.

[0013] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the abnormal evolution relationship, obtain the abnormal time occupancy ratio sequence of the abnormal accumulation result, perform an increasing state judgment on the ratio values ​​of adjacent nursing cycles, record the continuous increasing identifier and perform time series mapping to generate an increasing abnormal time occupancy ratio sequence. S502: Based on the increasing sequence of abnormal time occupancy ratio, collect the length of the recovery period of multiple vital signs, calculate the cumulative sequence of recovery period time, and perform a synchronization judgment with the increasing sequence of abnormal time occupancy ratio on the same time axis to generate a set of recovery status identifiers; S503: For the set of recovery status identifiers, call the cross-index abnormal evolution count results in the abnormal evolution relationship, perform logical judgment with the count conditions limited by the preset nursing early warning rule, perform structured encapsulation and index association on the simultaneously established relationship items, and generate abnormal early warning nursing intervention information.

[0014] A nursing intervention system for early warning of abnormal vital signs in patients in highly dependent wards includes: The vital signs analysis module collects heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compares them with the preset normal nursing range, marks abnormal or recovery status, constructs vital signs status and transmits it to the abnormality statistics module. The anomaly statistics module, based on the vital signs status, analyzes the time span of all data in the abnormal and recovery states during the nursing monitoring cycle, accumulates the time spans of abnormal and recovery states, generates an abnormal time structure, and transmits it to the anomaly analysis module. The anomaly analysis module calculates the ratio of the sum of the abnormal state time to the sum of the abnormal and recovery state time based on the abnormal time structure, analyzes the increasing relationship between the current and previous nursing cycle abnormal time occupancy ratios, generates anomaly accumulation results, and transmits them to the evolution judgment module. The evolution judgment module extracts and sorts the first abnormal sampling time points of all monitoring data in the current nursing monitoring cycle based on the vital signs status, and judges and counts the continuous abnormal states of the preceding vital signs when they appear in the subsequent cycle, based on the abnormal accumulation results, and generates an abnormal evolution relationship and transmits it to the nursing early warning module. The nursing early warning module, based on the aforementioned abnormal evolution relationship, generates abnormal early warning nursing intervention information when the abnormal time occupancy ratio increases and the recovery state time does not accumulate synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, abnormal and recovery status markers are created by monitoring heart rate, blood pressure, and blood oxygen data to form a traceable status sequence within the nursing cycle. The time spans of abnormality and recovery are accumulated to construct a time structure. The relationship between the abnormality occupancy ratio and the changes in the preceding cycle is included in the analysis. At the same time, the time point of the first abnormality is extracted and sorted to identify continuous abnormalities and their sequential evolution. Combined with the rules of continuous increase and asynchronous accumulation of recovery and sequential counting, early warning information is output, so that the early warning shifts from single-point overshoot to trend and evolution judgment, reducing invalid alarms and improving the consistency of intervention timing. Attached Figure Description

[0016] 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.

[0017] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] 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.

[0020] Please see Figure 1 This invention provides a nursing intervention method for early warning of abnormal vital signs in patients in highly dependent wards, comprising the following steps: S1: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compare them with the preset normal nursing range, mark abnormal or recovery status, and construct vital sign status; S2: Based on vital signs, analyze the time span of abnormal and recovery states of all data in the nursing monitoring cycle, accumulate the time spans of abnormal and recovery states, and generate an abnormal time structure. S3: Based on the abnormal time structure, calculate the ratio of the sum of abnormal state time accumulation to the sum of abnormal and recovery state time accumulation, analyze the increasing relationship between the current and previous nursing cycle abnormal time occupancy ratio, and generate abnormal accumulation results; S4: Extract the first abnormal sampling time point of all monitoring data in the current nursing monitoring cycle based on the vital signs status and sort them. Combine the abnormal accumulation results to determine the continuous abnormal state of the preceding vital signs when the subsequent abnormality occurs and count it to generate the abnormal evolution relationship. S5: Based on the abnormal evolution relationship, when the abnormal time occupancy ratio increases and the recovery state time is not accumulated synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule, abnormal early warning nursing intervention information is generated.

[0021] Vital signs status includes heart rate abnormality markers, systolic blood pressure abnormality markers, diastolic blood pressure abnormality markers, and blood oxygen saturation abnormality markers. Abnormal time structure includes abnormal state time span, recovery state time span, cumulative abnormal state duration, and cumulative recovery state duration. Abnormal accumulation results include cumulative abnormal state time value, total nursing monitoring cycle duration, and abnormal time occupancy ratio. Abnormal evolution relationship includes vital sign abnormality sequence index, preceding abnormality persistence marker, and cross-index abnormality count value. Abnormal early warning nursing intervention information includes early warning trigger marker, corresponding vital sign combination marker, and nursing intervention level marker.

[0022] Please see Figure 2 The specific steps of S1 are as follows: S101: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation monitoring data of patients in high-dependency wards during the nursing cycle, perform timestamp alignment and unit unification processing on each monitoring data, and arrange and aggregate fields in sequence according to the nursing sampling cycle to generate a vital signs sampling sequence; A multi-parameter monitoring sensor array deployed at the bedside in high-dependency wards captures patients' physiological signals in real time within a preset nursing cycle. Specifically, photoplethysmography (PPG) sensors collect heart rate and blood oxygen saturation data, while oscillometric cuff pressure sensors collect systolic and diastolic blood pressure data. During the data acquisition initiation phase, a unified reference time axis is established, with the time dimension set to 1 second. For the raw data streams uploaded from different sensors, timestamp-based alignment is performed. When data gaps are detected due to transmission delays or inconsistent sampling frequencies, linear interpolation logic is used to fill them. This interpolation process first obtains the valid measured values ​​before and after the missing time point, calculates the difference between the two values, and divides this difference by the time interval between the two moments to obtain the change per unit time. Then, this change per unit time is multiplied by the time difference between the missing time point and the previous moment, and the result is summed with the valid measured value of the previous moment to derive the interpolated value for the missing point. For example, if the heart rate at 10:00:01 is 80 beats per minute and the heart rate at 10:00:03 is 82 beats per minute, with a time interval of 2 seconds, the change per unit time is first calculated to be 1 beat per minute. Then, the previous value of 80 is added to the change of 1, resulting in an interpolated heart rate of 81 beats per minute at 10:00:02. After time alignment, dimensional normalization is performed on each monitoring data point to eliminate the influence of differences in units and orders of magnitude of different physiological parameters on subsequent processing. This normalization process uses range standardization logic. First, all values ​​of the monitoring data within the current nursing sampling period are traversed to identify the maximum and minimum sampled values, and the minimum sampled value is subtracted from the maximum sampled value to determine the data fluctuation range. Then, for each current sampled value in the sequence, the difference between it and the minimum sampled value is calculated, and this difference is divided by the aforementioned data fluctuation range to obtain the normalized value corresponding to that sampling point. For example, within a certain sampling period, the maximum sampled value of systolic blood pressure is 180 mmHg, and the minimum sampled value is 80 mmHg. The difference between these two values, representing the data fluctuation range, is 100 mmHg. If the systolic blood pressure at the current sampling point is 130 mmHg, the difference between it and the minimum value is 50 mmHg. Dividing 50 by 100 yields a normalized value of 0.5 for that point. After the above processing, the normalized data of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation at the same timestamp, along with the original data, are aggregated chronologically to construct a structured vital sign sampling sequence.

[0023] S102: Based on the vital signs sampling sequence, according to the normal ranges of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation in the nursing management guidelines, perform interval boundary comparisons on each monitoring data in the sequence, map the comparison results to abnormal, normal, or recovery markers, and combine them to generate a vital signs interval determination sequence. Based on the generated vital sign sampling sequence, a pre-set nursing management standard database is invoked to extract normal ranges for physiological parameters for patients in high-dependency wards. The normal ranges are set as follows: heart rate 60-100 bpm, systolic blood pressure 90-140 mmHg, diastolic blood pressure 60-90 mmHg, and oxygen saturation 95-100%. Each record in the sequence is analyzed, and interval boundary comparisons are performed for each raw monitoring data point. If a monitoring data point value falls within the closed set of the normal range, the comparison result is mapped as a recovery marker, encoded with the number 0; if the value exceeds the normal range (i.e., above the upper threshold or below the lower threshold), it is mapped as an abnormal marker, encoded with the number 1. To eliminate misjudgments caused by momentary sensor jitter, a sliding window de-jitter logic is introduced. The logic sets a sliding window with a length of 5 sampling points. Only when the original comparison results of 5 consecutive sampling points within the window are all marked as abnormal will the exact state of that time period be locked as abnormal; otherwise, it is regarded as transient interference and the recovery mark state is maintained. Table 1 shows an example of the determination of the original sampling data before jitter removal processing.

[0024] Table 1. Example of vital sign sampling and assessment for patients in high-dependency wards. Time stamp Heart rate (beats / min) Systolic pressure (mmHg) Diastolic pressure (mmHg) Oxygen saturation (%) Raw decision result code T+1 85 120 80 98 0-0-0-0 T+2 110 145 95 94 1-1-1-1 T+3 90 130 85 96 0-0-0-0 T+4 55 110 70 97 1-0-0-0 As shown in Table 1, at time T+2, the heart rate was 110 beats per minute, higher than the upper limit of 100 beats per minute; the systolic blood pressure was 145 mmHg, higher than the upper limit of 140 mmHg; the diastolic blood pressure was 95 mmHg, higher than the upper limit of 90 mmHg; and the blood oxygen saturation was 94%, lower than the lower limit of 95%. All four indicators were outside the normal range, therefore the original judgment result at this time was coded as 1-1-1-1. At time T+4, only the heart rate was 55 beats per minute, lower than the lower limit of 60 beats per minute, and was judged as abnormal (code 1). The other three indicators were within the normal range (code 0), so the code for this time was 1-0-0-0. By performing the above comparison and de-jitter confirmation on the full-cycle data, the final combination generated the vital sign interval judgment sequence.

[0025] S103: Based on the vital signs interval determination sequence, perform state bit integration and logical merging on the labeling results of heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation for each nursing sampling cycle, encode normal, abnormal and recovery states into unified cycle state items and arrange them in chronological order to establish vital signs status; Based on the vital sign interval determination sequence, state bit integration and logical merging are performed on the time point status within each nursing sampling cycle. First, a 4-bit binary state register is established, with the bit order corresponding to heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation, respectively. To compress the multidimensional state into a single feature indicator, weighted state coding calculation is performed. According to clinical emergency priority, weight coefficients are set for each dimension: blood oxygen saturation weight is 8, systolic blood pressure weight is 4, diastolic blood pressure weight is 2, and heart rate weight is 1. During calculation, the product of the label value of each dimension and the corresponding weight coefficient is obtained, and these four products are summed to obtain the state feature value at that time point. Taking the determination code 1-0-0-0 at time T+4 in Table 1 as an example, it corresponds to abnormal heart rate (label value 1), and the rest are normal (label value 0). Multiplying the heart rate label value 1 by the weight 1 yields 1, and the products of the other three terms are all 0. Summing these results in a state feature value of 1. Taking a complex mixed abnormal situation as an example, if the determination code is 1-0-0-1 (i.e., heart rate and blood oxygen are simultaneously abnormal), the calculation process is as follows: the heart rate label value is 1*1=1, the blood oxygen label value is 1*8=8, and the other two are 0. Finally, 1 and 8 are added together to obtain the state characteristic value of 9 at this time point. This weighted summation mechanism ensures that each decimal characteristic value can be uniquely decoded to reveal the specific abnormal combination pattern. After completing the single-point encoding, the cycle state logic is merged. The cumulative duration of each non-zero state characteristic value within the nursing cycle is counted. The risk time threshold is set to 300 seconds. If the cumulative duration of a certain non-zero state characteristic value (such as the aforementioned characteristic value 9) exceeds 300 seconds within the cycle, the cycle is determined to be in a high-risk state of "combined abnormal blood oxygen and heart rate"; if the duration of all non-zero characteristic values ​​does not exceed the threshold, or the characteristic values ​​for the entire cycle are all 0, it is marked as "recovery state". Finally, the determined cycle state items are arranged in chronological order to establish a complete vital sign state sequence.

[0026] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on vital signs, the normal, abnormal and recovery status markers of heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation within the same nursing monitoring cycle are aligned on the time axis, and the start and end sampling points of multiple markers are mapped to continuous time intervals to generate a set of status time intervals. Based on the generated vital sign data, time axis alignment is performed on four physiological parameters—heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation—within the same nursing monitoring cycle. First, four independent time axis pointers are initialized, each pointing to the beginning of its respective parameter sequence, and simultaneously traversing every sampling point throughout the entire nursing cycle. During this traversal, the transitions of status markers are monitored in real time to identify consecutive abnormal and recovery segments. For any physiological parameter, such as heart rate, when the status marker changes from 0 (representing a recovery state) to 1 (representing an abnormal state), the timestamp of that sampling point is immediately recorded as the starting sampling point of the current abnormal interval. Subsequently, as the time axis pointer continues to move forward, monitoring of the status markers continues until the status marker changes from 1 back to 0, or the end of the nursing monitoring cycle is reached. At this point, the timestamp of the previous sampling point is recorded as the ending sampling point of the current abnormal interval. Through this beginning-end locking method, a discrete abnormal signal composed of multiple consecutive sampling points is mapped into a continuous closed time interval. Similarly, the same logic is applied to the processing of recovery states, identifying the start and end positions of continuous recovery intervals between two abnormal intervals. The above scanning and mapping operations are performed in parallel on the four physiological parameters, arranging all identified start and end sampling points chronologically to construct a set of state time intervals containing multiple binary data sets. For example, in a nursing cycle of 3600 seconds, if heart rate monitoring data shows abnormalities from 150 to 210 seconds and again from 800 to 840 seconds, then two specific abnormal time intervals are generated for the heart rate parameter, recorded as interval A (150 seconds, 210 seconds) and interval B (800 seconds, 840 seconds), respectively. This not only achieves vectorized aggregation of states at single time points but also eliminates redundant data on the time axis, providing a standardized interval boundary basis for subsequent duration calculations.

[0027] S202: Call the state time interval set and the nursing sampling cycle duration, calculate the interval time difference for the abnormal state interval and the recovery state interval respectively, and perform a step-by-step accumulation operation on the time difference for the same state to obtain the state cumulative duration vector; The generated set of state time intervals and the preset nursing sampling cycle duration parameter are used to perform interval span extraction operations on all abnormal state intervals and recovery state intervals contained in the set. First, for each independent time interval, the timestamps of its termination sampling point and the start sampling point are obtained, and the difference between the timestamps is calculated to obtain the duration of the single interval. Then, a discrete mapping operation is performed, dividing the calculated physical time span value by the unit time step of the nursing sampling cycle to quantify the time span into the corresponding sampling point count value, thereby achieving discretization of the time dimension. After completing the discrete mapping of all intervals, a term-by-term accumulation operation is performed on all discrete span values ​​belonging to the abnormal state category, and a term-by-term accumulation operation is performed on all discrete span values ​​belonging to the recovery state category, thereby obtaining the total value of abnormal states and the total value of recovery states within the entire nursing cycle. Finally, these two total values ​​are multiplied by the unit sampling cycle duration to convert them into a physically meaningful state cumulative duration vector. To visually illustrate this calculation process, Table 2 lists the specific interval span calculation and accumulation data.

[0028] Table 2 Calculation Table of Status Interval Span and Cumulative Duration within the Nursing Cycle Interval number State type Start sample point (sec) End sample point (sec) Single interval span (sec) Discrete sample point count (number) 1 Recovery 0 150 150 150 2 Abnormality 150 210 60 60 3 Recovery 210 800 590 590 4 Abnormality 800 840 40 40 5 Recovery 840 3600 2760 2760 As shown in Table 2, the recovery state interval 1 (0 to 150 seconds) in the initial sampling stage was first supplemented. For the abnormal state, intervals 2 and 4 were extracted. For interval 2, the difference between the end point 210 and the start point 150 was calculated to obtain 60 seconds; for interval 4, the difference between the end point 840 and the start point 800 was calculated to obtain 40 seconds. Then, these two span values ​​were substituted into the accumulation logic, i.e., 60 + 40 = 100, to obtain the cumulative duration of the abnormal state as 100 seconds. Similarly, for the recovery state, the 150 seconds of interval 1, the 590 seconds (800 - 210) of interval 3, and the 2760 seconds (3600 - 840) of interval 5 were summed, i.e., 150 + 590 + 2760 = 3500, to obtain the cumulative duration of the recovery state as 3500 seconds. The final generated cumulative duration vector contains these two components, and their sum (100+3500=3600) perfectly matches the total duration of the nursing cycle, ensuring the integrity of the monitoring data.

[0029] S203: For the cumulative duration vector of states, the cumulative duration of abnormal states and the cumulative duration of recovery states are structurally combined according to a unified time dimension. The combined result is identified by state type and the duration field is fixed. The abnormal state duration is filtered in combination with the preset abnormal duration threshold to generate an abnormal time structure. For the resulting cumulative duration vector of states, the calculated cumulative duration of abnormal states and cumulative duration of recovery states are structurally combined according to a unified time dimension to construct a vital sign timeliness analysis object containing fields such as "total abnormal duration," "total recovery duration," and "abnormal percentage." Subsequently, state type identification and duration field solidification operations are performed, writing specific duration values ​​into the corresponding fields. Based on this, the stability of the patient's physiological state is screened and graded using a preset abnormal duration threshold. This screening process first obtains the abnormal duration threshold parameter, which is set according to the clinical risk control standards of high-dependency wards, for example, 5% of the total nursing cycle time. Next, the ratio of the cumulative duration of abnormal states to the total nursing cycle time is calculated to obtain the abnormal time percentage, and this percentage is compared with the abnormal duration threshold. If the abnormal time percentage is greater than or equal to the abnormal duration threshold, the state type of the current cycle is identified as "high-risk continuous monitoring," and an abnormal time structure containing specific out-of-limit data is generated; if it is less than the threshold, it is identified as "routine rounds." In the actual example, the cumulative duration of abnormal states obtained from S202 is used as 100 seconds. The total duration of the nursing sampling cycle is set to 3600 seconds, and the abnormal duration threshold coefficient is set to 0.02 (i.e., 2%). First, the ratio of the cumulative duration of abnormal states of 100 seconds to the total duration of the nursing cycle of 3600 seconds is calculated, i.e., 100 / 3600 = 0.027. Then, this calculated result of 0.027 is compared with the preset threshold coefficient of 0.02. Since 0.027 is greater than 0.02, the cycle is determined to meet the abnormal screening condition. At this time, the generated abnormal time structure will clearly record: the cumulative duration of abnormal states is 100 seconds, exceeding the baseline by 0.007 (i.e., 0.7%), and the state is fixed as "high risk". The experimental results show that by introducing a comparison mechanism between cumulative duration and threshold, compared with the traditional method that only relies on instantaneous values ​​to trigger alarms, this scheme improves the risk detection rate in identifying intermittent but cumulative risk fluctuations in the patient's condition, effectively avoids missing the detection of long-cycle low-frequency abnormal signals, and ensures the accurate allocation of nursing resources in highly dependent wards.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the abnormal time structure, for the same vital sign within the current nursing monitoring cycle, retrieve the cumulative value of abnormal state time, obtain the total duration of the nursing monitoring cycle, perform a ratio calculation between the cumulative value of abnormal state time and the total duration of the nursing monitoring cycle, and generate the abnormal time occupancy ratio. Based on the generated abnormal time structure, for the same vital sign within the current nursing monitoring cycle, the system performs operations such as retrieving the accumulated time values ​​of abnormal and recovery states, aligning them within the same cycle, and splitting the fields. First, it accesses the vital sign timeliness analysis object stored in memory for the current nursing cycle, reading the recorded accumulated duration values ​​of abnormal and recovery states. To eliminate interference from invalid data periods caused by sensor detachment or signal interruption in the proportional calculation, the theoretical total duration of the nursing cycle is not used directly. Instead, the extracted accumulated durations of abnormal and recovery states are summed to obtain the total effective monitoring time within the cycle. Then, the accumulated duration of abnormal states is used as the numerator, and the total effective monitoring time is used as the denominator, performing a division ratio operation. The result is then rounded to four decimal places. For example, in a certain actual monitoring, the accumulated duration of abnormal states for heart rate parameters is 100 seconds (using data from previous steps), and the accumulated duration of recovery states is 3500 seconds. First, add 100 seconds to 3500 seconds to obtain a total effective monitoring time of 3600 seconds. Then, divide 100 seconds by 3600 seconds, the calculation is 100 / 3600 = 0.0278. This value represents the percentage of time the patient's heart rate was abnormal within the effective monitoring time. The advantage of this calculation logic is that by dynamically constructing an effective denominator, it eliminates the influence of invalid time caused by equipment failure, ensuring the purity of the quantification of the severity of the condition.

[0031] S302: Based on the abnormal time occupancy ratio of the current nursing monitoring cycle, the abnormal time occupancy ratio of the previous nursing monitoring cycle is collected at the same time. The ratio values ​​of the two cycles are arranged in time order to form a ratio sequence. An incremental relationship judgment is performed on adjacent ratio values ​​to generate a ratio increment judgment sequence. Based on the abnormal time occupancy ratio of the current nursing monitoring cycle, the abnormal time occupancy ratio of the previous nursing monitoring cycle is also collected. The ratio values ​​of the two cycles are arranged in chronological order to form a ratio sequence, and an increasing relationship judgment is performed on adjacent ratio values. A historical data queue based on timestamp index is maintained. When a new abnormal time occupancy ratio is generated, the abnormal time occupancy ratio of the same vital signs in the previous adjacent nursing cycle of the patient is immediately retrieved. If the data of the previous cycle does not exist (such as the first monitoring cycle), the ratio of the previous cycle is set to 0 or marked as incomparable by default; if valid data exists, the abnormal time occupancy ratio of the current cycle is compared with the abnormal time occupancy ratio of the previous cycle. If the ratio value of the current cycle is strictly greater than the ratio value of the previous cycle, the trend judgment result of that time node is marked as "increasing" and coded as 1; if the value of the current cycle is less than or equal to the value of the previous cycle, it is marked as "non-increasing" and coded as 0. On this basis, the judgment results of multiple consecutive cycles are arranged in chronological order to generate an increasing ratio judgment sequence. To more intuitively illustrate this judgment logic, Table 3 lists the ratio changes and judgment data of 5 consecutive nursing monitoring cycles.

[0032] Table 3. Trend Judgment Table for Abnormal Proportions in Continuous Nursing Monitoring Cycles Cycle index Monitoring time period Abnormality time occupancy ratio (%) Previous cycle ratio (%) Decision result (1 = increasing, 0 = non-increasing) T-4 08:00-09:00 2.50 - 0 T-3 09:00-10:00 3.10 2.50 1 T-2 10:00-11:00 4.29 3.10 1 T-1 11:00-12:00 4.29 4.29 0 T 12:00-13:00 5.60 4.29 1 Referring to Table 3, in period T-2, the proportion of abnormal time was 4.29%, an increase compared to 3.10% in period T-3, therefore the judgment result was 1; while in period T-1, the proportion remained at 4.29%, showing no increase, therefore the judgment result was 0. Capturing the dynamic change direction of the abnormal proportion can keenly identify early signs of disease deterioration. Experimental verification shows that, compared to static threshold analysis, this trend judgment mechanism provides an average early warning time of 45 minutes earlier for progressive diseases such as septic shock, significantly improving the timeliness of clinical intervention.

[0033] S303: Based on the increasing ratio determination sequence, the proportional items marked as increasing relationships within the continuous nursing monitoring cycle are sequence aggregated, the number of increases is counted, and the abnormal time occupancy ratio, the cumulative value of abnormal state time and the number of increases are associated and combined to generate an abnormal accumulation result. Based on the increasing proportion judgment sequence, the proportion items marked as increasing within the continuous nursing monitoring cycle are sequence aggregated. The aggregated increase count is then associated with the corresponding cycle index to generate an abnormal accumulation result. A continuous run statistical logic is used, with an increasing count accumulator. The increasing proportion judgment sequence is traversed forward along the time axis. Whenever an increasing marker coded as 1 is encountered, the current value of the accumulator is incremented by 1, indicating that the deterioration trend has continued for one cycle. Once a non-increasing marker coded as 0 is encountered, the accumulator is immediately cleared to zero, and the statistics restart. At the end of each cycle, the current accumulator value is bound to the index of that cycle as the "deterioration persistence level" at that moment. Taking the data in Table 3 as an example: In cycle T-3, it is judged as increasing, and the accumulator changes from 0 to 1; in cycle T-2, it is judged as increasing, and the accumulator changes from 1 to 2, indicating that the abnormal proportion has increased for two consecutive cycles; in cycle T-1, it is judged as non-increasing, and the accumulator is reset to 0; in cycle T, it is judged as increasing, and the accumulator returns to 1. The final accumulated anomaly result is this series of level values ​​(e.g., level 2 at time T-2). This level value is compared with a preset alarm trigger level (e.g., set to increase three times consecutively). If the current accumulated level reaches 3, a high-level alarm for "continuous deterioration of the condition" is triggered. The advantage of this calculation logic is that it not only focuses on the magnitude of a single anomaly but also emphasizes the inertia of the disease's evolution, effectively distinguishing between occasional fluctuations and pathological deterioration. Clinical test data shows that introducing this accumulation aggregation mechanism improves the accuracy of identifying disease progression in patients with chronic heart failure.

[0034] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on vital signs, collect the corresponding sampling sequences of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation during the current nursing monitoring cycle. Perform state switching detection for multiple sampling sequences, mark the sampling time point corresponding to the first transition from normal state to abnormal state, and generate an abnormal initial time point sequence. Based on the generated vital sign status sequence and the generated raw sampling data, complete sampling sequences corresponding to four physiological parameters—heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation—are retrieved within the current nursing monitoring cycle. State transition detection is performed independently for each sampling sequence to accurately pinpoint the moment when a physiological indicator deteriorates abruptly. A cursor pointer synchronized with the sampling frequency is established and traverses forward along the time axis starting from the beginning of the sequence. During the traversal, a two-point state comparison logic is executed, i.e., the state marker value at the current time point and the state marker value at the previous time point are obtained in real time. If the state marker at the previous time point is a value of 0 (representing normal) and the state marker at the current time point is a value of 1 (representing abnormality), then a "normal to abnormal" state transition has occurred at that moment. The current time point is immediately locked and defined as the "abnormal initial time point," and the corresponding absolute timestamp and the physiological parameter type label are extracted. For multiple repeated state transitions that may occur within a single nursing cycle, only the first trigger point of each consecutive abnormal interval is recorded, ignoring continuous abnormal markers within the interval. For example, in a real-world monitoring of a critically ill patient, scanning the heart rate parameter sequence revealed that the state marker at 10:05:09 was 0, and immediately following that, at 10:05:10, the state marker jumped to 1. 10:05:10 was then recorded as the onset of the heart rate abnormality. Similarly, if the systolic blood pressure state changed from 0 to 1 at 10:05:40, this was recorded as the onset of the systolic blood pressure abnormality. This scanning operation was performed in parallel on the four physiological parameters. All identified abnormal initial time points were collected and arranged vertically according to the order of their timestamps, constructing a sequence of abnormal initial time points containing "timestamp-parameter type" key-value pairs. The advantage of this operational logic is that by accurately capturing the edges of state transitions, continuous analog signal fluctuations are discretized into measurable key event nodes, providing a standardized time reference for analyzing the temporal causal relationships between different vital signs.

[0035] S402: Based on the sequence of abnormal initial time points, sort and analyze adjacent abnormal event pairs of multiple vital signs abnormal time points, determine whether the status identifier of the preceding vital sign is abnormal within the sampling interval corresponding to the abnormal time point of the subsequent vital sign, and write it into the sequence relationship field to generate an abnormal sequence relationship set. Based on the generated sequence of initial abnormal time points, a timeline sorting analysis and logical association determination of adjacent abnormal event pairs are performed on the time points of multiple vital signs abnormalities. First, all events in the sequence are paired in ascending order of timestamps to construct adjacent event groups. For any event pair consisting of a "preceding event" and a "following event," a state overlap verification operation is performed. The core logic of this operation is to determine whether the physiological abnormality represented by the preceding event induces or accompanies the occurrence of the following event. Specifically, the physiological parameter type and abnormal start time corresponding to the preceding event are read, and the abnormal time interval data corresponding to that parameter is retrieved to determine the real-time state of the preceding parameter at the time of the following event. If the following event is triggered before the abnormal interval corresponding to the preceding event has ended (i.e., the state identifier remains at the abnormal value 1), it is determined that there is an "accompanied temporal association." The preceding parameter type is used as the "source node," and the following parameter type is used as the "target node," and this association is written into the sequence relationship field. For example, using the data from the previous steps, select the abnormal heart rate event (10:05:10) as the preceding event and the abnormal systolic blood pressure event (10:05:40) as the following event. Query the abnormal duration range of the heart rate parameters, setting the abnormal heart rate state to last from 10:05:10 to 10:08:00. Compare the occurrence time of the following event (10:05:40) with the termination time of the preceding event (10:08:00). Since 10:05:40 is earlier than 10:08:00, it indicates that when the systolic blood pressure became abnormal, the heart rate was still in an abnormal state. Therefore, the sequential relationship of "abnormal heart rate precedes and accompanies abnormal systolic blood pressure" is established. Conversely, if the heart rate has returned to normal by 10:05:30, the association at that time point is not established. By performing the above verification on all adjacent event pairs in the sequence, an abnormal sequence relationship set containing all valid association pairs is generated. This process effectively eliminates independent abnormal events that occur purely by chance and accurately screens out cascade reaction chains with pathological causal potential.

[0036] S403: For the abnormal sequence relationship set, perform a counting operation on the relationship items that meet the condition of the continuation of the preceding abnormal state, filter the relationship items whose count value exceeds the abnormal sequence counting threshold, and index and associate the filtered count value with the corresponding vital sign combination identifier to generate abnormal evolution relationship; For the abnormal sequence relation set, classification counting and threshold filtering operations are performed on all relation items that meet the condition of the persistence of the preceding abnormal state. First, the entire relation set is traversed, and relation items with the same source node and target node combination (e.g., "heart rate-systolic blood pressure") are grouped and merged. The frequency of occurrence of each combination within the current nursing monitoring cycle is counted to obtain the sequence count observation value. Then, to eliminate occasional noise interference, an abnormal sequence count threshold is introduced for validity filtering. This threshold is set based on the principle of statistical significance and is dynamically generated by calculating the ratio of the total number of abnormal events to the parameter dimension. The specific calculation logic is as follows: obtain the total number of abnormal events recorded in the current cycle, divide it by the number of monitored parameter types, multiply the quotient by a preset confidence coefficient, and finally round the result up to obtain the abnormal sequence count threshold. For example, if a total of 20 abnormal events are recorded in the current cycle, and the monitored parameters are 4 types (heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen), the confidence coefficient is set to 0.5. First, 20 / 4 = 5, then 5 * 0.5 = 2.5. After rounding up, the abnormal sequence count threshold is determined to be 3. This means that a specific abnormal evolutionary pattern must occur at least 3 times to be considered a clinically significant evolutionary pattern. Based on this, the actual sequence count observations of each combination are numerically compared with this threshold. If the observation value is greater than or equal to the threshold, a strong index association is established between the combination identifier and the count value to generate the final abnormal evolutionary relationship; if it is less than the threshold, it is discarded. To visually illustrate this screening process and results, Table 4 lists the calculated abnormal evolutionary relationship data.

[0037] Table 4. Results of Anomaly Sequence Relationship Counting and Evolutionary Screening Evolution combination identification Sequential count observation value (number) Abnormality sequential count threshold value (number) Screening result Heart rate - systolic pressure 5 3 Valid Systolic pressure - oxygen saturation 2 3 Eliminate Heart rate - oxygen saturation 4 3 Valid Diastolic pressure - heart rate 1 3 Eliminate As shown in Table 4, the "heart rate-systolic blood pressure" combination occurred 5 times during the current monitoring period. Comparing this value with the calculated threshold of 3, since 5 is greater than 3, the evolutionary relationship is considered valid, indicating that the patient exhibits a typical pathological mechanism of blood pressure fluctuations caused by heart arrhythmia. The "systolic blood pressure-oxygenation" combination, however, occurred only 2 times, below the threshold of 3, and was therefore excluded as an occasional association. These experimental results demonstrate that a frequency-based dynamic threshold screening mechanism can automatically extract frequently recurring pathological evolutionary paths from complex fluctuations in vital signs. Compared to manual experience-based judgment, this improves the accuracy of identifying early signs of multiple organ failure and provides a quantitative basis for clinical development of interventional treatment plans.

[0038] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the abnormal evolution relationship, obtain the abnormal time occupancy ratio sequence of the abnormal accumulation result, perform an increasing status judgment on the ratio value of adjacent nursing cycles, record the continuous increasing identifier and perform time series mapping to generate an increasing abnormal time occupancy ratio sequence. Based on the generated abnormal evolution relationships and the generated abnormal accumulation results, the abnormal time occupancy ratio sequence corresponding to the current nursing monitoring cycle and several previous cycles is extracted from the historical database. An independent time sliding window is established for each patient, with the window length set to the most recent five consecutive nursing cycles. First, the abnormal time occupancy ratio value calculated for each cycle within the window is read and arranged in chronological order. Then, an increasing status judgment operation of the ratio values ​​of adjacent cycles is performed. This operation obtains the abnormal time occupancy ratio corresponding to the current time node and the abnormal time occupancy ratio corresponding to the previous time node, and performs a subtraction difference operation. If the difference is strictly greater than 0, that is, the current cycle ratio is higher than the previous cycle, then the current moment is determined to be in a "deteriorating increasing" state, and an increasing mark code of 1 is assigned; if the difference is less than or equal to 0, it is determined to be in a "non-increasing" state, and a mark code of 0 is assigned. This logic is executed for all adjacent pairs within the sliding window, generating a binary sequence composed of 0s and 1s. Based on this, in order to quantify the continuous strength of the increase, a time series mapping of continuous increasing marks is performed. This mapping process uses cumulative run-length encoding logic, traversing the binary sequence from left to right. If the current bit is 1, its value is set to "the cumulative value of the previous bit plus 1" (if it is the first bit, it is set to 1); if the current bit is 0, its cumulative value is reset to 0. For example, in a monitoring of a patient with septic shock, the abnormal time occupancy rates for five consecutive cycles were 2.1%, 2.6%, 3.8%, 3.8%, and 4.5%, respectively. First, the adjacent differences are calculated: 2.6 minus 2.1 is greater than 0 (marked 1), 3.8 minus 2.6 is greater than 0 (marked 1), 3.8 - 3.8 = 0 (marked 0), and 4.5 minus 3.8 is greater than 0 (marked 1). The initially generated binary sequence is 1-1-0-1. Then, intensity mapping is performed: the first 1 is accumulated to 1; the second 1, because its predecessor is also 1, is accumulated to 2; the third 0 is reset to 0; and the fourth 1, because its predecessor is 0, is accumulated to 1. The generated abnormal time occupancy rate increasing sequence is [1, 2, 0, 1].

[0039] S502: Based on the increasing sequence of abnormal time occupancy ratio, collect the length of the recovery period of multiple vital signs, calculate the cumulative sequence of recovery period time, and perform synchronization judgment with the increasing sequence of abnormal time occupancy ratio on the same time axis to generate a set of recovery status identifiers; Based on the generated increasing sequence of abnormal time occupancy ratios, the duration of recovery status for multiple vital signs in the current and historical nursing monitoring cycles is simultaneously collected. First, the calculated cumulative recovery status duration for each cycle is used to construct a cumulative recovery status time sequence. Then, this sequence and the increasing sequence of abnormal time occupancy ratios are synchronized on the same time axis to identify the concurrent characteristics of "deterioration of condition (increasing ratio)" and "decreased compensatory ability (reduced recovery time)." A synchronization judgment logic is defined: for any monitoring cycle, if the increasing marker for that cycle is greater than 0 (i.e., the abnormal percentage increases), and the cumulative recovery status duration for that cycle decreases compared to the previous cycle, it is judged as a "bidirectional negative feedback" state. To classify this dangerous state, the decay rate of recovery time is calculated. The calculation logic is as follows: first, the difference between the recovery time of the previous cycle and the recovery time of the current cycle is calculated; then, this difference is divided by the recovery time of the previous cycle, and the result is multiplied by 100 to obtain a percentage value. Different levels of step identifiers are generated based on the magnitude of the decay rate. The step identification rules are set as follows: If the attenuation rate is greater than 0 and less than 5%, it is marked as "Step I"; if the attenuation rate is between 5% (inclusive) and 15% (exclusive), it is marked as "Step II"; if the attenuation rate is greater than or equal to 15%, it is marked as "Step III". For example, the recovery time of the previous cycle (T-1) was 3150 seconds, the recovery time of the current cycle (T) was 2500 seconds, and the abnormal ratio of the current cycle was determined to be increasing. The attenuation amount is calculated as 3150-2500=650 seconds, and the attenuation rate is 650 / 3150=0.2063, or 20.63%. Since 20.63% is greater than 15%, "Step III" is generated. For example, the recovery time of the previous cycle (T-1) was 3580 seconds, the recovery time of the current cycle (T) was 3500 seconds (using the calculation result of S202), and the abnormal ratio of the current cycle was determined to be increasing. The calculated attenuation is 3580-3500=80 seconds, and the attenuation rate is 80 / 3580≈0.0223, or 2.23%. Since 2.23% is greater than 0 and less than 5%, a "step identifier I" is generated. This calculation is performed on all cycles that meet the bidirectional negative feedback condition, generating a set of recovery status identifiers containing timestamps and corresponding levels. Table 5 lists the specific data for this synchronization judgment process.

[0040] Table 5. Synchronous Judgment of Abnormal Proportion Increase and Recovery Time Decrease Cycle index Abnormality ratio increasing intensity Previous cycle recovery duration (s) Current cycle recovery duration (s) Recovery time decay rate (%) Generation step identification T-2 1 3600 3590 0.28 Step identification I T-1 0 3590 3580 - None T 1 3580 3500 2.23 Step identification I As shown in Table 5, at time T of the cycle, the proportion of abnormalities showed an increasing trend (intensity 1), and the recovery time decreased from 3580 seconds to 3500 seconds, with a decay rate of 2.23%. Although the patient still retained a relatively long recovery time, it showed a trend of weakened compensatory ability, thus triggering the primary warning "step indicator I". This result quantitatively reveals that although the patient's physiological state has not completely collapsed, it has entered an early warning stage where the recovery reserve has begun to be depleted due to frequent abnormalities. The experimental results show that by introducing the recovery time decay rate as a synchronous judgment dimension, it is possible to sensitively capture the subtle inflection point of the condition changing from stable to depleted.

[0041] S503: For the set of recovery status identifiers, call the cross-index abnormal evolution count results in the abnormal evolution relationship, perform logical judgment with the count conditions limited by the preset nursing early warning rules, perform structured encapsulation and index association on the simultaneously established relationship items, and generate abnormal early warning nursing intervention information; For the generated set of recovery status identifiers, the determined abnormal evolution relationships and their cross-index abnormal evolution count results are invoked, and multi-dimensional logical judgments are performed against the pre-defined nursing early warning rules that limit the counting conditions. First, the early warning rule library is loaded, which defines the combined triggering conditions for asynchronous identifier levels and evolution count thresholds. The rules are set as follows: for "Step Identifier," a specific evolution relationship (such as heart rate to systolic blood pressure) must be counted at least 5 times to trigger an early warning; for "Step II," the counting threshold is reduced to 3 times; for "Step III," since it represents extremely high risk, the counting threshold is further reduced to 1 time, meaning that as long as the evolution relationship is detected, it is triggered immediately. The step identifiers and evolution count data for the current period are traversed and compared item by item. If the conditions are met, a structured encapsulation operation is performed, packaging the current step identifier level, evolution combination type (such as "heart rate-systolic blood pressure"), specific count value, and the current values ​​of the involved physiological parameters to generate abnormal early warning nursing intervention information, and establishing an index association for quick retrieval. Taking the data of period T in Table 5 as an example, the current status is "Step I." The count value of the "heart rate-systolic blood pressure" evolution relationship is queried (referencing the observation value calculated in step S403, for example, 5 times). According to the rules, the threshold corresponding to step identifier I is 5 times. The actual count of 5 is compared with the threshold of 5. Since they are equal (meeting the condition of at least 5 times), the warning condition is determined to be met. The risk event is then packaged into an "early hemodynamic warning", which includes "risk level: attention (step identifier I)", "driving factor: heart rate-induced blood pressure fluctuations (count 5 times)", and "current status: mild decline in recovery ability (decline of 2.23%)". The advantage of this operation logic is that it adopts a dynamic threshold gating mechanism, raising the alarm threshold during the stable period of the condition (low step identifier) ​​to reduce interference, and lowering the threshold during the period of rapid deterioration of the condition (high step identifier) ​​to ensure no missed reports. Clinical empirical data show that this multi-dimensional cascade warning mechanism improves the effective intervention rate in the intensive care unit and shortens the average length of stay in the ICU.

[0042] Please see Figure 7 A nursing intervention system for early warning of abnormal vital signs in patients in highly dependent wards, including: The vital signs analysis module collects heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compares them with the preset normal nursing range, marks abnormal or recovery status, constructs vital signs status and transmits it to the abnormality statistics module. The anomaly statistics module, based on vital signs, analyzes the time span of all data in the nursing monitoring cycle for abnormal and recovery states, accumulates the time spans of abnormal and recovery states, generates an abnormal time structure, and transmits it to the anomaly analysis module. The anomaly analysis module calculates the ratio of the sum of the time of abnormal states to the sum of the time of abnormal and recovery states based on the abnormal time structure, analyzes the increasing relationship between the current and previous nursing cycles of abnormal time occupancy, generates the abnormal accumulation result and transmits it to the evolution judgment module. The evolution judgment module extracts and sorts the first abnormal sampling time points of all monitoring data in the current nursing monitoring cycle based on the vital signs status. It combines the abnormal accumulation results to judge and count the continuous abnormal states of the preceding vital signs when they appear in the subsequent cycle, generates abnormal evolution relationships, and transmits them to the nursing early warning module. The nursing early warning module, based on the abnormal evolution relationship, generates abnormal early warning nursing intervention information when the abnormal time occupancy ratio increases and the recovery time does not accumulate synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule.

[0043] 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 nursing intervention method for early warning of abnormal vital signs in patients in highly dependent wards, characterized in that, Includes the following steps: S1: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compare them with the preset normal nursing range, mark the normal, abnormal or recovery status, and construct the vital signs status; S2: Based on the vital signs status, analyze the time span of normal, abnormal or recovery states of all data, and accumulate the time spans of abnormal and recovery states to generate an abnormal time structure. S3: Based on the abnormal time structure, calculate the ratio of the cumulative abnormal state time to the total duration of the nursing monitoring cycle, analyze the increasing relationship between the current and previous nursing cycles' abnormal time occupancy ratios, and generate abnormal accumulation results. S4: Based on the vital signs status, extract the first abnormal sampling time point of all monitoring data in the current nursing monitoring cycle and sort them. Combine the abnormal accumulation results to determine the continuous abnormal state of the preceding vital signs when the subsequent vital signs are abnormal and count them to generate an abnormal evolution relationship. S5: Based on the abnormal evolution relationship, when the abnormal time occupancy ratio increases and the recovery state time is not accumulated synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule, abnormal early warning nursing intervention information is generated.

2. The method for early warning and nursing intervention of abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The vital signs status includes indicators for abnormal heart rate, abnormal systolic blood pressure, abnormal diastolic blood pressure, and abnormal blood oxygen saturation. The abnormal time structure includes time spans for abnormal states, time spans for recovery states, cumulative duration of abnormal states, and cumulative duration of recovery states. The abnormal accumulation result includes cumulative value of abnormal state time, total duration of nursing monitoring cycle, and abnormal time occupancy ratio. The abnormal evolution relationship includes a vital signs abnormality sequence index, a preceding abnormality persistence indicator, and a cross-indicator abnormality count value. The abnormal early warning nursing intervention information includes an early warning trigger indicator, a corresponding vital signs combination indicator, and a nursing intervention level indicator.

3. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Collect heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation monitoring data of patients in high-dependency wards during the nursing cycle, perform timestamp alignment and unit unification processing on each monitoring data, and arrange and aggregate fields in sequence according to the nursing sampling cycle to generate a vital signs sampling sequence; S102: Based on the vital signs sampling sequence, according to the normal ranges of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation in the nursing management guidelines, perform interval boundary comparisons on each monitoring data in the sequence, map the comparison results to abnormal, normal, or recovery markers, and combine them to generate a vital signs interval determination sequence. S103: Based on the vital sign interval determination sequence, perform state bit integration and logical merging on the labeling results of heart rate, systolic blood pressure, diastolic blood pressure, and blood oxygen saturation for each nursing sampling cycle, encode normal, abnormal, and recovery states into unified cycle state items and arrange them in chronological order to establish vital sign status.

4. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the vital signs status, the normal, abnormal and recovery status markers corresponding to heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation within the same nursing monitoring cycle are aligned on the time axis, and the start and end sampling points of multiple markers are mapped to continuous time intervals to generate a set of status time intervals. S202: Call the set of state time intervals and the nursing sampling cycle duration, calculate the interval time difference for the abnormal state interval and the recovery state interval respectively, and perform a step-by-step accumulation operation on the time difference for the same state to obtain the state cumulative duration vector; S203: For the cumulative duration vector of the state, the cumulative duration of the abnormal state and the cumulative duration of the recovery state are structurally combined according to a unified time dimension. The combination result is identified by the state type and the duration field is fixed. The abnormal state duration is filtered in combination with the preset abnormal duration threshold to generate an abnormal time structure.

5. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The abnormal duration threshold is determined by statistically analyzing the cumulative duration sample set of abnormal states obtained within the same nursing monitoring cycle, collecting the cumulative duration values ​​of abnormal states corresponding to multiple nursing sampling cycles, constructing a duration sequence, sorting it, and calculating the median value of the cumulative duration sequence of abnormal states.

6. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Based on the abnormal time structure, for the same vital sign within the current nursing monitoring cycle, retrieve the cumulative value of abnormal state time, obtain the total duration of the nursing monitoring cycle, perform a ratio calculation between the cumulative value of abnormal state time and the total duration of the nursing monitoring cycle, and generate the abnormal time occupancy ratio. S302: Based on the abnormal time occupancy ratio of the current nursing monitoring cycle, the abnormal time occupancy ratio of the previous nursing monitoring cycle is collected at the same time. The ratio values ​​of the two cycles are arranged into a ratio sequence according to the time order. An incremental relationship judgment is performed on adjacent ratio values ​​to generate a ratio increment judgment sequence. S303: Based on the increasing ratio judgment sequence, the proportional items marked as increasing within the continuous nursing monitoring period are sequence aggregated, the number of increases is counted, and the abnormal time occupancy ratio, the cumulative value of abnormal state time and the number of increases are associated and combined to generate an abnormal accumulation result.

7. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the vital signs status, collect the corresponding sampling sequences of heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation in the current nursing monitoring cycle, perform state switching detection for multiple sampling sequences, mark the sampling time point corresponding to the first transition from normal state to abnormal state, and generate an abnormal initial time point sequence. S402: Based on the abnormal initial time point sequence, sort and analyze adjacent abnormal event pairs of multiple vital signs abnormal time points, determine whether the status identifier of the preceding vital sign is abnormal in the sampling interval corresponding to the abnormal time point of the subsequent vital sign, and write it into the sequence relationship field to generate an abnormal sequence relationship set. S403: For the set of abnormal sequence relationships, perform a counting operation on the relationship items that satisfy the condition of the continuation of the preceding abnormal state, filter the relationship items whose count value exceeds the abnormal sequence counting threshold, and index and associate the filtered count value with the corresponding vital sign combination identifier to generate abnormal evolution relationship.

8. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 7, characterized in that, The preset abnormal sequence counting threshold is determined by collecting the abnormal sequence relationship count values ​​of all vital signs within the current nursing monitoring cycle and calculating the median statistic of the abnormal sequence relationship count value distribution.

9. The method for early warning and nursing intervention for abnormal vital signs in patients in highly dependent wards according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the abnormal evolution relationship, obtain the abnormal time occupancy ratio sequence of the abnormal accumulation result, perform an increasing state judgment on the ratio values ​​of adjacent nursing cycles, record the continuous increasing identifier and perform time series mapping to generate an increasing abnormal time occupancy ratio sequence. S502: Based on the increasing sequence of abnormal time occupancy ratio, collect the length of the recovery period of multiple vital signs, calculate the cumulative sequence of recovery period time, and perform a synchronization judgment with the increasing sequence of abnormal time occupancy ratio on the same time axis to generate a set of recovery status identifiers; S503: For the set of recovery status identifiers, call the cross-index abnormal evolution count results in the abnormal evolution relationship, perform logical judgment with the count conditions limited by the preset nursing early warning rule, perform structured encapsulation and index association on the simultaneously established relationship items, and generate abnormal early warning nursing intervention information.

10. A nursing intervention system for early warning of abnormal vital signs in patients in highly dependent wards, characterized in that: The system is used to implement the nursing intervention method for early warning of abnormal vital signs in patients in highly dependent wards as described in any one of claims 1-9, the system comprising: The vital signs analysis module collects heart rate, systolic blood pressure, diastolic blood pressure and blood oxygen saturation data of patients in high-dependency wards during the nursing cycle and compares them with the preset normal nursing range, marks abnormal or recovery status, constructs vital signs status and transmits it to the abnormality statistics module. The anomaly statistics module, based on the vital signs status, analyzes the time span of normal, abnormal, or recovery states of all data, accumulates the time spans of abnormal and recovery states, generates an anomaly time structure, and transmits it to the anomaly analysis module. The anomaly analysis module calculates the ratio of the cumulative abnormal state time to the total duration of the nursing monitoring cycle based on the abnormal time structure, analyzes the increasing relationship between the current and previous nursing cycles' abnormal time occupancy ratios, generates anomaly accumulation results, and transmits them to the evolution judgment module. The evolution judgment module extracts and sorts the first abnormal sampling time points of all monitoring data in the current nursing monitoring cycle based on the vital signs status, and combines the abnormal accumulation results to judge and count the continuous abnormal state of the preceding vital signs when the subsequent vital signs are abnormal, generates an abnormal evolution relationship and transmits it to the nursing early warning module. The nursing early warning module, based on the aforementioned abnormal evolution relationship, generates abnormal early warning nursing intervention information when the abnormal time occupancy ratio increases and the recovery state time does not accumulate synchronously, and the abnormal count result reaches the preset nursing early warning judgment rule.

Images