Gynecological surgery risk intelligent assessment system based on big data

Abstract

This invention relates to the field of data processing technology, specifically to a big data-based intelligent risk assessment system for gynecological surgery. The system includes a cycle positioning module, a dominant indicator identification module, a collaborative fluctuation detection module, a risk trend linkage module, and a graded path adjustment module. In this invention, by introducing and grouping time series data of menstrual cycles, biochemical hormones, and body temperature fluctuations during data processing, the dominant change patterns at each stage are identified through feature classification and trend extraction. Endocrine changes are cyclically mapped to preoperative clinical events. During surgery, nodes of physiological rhythm imbalance are dynamically identified based on the collaborative relationship of signs and trends. Furthermore, risk trend linkage segments are tracked in the postoperative indicator trajectory, and unstable indicators are matched and optimized with the risk path structure. This creates a closed-loop association between the preoperative cycle structure and the dynamics of intraoperative and postoperative signs, improving the individual adaptability and temporal consistency of the overall risk assessment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent risk assessment system for gynecological surgery based on big data. Background Technology

[0002] The field of data processing technology involves technologies related to the organization, transformation, analysis, and structuring of collected, acquired, or stored data. Core aspects include data format standardization, data cleaning and normalization, feature extraction and construction, design of data input structures required for model building, and data analysis processes and logical configurations for specific scenarios. This technology has wide applications in various industries such as medicine, finance, transportation, and manufacturing. Particularly in the medical industry, data processing can be used for assisted diagnosis, risk assessment, treatment plan optimization, and preoperative decision support. By combining multi-source data such as medical records, physiological parameters, and image data, and by developing data indicator systems and establishing judgment rules, clinical decision support systems can be constructed. Traditional gynecological surgery risk assessment systems rely on preoperative examination results and medical history data. Clinicians use their experience to manually assess the risk level or refer to paper-based scoring sheets. However, traditional methods rely on manual scoring based on factors such as age, surgical history, chronic disease status, and current lesion characteristics to evaluate potential risks and postoperative complications during surgery. These methods involve manual scoring based on criteria like the modified American College of Anesthesiologists (ACAS) grading system or a modified surgical risk index. Key information is manually summarized from medical records to form a judgment. Some systematized processes may also utilize manual input in Excel or partially embedded electronic medical records for data recording and preliminary risk classification.

[0003] Current technologies primarily rely on physician experience and manual scoring scales to assess preoperative risks. However, they cannot accurately capture the cyclical physiological differences in female patients, and the scoring dimensions are insufficient to cover factors influencing the endocrine cycle. This can lead to the overlooking of abnormal changes in individual vital signs at specific surgical stages. Manual summarization of vital sign information carries the risk of omissions and biased judgments. Under conditions where medical history and multi-source vital sign data are intertwined, traditional procedures struggle to establish a dynamic linkage between preoperative and postoperative risk chains. This results in abnormal intraoperative vital sign responses not being predicted in advance and untimely assessment of postoperative recovery risk points, affecting the timeliness and accuracy of overall decision-making. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a big data-based intelligent risk assessment system for gynecological surgery.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a big data-based intelligent risk assessment system for gynecological surgery includes: The cycle positioning module acquires the patient's menstrual cycle start time and hormone test data, sorts the cycle time axis, constructs cycle segment groups based on the start time point, marks the cycle segment groups in the preoperative examination data record items and performs collection processing to generate a cycle synchronization physiological feature set. The dominant indicator identification module, based on the set of periodic synchronous physiological features, calls the estrogen values, progesterone trend trajectory and body temperature fluctuation frequency band within the period segment, classifies the change amplitude and trend direction of the three features within each period segment, sets the features that show a continuous upward trend and maintain the dominant change pattern in the target period segment as the stage dominant features, and associates them with the clinical event records before gynecological surgery to generate a period dominant indicator mapping table. The collaborative fluctuation detection module, based on the cycle-dominant index mapping table, calls the heart rate flow curve, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory, and body temperature change segments from the real-time monitoring data during gynecological surgery. It judges the trend consistency and synchronous response cycle between each pair of data, identifies the trend reversal turning point, and judges the rhythm difference between the signs before and after. If the collaborative state has not recovered and the reversal trend continues to extend for multiple consecutive time periods, the time period is marked as the sign collapse segment. The module summarizes the types of signs and corresponding time anchors within the collapse segment and generates an interactive table of lost signs during surgery. The risk trend linkage module, based on the intraoperative lost sign interaction table, calls the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment from the postoperative monitoring records of gynecological surgery. It pairs the disintegration time anchor point with the postoperative indicator trend trajectory, and groups the corresponding indicators with similar time and continuously upward trend into the linkage indicator group, generating a postoperative indicator risk linkage trajectory map.

[0006] As a further embodiment of the present invention, the cycle synchronization physiological feature set includes cycle segment labels, synchronized sign trajectories, and endocrine change identifiers; the cycle dominant indicator mapping table includes dominant feature types, corresponding cycle segment numbers, and associated clinical event codes; the intraoperative disconnected sign interaction table includes sign type indexes, trend reverse anchors, and synergistic failure segments; and the postoperative indicator risk linkage trajectory diagram includes linkage indicator group classification, trend resonance time axis, and fluctuation persistence parameters.

[0007] As a further aspect of the present invention, the periodic positioning module includes: The menstrual cycle initiation identification submodule obtains the patient's menstrual cycle initiation time and patient hormone test data, extracts the sequence of continuous bleeding time points from the patient's menstrual records, determines the interval value of adjacent time points, filters time points that meet the cycle interruption conditions, calls the bleeding initiation records within the cycle segment, performs time axis annotation and sorting, and generates a list of cycle initiation times. The biochemical hormone time analysis submodule performs interval landing point judgment on the time point according to the cycle start time list, matches the corresponding cycle segment, and filters the detection points through the time distribution density of hormone data to obtain the hormone detection time mapping table. The cycle synchronization feature construction submodule calls the hormone detection time mapping table to construct the time series of hormone indicators for the cycle, calculates the aggregated value of hormone synchronization perturbation, performs position matching and gradient amplitude threshold judgment on the aggregated value of cycle hormone synchronization perturbation with the ovulation segment range, and establishes a set of cycle synchronization physiological features.

[0008] As a further aspect of the present invention, the dominant indicator identification module includes: The feature trend classification submodule, based on the periodic synchronous physiological feature set, calls the estrogen values, progesterone trend trajectory, and body temperature fluctuation frequency band within the period segment to sequentially obtain estrogen time series samples, progesterone segment samples, and body temperature spectrum samples. According to the division of the period segment, the three samples are segmented trend extracted, the interaction trend response value is calculated, and interval clustering is performed based on the range and polarity direction of the interaction trend response value to obtain physiological feature trend classification label groups. The phase-dominant extraction submodule classifies the physiological feature trend into a label group, calls the segment label classified as continuously rising, filters estrogen, progesterone and body temperature feature segments that are in an upward state for two or more consecutive segments, and determines whether the cumulative value of persistence and amplitude is the peak value of the real-time period segment. If the condition of continuous amplitude superposition is met, the corresponding feature is set as the dominant feature of the period segment, and the period-dominant physiological feature sequence is obtained. The clinical event mapping submodule calls the cycle-dominant physiological feature sequence, compares it with the preoperative clinical event records of gynecological surgery within the corresponding cycle segment, matches event feature items within a similar time range by comparing the event occurrence time with the time period of the dominant feature, records the corresponding cycle identifier and dominant feature label, and obtains the cycle-dominant indicator mapping table.

[0009] As a further aspect of the present invention, the cooperative fluctuation detection module includes: The data stream receiving submodule calls the periodic dominant index mapping table to sequentially collect heart rate flow curves, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory and body temperature change segments from the real-time monitoring data during gynecological surgery. The original data is synchronized and aligned according to the corresponding time anchor points. By recalibrating the time series data according to a unified sampling period, a structured vital signs time series is generated. The rhythm consistency determination submodule extracts the fluctuation amplitude, cycle length and phase difference between any two pairs of vital signs data within adjacent windows based on the structured vital sign time series. According to the set rhythm phase threshold and amplitude difference critical coefficient, it performs sliding window determination on the time series data of multiple pairs of vital signs, records whether the trend direction is consistent and whether the synchronization response cycle is within the rhythm matching interval, and generates a vital sign rhythm consistency identifier sequence. The trend reversal identification submodule calls the vital sign rhythm consistency identifier sequence to locate the positions where trend consistency is missing and rhythm coordination is ineffective in multiple consecutive windows, continuously tracks the continuous extension of the reverse trend, marks the vital sign type and corresponding time anchor point within the time interval, and obtains the intraoperative lost vital sign interaction table.

[0010] As a further aspect of the present invention, the risk trend linkage module includes: The postoperative indicator extraction submodule calls the sign time anchor point in the intraoperative lost sign interaction table to obtain the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment in the postoperative monitoring record of gynecological surgery. It performs trend curve extraction and time series standardization processing on the three types of postoperative indicators respectively, unifies multiple data to an aligned time axis based on the postoperative start time, and generates a standardized postoperative indicator trend sequence. The time anchor matching submodule constructs a fixed time window for the time anchor based on the standardized postoperative indicator trend sequence. Within the window, it calculates the trend slope of three types of data: postoperative fever, white blood cell count, and heart rate. Indicators with a slope that is continuously greater than zero and lasts for more than half of the time window are marked with a trend. The time difference between the trend mark and the time anchor is used as a matching factor to obtain the postoperative indicator linkage node set. The linkage trajectory submodule calls the postoperative indicator linkage node set. For each set of successfully matched indicator data, it constructs the linkage trend trajectory of three types of indicators, namely postoperative fever, white blood cell count, and heart rate, according to the order of intraoperative time anchor points. The time series in the trajectory is processed by multidimensional merging, and the indicator type, trend direction, and time position in the linkage segment are integrated and arranged in sequence to form a risk trend evolution channel, generating a postoperative indicator risk linkage trajectory map.

[0011] As a further aspect of the present invention, the system also includes a hierarchical path adjustment module: The risk assessment path adjustment module calls the core indicator set in the real-time risk assessment path based on the postoperative indicator risk linkage trajectory map, extracts the vital signs in the path and matches them with the vital signs in the linkage trajectory, and determines whether the trend stabilization indicators in the linkage trajectory include the core indicators in the assessment path. If the match is incomplete, the assessment path structure is adjusted, the assessment priority of the matching items is optimized, and the risk response structure of the assessment level is generated. The risk response structure for the assessment level includes matching body collection, priority assessment path, and adjustment level nodes.

[0012] As a further aspect of the present invention, the graded path adjustment module includes: The core indicator matching submodule calls the postoperative indicator risk linkage trajectory map, extracts the name of the vital signs, indicator weight and corresponding risk range of the path under the core indicator set in the real-time risk assessment level path, matches according to the indicator name, judges the trend stability of indicators whose trend direction remains unchanged in the linkage trajectory map, filters the vital signs whose trend duration exceeds the average assessment cycle, classifies the linkage indicators according to whether they are in the core indicator set, and obtains the trend stable core indicator matching table. The assessment structure reconstruction submodule, based on the trend stability core indicator matching table, adjusts the path structure for those with incomplete matching, rearranges the priority order of the matched indicators in the core indicator set, extracts the time position, trend length and amplitude fluctuation of the indicators in the postoperative indicator risk linkage trajectory diagram, constructs the indicator key ranking vector, embeds it into the original risk assessment level path structure, replaces the original priority arrangement, and generates the assessment level risk response structure.

[0013] As a further aspect of the present invention, the amplitude fluctuation is the difference between the peak value and the trough value within the continuous trend segment of the postoperative indicator risk linkage trajectory diagram; The trend length is the length of the time period in which the continuous trend direction remains unchanged in the postoperative indicator risk linkage trajectory diagram, and only when the trend length is greater than the set threshold for the duration of the trough trend, the corresponding indicator in the indicator key ranking vector is given a positive weight. The threshold for the duration of the trough trend is the weighted average of the average trend duration of the core indicator's concentrated vital signs. The priority order is rearranged according to the weight values ​​of the indicators in the indicator criticality ranking vector from largest to smallest. In the ranking result, indicators with equal weights are arranged according to the order of their first appearance in the postoperative indicator risk linkage trajectory diagram.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by incorporating and grouping time series data of menstrual cycles, biochemical hormones, and body temperature fluctuations during data processing, and identifying dominant changes in stages through feature classification and trend extraction, endocrine changes are periodically mapped to preoperative clinical events, thus achieving a stage-specific characterization of the individual physiological state of female patients. During surgery, nodes of physiological rhythm imbalance are dynamically identified based on the synergistic relationship of signs and trends, and risk trend linkage segments are tracked in the postoperative indicator trajectory. Unstable indicators and risk path structures are matched and optimized, so that the preoperative cycle structure and the dynamics of intraoperative and postoperative signs form a closed loop. The resulting assessment structure has the ability to verify continuous nodes, identify physiological rhythm variations, and optimize postoperative risk responses, which can improve the individual adaptability and temporal consistency of the overall risk assessment. Attached Figure Description

[0015] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the periodic positioning module in this invention; Figure 3 This is a flowchart of the main indicator identification module in this invention; Figure 4 This is a flowchart of the collaborative fluctuation detection module in this invention; Figure 5 This is a flowchart of the risk trend linkage module in this invention; Figure 6 This is a flowchart of the grade path adjustment module in this invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] Please see Figure 1 The big data-based intelligent risk assessment system for gynecological surgery includes: The cycle positioning module acquires the patient's menstrual cycle start time and hormone test data, sorts the cycle time axis, constructs cycle segment groups based on the start time point, marks the cycle segment groups in the preoperative examination data record items and performs collection processing to generate a cycle synchronization physiological feature set. The dominant indicator identification module, based on the cycle-synchronized physiological feature set, calls the estrogen values, progesterone trend trajectory, and body temperature fluctuation frequency band within the cycle segment. It classifies the change amplitude and trend direction of the three features within each cycle segment segment by segment. The feature that shows a continuous upward trend and maintains the dominant change pattern in the target cycle segment is set as the stage dominant feature and associated with the clinical event record before gynecological surgery to generate a cycle dominant indicator mapping table. The collaborative fluctuation detection module, based on the periodic dominant index mapping table, calls the heart rate flow curve, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory and body temperature change segments from the real-time monitoring data during gynecological surgery. It judges the trend consistency and synchronous response cycle between each pair of data, identifies the trend reversal turning point and judges the rhythm difference between the signs before and after. If the collaborative state has not recovered and the reversal trend continues to extend in multiple consecutive time periods, the time period is marked as the sign collapse segment. It summarizes the types of signs in the collapse segment and the corresponding time anchor points, and generates an interactive table of lost signs during surgery. The risk trend linkage module uses the intraoperative lost sign interaction table to call the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment from the postoperative monitoring record of gynecological surgery. It pairs the disintegration time anchor point with the postoperative indicator trend trajectory, and groups the corresponding indicators with similar time and continuously upward trend into the linkage indicator group to generate the postoperative indicator risk linkage trajectory map. The risk assessment path adjustment module calls the core indicator set in the real-time risk assessment path based on the postoperative indicator risk linkage trajectory map, extracts the vital signs in the path and matches them with the vital signs in the linkage trajectory, and determines whether the trend stabilization indicators in the linkage trajectory include the core indicators in the assessment path. If the match is incomplete, the assessment path structure is adjusted, the assessment priority of the matching items is optimized, and the risk response structure of the assessment level is generated. The cycle-synchronized physiological feature set includes cycle segment labels, synchronized sign trajectories, and endocrine change identifiers. The cycle-dominant indicator mapping table includes dominant feature types, corresponding cycle segment numbers, and associated clinical event codes. The intraoperative lost-sign interaction table includes sign type indexes, trend reverse anchors, and synergy failure segments. The postoperative indicator risk linkage trajectory diagram includes linkage indicator group classifications, trend resonance time axes, and fluctuation persistence parameters. The assessment level risk response structure includes matched sign sets, priority assessment paths, and adjustment level nodes.

[0019] Please see Figure 2 The periodic positioning module includes: The menstrual cycle initiation identification submodule obtains the patient's menstrual cycle initiation time and patient hormone test data, extracts the sequence of continuous bleeding time points from the patient's menstrual records, determines the interval value of adjacent time points, filters time points that meet the cycle interruption conditions, calls the bleeding initiation records within the cycle segment, performs time axis annotation and sorting, and generates a list of cycle initiation times. The continuous bleeding time series during menstruation was extracted from the patient's menstrual records. The start time of each bleeding event was recorded as a timestamp. The time interval difference between adjacent bleeding start times was calculated sequentially. Each time difference was compared with the known standard interval range of a normal menstrual cycle, which is 25 to 35 days. If a time difference falls within this range during the calculation, the corresponding bleeding start time is used as a candidate menstrual cycle start time for preliminary screening. At the same time, the stability index of adjacent cycles is calculated for the candidate time points that meet the conditions. The standard deviation σ within three cycles is used to judge. If σ ≤ 2 days, the cycle regularity is considered to meet the requirements and is determined as the cycle start time. A list of cycle start times is constructed based on the start time of each cycle, as shown in Table 1. Table 1: Periodic Segmentation Table; As shown in Table 1, the stability score of the cycle starting point is obtained by statistically analyzing the standard deviation of three consecutive cycles. When it is less than 2 days, the cycle can be determined to be stable. Taking the patient's menstrual start dates recorded on January 1st, January 30th, and February 28th, 2025 as examples, the cycle intervals were 29 and 30 days, with a standard deviation of [missing information]. ≈0.5 days, which meets the cycle determination condition, so it is recorded in the cycle start time list in sequence.

[0020] The biochemical hormone time analysis submodule performs interval drop-off judgment on time points based on the cycle start time list, matches the corresponding cycle segment, and filters the detection points through the time distribution density of hormone data to obtain the hormone detection time mapping table. The cycle is divided into 5 time zones: menstrual period, early follicular phase, mid-follicular phase, ovulation period, and luteal phase. A zone-matching operation is performed on the hormone testing time points, and the difference between the testing time and the corresponding cycle start time needs to be calculated in this step. Based on the cycle length, the above regions are divided, and the time range for each region is set as follows (referencing relevant literature): menstrual period is from the start of menstruation to day 5; early follicular phase is day 6 to 10; mid-follicular phase is day 11 to 14; ovulation period is day 15 to 17; and the luteal phase is from day 18 to the end of the cycle. The range of values ​​directly determines the cycle stage to which the detection time point belongs, and a mapping table is constructed based on this to record the relationship between hormone values ​​and cycle position. For example, if a patient's estrogen is measured on day 13 of cycle 2, =13, falling into the mid-follicular phase, therefore classified as the mid-follicular phase, and the mapping information is recorded; the density of detection points in each phase is calculated, the detection frequency per unit time is counted, high-density monitoring points are selected, and the density standard is set to ≥3 detections per phase. If the phase density meets the standard, the phase hormone change data is included in the subsequent feature extraction process to obtain the hormone detection time mapping table.

[0021] The periodic synchronization feature construction submodule calls the hormone detection time mapping table to construct the periodic hormone index time series, using the formula: ; Calculate the aggregated value of hormone synchronization perturbation, match the aggregated value of cycle hormone synchronization perturbation with the range of ovulation segment and determine the gradient amplitude threshold, and establish a set of cycle synchronization physiological features; in, This indicates the aggregation value of hormone synchronization perturbation. Indicates the first Estrogen levels at various time points This represents the average estrogen level within a real-time cycle. Indicates the first The rate of change in hormone levels at different time points Indicates the first The position of the time point on the periodic time axis. Indicates the center position of the ovulation time segment. Indicates the first The extent of diffusion of changes in the hormone gradient at different time points. Indicates the number of time points; Formula calculation logic: By integrating the deviation of hormone values ​​from the cycle mean, the rate of hormone change, the time difference from the ovulation reference point, and the density of monitoring points, a comprehensive calculation process involving multiple factors is formed. First, the absolute value of the difference between hormone values ​​and the cycle mean... This reflects the degree of fluctuation at this detection point during cyclical hormone changes, and the corresponding rate of change. This measures the magnitude of hormone level changes; the faster the change, the more significant the change. The time difference between the result and the ovulation reference point is also considered. The sums are then taken as the reciprocal to form a weighted factor for the period closest to ovulation. The closer to ovulation, the larger the value. This is then divided by the square root of the density factor. In order to balance the impact of the detection time distribution on the overall results, the above four types of indicators jointly affect the weighted value at each time point through the multiplication and division structure, forming the hormone synchronous perturbation aggregate value within the period; Hormone synchronization perturbation aggregation value is a comprehensive numerical indicator that measures the intensity and speed of hormone changes during the menstrual cycle and their synchronicity with ovulation time. The higher the value, the greater and faster the hormone fluctuations during the cycle, and the more concentrated they are in the time region close to ovulation. It is calculated by integrating four factors: hormone amplitude deviation, speed, time distance, and detection density. It is an important basis for identifying abnormal hormone changes and ovulation rhythm. Parameter meaning and calculation process: Indicates the first Hormone levels at specific time points; This represents the average value of the hormone during the current cycle; For the first The rate of change in hormones at a given point is defined as... ; This represents the time elapsed between the current point and the start of the cycle. The reference point for ovulation is set at day 14. The detection time interval between this point and the previous point. Given the total number of detection points within a cycle, taking a certain cycle as an example, the recorded hormone value array is [80, 130, 190, 250], corresponding to detection times of [11, 12, 13, 14] days, we can obtain: ; each : , , ; each : Detection time [11, 12, 13, 14]; ; each The detection interval is 1. Substitute into the formula and calculate each term: Point 2: ; Point 3: ; Point 4: ; Substitute into the formula to calculate: ; The results indicate that the hormone synchronization perturbation aggregation value is 860.98, which means that the hormone value in the current cycle deviates greatly from the mean and changes rapidly. Ovulation is approaching and the HA value is high, indicating that the ovulation period is significant. The advantage of the formula is that it can aggregate and quantify the synchronous changes of hormones by combining multiple factors such as the degree of deviation of hormone values, the rate of change, and the degree of proximity of time. After calculating the periodic hormone synchronization aggregation value, the corresponding value for each period is... Values ​​are segmented and clustered according to The values ​​are divided into intervals, which will be divided into low... Segment (HA<200), Middle Section (200≤ <500), High part( (≥500), and construct an ovulation location matching mapping table.

[0022] Please see Figure 3 The main indicator identification module includes: The feature trend classification submodule, based on the periodic synchronous physiological feature set, retrieves estrogen values, progesterone trend trajectories, and body temperature fluctuation frequency bands within the period segment. It sequentially obtains estrogen time-series samples, progesterone segment samples, and body temperature spectrum samples. According to the period segment division, it performs segmented trend extraction on the three samples using the following formula: ; Calculate the interaction trend response value, perform interval clustering based on the range and polarity of the interaction trend response value, and obtain physiological characteristic trend classification label groups; in, Indicates the interactive trend response value. Indicates the first The first in the periodic segment Estrogen changes at each sampling point Indicates the first The body temperature fluctuation coefficient after standardization of each sampling point Indicates the first Duan Di In the class of samples, the first Modulation factor at each sampling point This represents the total number of sampling points within the periodic interval; Formula calculation logic: By weighting and fusing hormonal fluctuations, body temperature changes, and regulatory factors for each type of physiological sample within the cycle, a quantitative indicator reflecting the cyclical changes in physiological state is formed. Calculate hormone bias, i.e., the first The difference between the hormone value of each sample point and the mean of the class samples is used to represent the degree of local hormone abnormality. The standardized body temperature fluctuation coefficient is then used. Multiplied by the adjustment factor To quantify the amplification or inhibition of hormone fluctuations by body temperature, the regulatory factor is set by the position of the sample point in the sampling segment, reflecting the significance of structural regulation. The absolute value of the sum of the two factors can eliminate the influence of positive and negative differences. The summation of the sample points and the division by the number of periods are then performed. The average response intensity of TG was obtained, which served as the aggregate expression of physiological trend changes within the cycle. The interaction trend response value represents the synchronous coupling strength between hormone fluctuations and body temperature fluctuations within a specific period. It combines the degree of hormone variation and the amplitude of body temperature change, and reflects the interaction trend between different physiological parameters according to the weighting of regulatory factors. The higher the value, the more drastic the physiological changes in this type of sample, and the clearer the trend, which is convenient for subsequent classification and identification. Parameter meaning and calculation process: The starting point for sample extraction was the start and end times recorded by the segment number in the period segment division table. Within each period, the period from day 10 to day 17 was the period of high progesterone variation, and the sampling frequency was set to once a day, corresponding to the extraction of 8 progesterone sample points. The sampling area for body temperature spectrum samples was set from day 6 to day 20 of the period. Three groups of samples were extracted using the spectral window sliding method, with each group lasting 5 days and the window step size being 3 days, forming three time period samples to represent the body temperature changes at different stages of the period. The extraction interval for estrogen sample sequences was set from day 3 to day 16 of the period, and was extracted once a day, for a total of 14 sample points. The three types of samples were numbered as type 1, type 2, and type 3 samples according to the extraction order, and the sample sequences for each type in Q periods were constructed. Indicates the first In the class of samples, the first The estrogen variation value corresponding to each point is calculated as follows: , This represents the hormone level at the current sampling point. For the first Class sample mean; For the first The body temperature amplitude coefficient after standardization is defined as the normalized value of the difference between the temperature and the sample mean multiplied by the coefficient α. α was set to 1.5 to enhance sensitivity to temperature abrupt changes; For the sample number The first paragraph The modulation factor of the point, according to the first After the position of the point in the sample segment is normalized, it is weighted with 0.5 for the first segment, 1.0 for the middle segment, and 1.2 for the last segment. The regulatory effect of temperature on hormones is quantified by the weighted structure. Q represents the total number of samples in the period, which is set to 5 period samples, i.e. Q=5. The "addition" and "multiplication" used in the formula structure reflect the linear relationship between multiple factors, while taking the absolute value reflects the non-directional influence of the change, and then averaging to eliminate the influence of the number of periods on the scale of the result. The actual data is as follows: In periods 1 to 5, the hormone values ​​at point 1 of the first type of sample are [120, 115, 123, 119, 117], with a mean of 120.8. The corresponding temperature-standardized values ​​are [0.5, 0.4, 0.6, 0.5, 0.45]. Assuming the modulation factor μ is [1.0, 1.0, 1.0, 1.0, 1.0], the calculation is as follows: Item 1: ; Item 2: ; Item 3: ; Item 4: ; Item 5: ; Substitute into the formula: ; Table 2: Sample values ​​of physiological characteristics and Value calculation table; As shown in Table 2, the first type of samples were obtained. The value is 2.63. This value will be used in the subsequent judgment of physiological characteristic trend clustering as a trend response index for this type of sample. The advantage of the formula is that by multiplying the estrogen variation value with the body temperature fluctuation and introducing a modulation factor, the response performance of the trend fluctuation within the sample is enhanced, so that the signals of different physiological channels can be uniformly included in the aggregation judgment.

[0023] The phase-dominant extraction submodule categorizes labels based on physiological feature trends, calls up labels for segments that are continuously rising, filters estrogen, progesterone, and body temperature feature segments that are rising for two or more consecutive segments, and determines whether the cumulative value of persistence and amplitude is the peak value of the real-time period segment. If the condition of continuous amplitude superposition is met, the corresponding feature is set as the dominant feature of the period segment, and the period-dominant physiological feature sequence is obtained. In the completed Perform continuous trend clustering calculations on the indicator vectors, and set a threshold for continuous increase. Set to 1.0, the continuous trend points must meet the following requirements. The value increases monotonically and continuously across the three sampling segments. If the TG point sequence is [1.2, 2.3, 3.0], and the difference between each item is greater than a set threshold. If the temperature amplitude is higher than the peak value of the cycle hormone, it is identified as an upward trend segment. The analysis then determines whether the estrogen levels in this segment are within the peak range of the cycle hormone. If the hormone values ​​at the corresponding sampling points are all greater than the peak value threshold (set to 180 pmol / L), and the hormone values ​​in the three segments are 185, 210, and 198 respectively, then the peak condition is met, and the trend segment is considered to have dominant characteristics. Next, the analysis examines whether the temperature amplitude shows a continuous upward trend. The standard for a rise in body temperature is defined as a continuous increase for three consecutive days with a daily difference ≥ 0.1℃. If the temperature records are 36.5, 36.7, and 36.9, then this is considered a dominant trend segment. The segments that meet the criteria of trend, peak value, and amplitude are then identified as dominant trend segments. The label is marked as the dominant label and classified into the cycle-dominant physiological characteristic sequence.

[0024] The clinical event mapping submodule calls the cycle-dominant physiological feature sequence, compares it with the gynecological surgery preoperative clinical event records within the corresponding cycle segment, matches event feature items within a similar time range by comparing the event occurrence time with the time period of the dominant feature, records the corresponding cycle identifier and dominant feature label, and obtains the cycle-dominant indicator mapping table. Specific time points such as gynecological surgery, ovulation induction drug use, and infertility diagnosis were located in patient records, and an event time series table was constructed. The table records the event occurrence time, event type, cycle number, number of days before and after the event, and relative distance from the dominant feature location. The time point of each event was compared with the dominant feature of the cycle. The absolute distance difference is calculated, with the distance unit being days. The matching window is set to ±3 days, meaning if the event occurs during the dominant period... Within 3 days before and after a point, it is considered a relevant event and a tag is mapped. For example, if event A occurs on day 13 of the cycle, it is considered the dominant event. If the feature appears on day 12, the difference is 1 day, which satisfies the matching window. Event A is mapped to the dominant feature and the mapping table is shown below. Table 3: Mapping Table of Clinical Events and Dominant Characteristics; Table 3 lists the clinical events and their main components. The mapping relationship of features shows that events A and C both overlap with the dominant feature in time, so they can be included in the analysis label sequence for subsequent analysis.

[0025] Please see Figure 4 The collaborative fluctuation detection module includes: The data stream receiving submodule calls the cycle-dominant indicator mapping table, sequentially collecting heart rate flow curves, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory, and body temperature change segments from real-time monitoring data during gynecological surgery. The original data is synchronized and aligned according to the corresponding time anchor points. By recalibrating the time series data according to a unified sampling period, a structured vital signs time series is generated. The composition of the "Periodic Dominant Indicator Mapping Table" needs to be clearly defined, including multiple feature values ​​used to determine the periodicity of vital sign data and their corresponding sampling periods. Typical frequency domain features (such as dominant frequency, sampling period, and data window length) should be defined for heart rate (HR), blood pressure (BP), blood oxygen saturation (SpO2), and body temperature (Temp). For example, heart rate is 1Hz, blood pressure is 0.5Hz, blood oxygen saturation is 0.1Hz, and body temperature is 0.01Hz. In gynecological surgery scenarios, these four types of data are acquired in real time from multimodal monitoring devices (such as the Philips IntelliVue series). Each vital sign data point is collected into a buffer via serial port or Ethernet port, and the buffer is maintained according to the FIFO principle. Continuing the update, the heart rate flow curve is extracted within the set sampling period, with values ​​recorded once per second (BPM), such as 70, 72, 74, etc.; blood pressure is recorded as systolic and diastolic blood pressure values ​​every 5 seconds, such as [120 / 80], [122 / 81], etc.; percutaneous oxygen saturation is recorded as SpO2 percentage values ​​once per second, such as 98%, 97%; body temperature is recorded once per minute, such as 36.5℃, 36.6℃. According to the anchor point sequence set by the surgical timeline (such as key time nodes such as scalpel insertion, anesthesia onset, instrument insertion, etc.), the various types of data are linearly interpolated and aligned using timestamps, unifying the original data at different frequencies to the same time reference, resulting in a structured vital sign time series.

[0026] The rhythm consistency determination submodule is based on structured vital sign time series. It extracts the fluctuation amplitude, period length and phase difference between any two vital sign data in adjacent windows. According to the set rhythm phase threshold and amplitude difference critical coefficient, it performs sliding window determination on the time series data of multiple pairs of vital signs, records whether the trend direction is consistent and whether the synchronization response period is in the rhythm matching interval, and generates a vital sign rhythm consistency identifier sequence. First, key rhythm parameters between various vital signs need to be extracted within the same time window. For example, the fluctuation amplitude of heart rate and blood pressure within the 0-30s window can be obtained by subtracting the minimum value from the maximum value. If the heart rate is set to [70, 74, 73], the amplitude is 4; if the estimated arterial blood pressure is [110, 115, 108], the amplitude is 7. The period length can be estimated using the autocorrelation function. For example, if the maximum value of Rxx in the heart rate time series Rxx(τ) occurs when τ=5s, the period is 5 seconds. The phase difference can be calculated from the lag corresponding to the maximum cross-correlation value of the two series. The maximum cross-correlation value of heart rate and blood pressure is set... If the value appears at a lag of +2 seconds, the phase difference is 2 seconds. The system calculates the amplitude difference between heart rate and blood pressure according to the set threshold, such as the rhythm phase threshold Δφ being ±3s and the amplitude difference critical coefficient being 20%. The difference is |4-7| / ((4+7) / 2)=3 / 5.5≈54.5%. If the threshold is exceeded, the vital signs are considered to lack rhythm consistency. During the repeated judgment process with a sliding window of 10 seconds, if the trend direction (such as rising or falling) is inconsistent or the period mismatch window is greater than 3, the trend inconsistency in that time period is recorded, and a vital sign rhythm consistency identifier sequence is generated.

[0027] The trend reversal identification submodule calls the vital sign rhythm consistency identifier sequence to locate the location of trend consistency loss and rhythm coordination failure in multiple consecutive windows, continuously tracks the continuous extension of the reverse trend, marks the vital sign type and corresponding time anchor point within the time interval, and obtains the intraoperative lost vital sign interaction table. Load the consistency records of vital signs. Set HR-BP and HR-SpO2 to "mismatch" in three consecutive windows from 00:01:00 to 00:01:30. The system identifies this as a segment with missing trend consistency. By analyzing the direction of change of vital sign values ​​in each window, such as HR [76→78→81] indicating a continuous increase and BP [110→107→105] indicating a continuous decrease, if the trend direction is opposite, it is marked as a reverse trend. If three consecutive windows maintain a reverse trend, it is marked as a continuous reverse trend segment. The system records the start time point as 00:01:00 and the end time as 00:01:30. It marks the vital sign types HR and BP, and HR and SpO2 involved in the segment as missing interaction pairs in sequence. At the same time, the time anchor points in the window are marked synchronously. For example, the interval corresponds to the start and end time of the intraoperative "uterine clamping" operation. Construct an intraoperative missing vital sign interaction table.

[0028] Please see Figure 5 The risk trend linkage module includes: The postoperative indicator extraction submodule calls the sign time anchor point in the intraoperative lost sign interaction table to obtain the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment in the postoperative monitoring record of gynecological surgery. The three types of postoperative indicators are subjected to trend curve extraction and time series standardization processing respectively. Multiple data are unified to the aligned time axis based on the postoperative start time to generate a standardized postoperative indicator trend sequence. Extract the start and end times of the loss of vital signs, such as [00:01:00–00:01:30], corresponding to the intraoperative "uterine clamping" node. Retrieve postoperative data within 24 hours after the anchor point in the postoperative monitoring record. For example, the fever curve can be obtained from the hourly temperature value obtained from the electronic medical record temperature chart, set to [36.8, 37.2, 37.8, 38.2, 38.5]. The white blood cell response trajectory can be obtained from the WBC values ​​obtained from the postoperative 0h, 6h, 12h, and 24h blood test reports, such as [7.8×10]. 9 / L, 9.2×10 9 / L, 11.3×10 9 / L, 12.5×10 9 [ / L], the heart rate recovery segment is obtained by extracting heart rate values ​​every 15 minutes from postoperative monitoring records, such as [86, 84, 82, 80]; for the three types of indicators, an aligned time axis with the postoperative start time as T0 is constructed respectively, and a sequence T0=[0, 15, 30, ..., 1440] is established with 15 minutes as the time unit. Interpolation and smoothing algorithms (such as linear interpolation, moving average) are used to map WBC, Temp, and HR values ​​to this time axis respectively. Time normalization is performed on data with different sampling frequencies to form a standardized postoperative indicator trend sequence.

[0029] The time anchor matching submodule constructs a fixed time window for the time anchor based on the standardized postoperative indicator trend sequence. Within the window, it calculates the trend slope of three types of data: postoperative fever, white blood cell count, and heart rate. Indicators with a slope that is continuously greater than zero and lasts for more than half of the time window are marked with a trend. The time difference between the trend mark and the time anchor is used as a matching factor to obtain the set of postoperative indicator linkage nodes. Using each postoperative time anchor point (e.g., postoperative start T0) as a reference, a fixed time window W is constructed, with a length of 180 minutes. The trend slope of each type of indicator curve within the window is calculated, and the slope k can be obtained from the linear fitting formula k=(y n -y1) / (t n-t1) yields the following: In the temperature sequence [36.8, 37.0, 37.5, 38.0], k = (38.0 - 36.8) / (180 - 0) = 0.0067℃ / min. If the length of a segment with a slope continuously greater than 0 exceeds 90 minutes (half of the window), the trend is marked as "rising". For WBC values ​​such as [8.5, 9.5, 10.6, 11.7], the slope is calculated to be (11.7 - 8.5) / 180 = 0.0178 × 10⁻¹⁰. 9 / L·min, satisfying a continuous upward trend; if the heart rate is [86, 85, 84, 83] and the slope is negative, it is not marked; the difference between the time of occurrence of the marked trend and the T0 time is calculated as the matching factor. If ΔT is within 0-180min, the matching is successful, and the postoperative indicator linkage node set is obtained.

[0030] The linkage trajectory submodule calls the postoperative indicator linkage node set. For each set of successfully matched indicator data, it constructs the linkage trend trajectory of three types of indicators, namely postoperative fever, white blood cell count, and heart rate, according to the order of intraoperative time anchor points. It performs multi-dimensional merging processing on the time series in the trajectory, integrates the indicator type, trend direction, and time position in the linkage section, arranges them in sequence to form a risk trend evolution channel, and generates a postoperative indicator risk linkage trajectory map. For successfully matched indicators such as Temp and WBC, a postoperative linkage trend trajectory is constructed according to the intraoperative time anchor point sequence, i.e., nodes such as "uterine clamping" → "suture hemostasis" → "debridement and drainage" as T1, T2, and T3. After each anchor point, it is checked whether there are Temp, WBC, and HR indicators that simultaneously have trend markers. If the time difference of the three indicators is within the allowable range (e.g., 30 minutes), they are merged into the same linkage segment, such as [postoperative T1+45min, Temp↑, WBC↑]; the linkage trajectory sequence is constructed as: [T1+45min, [Temp↑, WBC↑]]. [T2+60min, [Temp→, WBC↑, HR↓]]; Perform multidimensional merging processing on each trajectory segment to encode and integrate the states of each indicator within the same time period, such as ↑=1, ↓=-1, →=0, forming a multidimensional state vector. Set [1, 1, 0] to represent Temp↑, WBC↑, HR→. Pair the state vector with the corresponding time and sort them, such as [(T1+45min, [1, 1, 0]), (T2+60min, [0, 1, -1])]. Connect them in chronological order to form a risk trend evolution channel and form a postoperative indicator risk linkage trajectory map.

[0031] Please see Figure 6 The graded path adjustment module includes: The core indicator matching submodule calls the postoperative indicator risk linkage trajectory map, extracts the names of vital signs, indicator weights and corresponding risk ranges under the path of the core indicator set in the real-time risk assessment level path, matches them according to the indicator names, determines the trend stability of indicators with continuous and unchanged trend direction in the linkage trajectory map, filters vital signs whose trend duration exceeds the average assessment cycle, classifies the linkage indicators according to whether they are in the core indicator set, and obtains the trend stable core indicator matching table. Read the core indicator set from the risk level path. The core indicator set includes vital sign names [body temperature (Temp), white blood cell count (WBC), heart rate (HR)], indicator weights [Temp: 0.35, WBC: 0.4, HR: 0.25], and their risk ranges such as [Temp > 38.5℃, WBC > 12 × 10⁻⁶]. 9 [ / L, HR>100bpm], the system extracts the trend direction sequence and time position of each indicator under the path from the linkage trajectory map, and judges whether the trend direction remains unchanged in the continuous segment. The Temp value sequence is set as [37.5, 37.8, 38.2, 38.6, 38.9] and it continues to rise for more than 60 minutes. Then the trend direction "↑" is considered stable. The trend stability judgment method is: set the average evaluation period Tavg=45 minutes. If the trend duration Ts≥Tavg, the trend is marked as a stable item. For example, if the Temp trend segment Ts=60min>45min, it is retained. Items that appear in the core indicator set in the trend stable items are marked as "matched". For example, Temp and WBC are matched items. HR is marked as "non-matched" or "non-core" because it has no continuous trend or is not in the core indicator set. The trend stable core indicator matching table is output.

[0032] The assessment structure reconstruction submodule is based on the trend stable core indicator matching table. It adjusts the path structure that does not match completely, rearranges the priority order of the matched indicators in the core indicator set, extracts the time position, trend length and amplitude fluctuation of the indicators in the postoperative indicator risk linkage trajectory map, constructs the indicator key ranking vector, embeds it into the original risk assessment level path structure, replaces the original priority arrangement, and generates the assessment level risk response structure. The system identifies path nodes in the assessment level path structure that lack matching indicators. For example, if the original path is [Temp→HR→WBC], and HR fails to match successfully in the linkage trajectory diagram, the path structure needs to be adjusted. The system retains the matched Temp and WBC in the path, removes HR, and reconstructs the path to [Temp→WBC]. Then, it extracts the trend time position, length, and fluctuation amplitude from the matched items. The fluctuation amplitude ΔV can be calculated from the difference between the maximum and minimum values. Setting the Temp segment to [37.8, 38.9], then ΔV = 1.1℃. The trend length L_t is the trend duration, such as 60 minutes. The system constructs an index using the time position Tstart, ΔV, and Lt of each indicator as a three-dimensional vector Vi = [Tstart, ΔV, Lt]. The key ranking vectors are defined as follows: Temp is [60, 1.1, 60], and WBC is [45, 4.8, 90] (WBC increases from 7.9 to 12.7). The ranking priority is then adjusted according to the indicator weights, with a priority score S = α * ΔV + β * Lt, where α = 0.6 and β = 0.4. The Temp score is 0.6 × 1.1 + 0.4 × 60 = 0.66 + 24 = 24.66, and the WBC score is 0.6 × 4.8 + 0.4 × 90 = 2.88 + 36 = 38.88. WBC has a higher priority than Temp. A new risk response structure is formed by embedding this structure into the reconstructed path: [WBC → Temp], replacing the original path [Temp → HR → WBC], thus generating the assessment level risk response structure.

[0033] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A big data based intelligent risk assessment system for gynecological surgery, characterized in that, The system includes: The cycle positioning module acquires the patient's menstrual cycle start time and hormone test data, sorts the cycle time axis, constructs cycle segment groups based on the start time point, marks the cycle segment groups in the preoperative examination data record items and performs collection processing to generate a cycle synchronization physiological feature set. The dominant indicator identification module, based on the set of periodic synchronous physiological features, calls the estrogen values, progesterone trend trajectory and body temperature fluctuation frequency band within the period segment, classifies the change amplitude and trend direction of the three features within each period segment, sets the features that show a continuous upward trend and maintain the dominant change pattern in the target period segment as the stage dominant features, and associates them with the clinical event records before gynecological surgery to generate a period dominant indicator mapping table. The collaborative fluctuation detection module, based on the cycle-dominant index mapping table, calls the heart rate flow curve, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory, and body temperature change segments from the real-time monitoring data during gynecological surgery. It judges the trend consistency and synchronous response cycle between each pair of data, identifies the trend reversal turning point, and judges the rhythm difference between the signs before and after. If the collaborative state has not recovered and the reversal trend continues to extend for multiple consecutive time periods, the time period is marked as the sign collapse segment. The module summarizes the types of signs and corresponding time anchors within the collapse segment and generates an interactive table of lost signs during surgery. The risk trend linkage module, based on the intraoperative lost sign interaction table, calls the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment from the postoperative monitoring records of gynecological surgery. It pairs the disintegration time anchor point with the postoperative indicator trend trajectory, and groups the corresponding indicators with similar time and continuously upward trend into the linkage indicator group, generating a postoperative indicator risk linkage trajectory map.

2. The big data based intelligent gynecological surgery risk assessment system according to claim 1, wherein, The cycle-synchronized physiological feature set includes cycle segment labels, synchronized sign trajectories, and endocrine change identifiers. The cycle-dominant indicator mapping table includes dominant feature types, corresponding cycle segment numbers, and associated clinical event codes. The intraoperative disconnected sign interaction table includes sign type indexes, trend reversal anchors, and synergistic failure segments. The postoperative indicator risk linkage trajectory diagram includes linkage indicator group classifications, trend resonance time axes, and fluctuation persistence parameters.

3. The intelligent risk assessment system for gynecological surgery based on big data according to claim 1, characterized in that, The periodic positioning module includes: The menstrual cycle initiation identification submodule obtains the patient's menstrual cycle initiation time and patient hormone test data, extracts the sequence of continuous bleeding time points from the patient's menstrual records, determines the interval value of adjacent time points, filters time points that meet the cycle interruption conditions, calls the bleeding initiation records within the cycle segment, performs time axis annotation and sorting, and generates a list of cycle initiation times. The biochemical hormone time analysis submodule performs interval landing point judgment on the time point according to the cycle start time list, matches the corresponding cycle segment, and filters the detection points through the time distribution density of hormone data to obtain the hormone detection time mapping table. The cycle synchronization feature construction submodule calls the hormone detection time mapping table to construct the time series of hormone indicators for the cycle, calculates the aggregated value of hormone synchronization perturbation, performs position matching and gradient amplitude threshold judgment on the aggregated value of cycle hormone synchronization perturbation with the ovulation segment range, and establishes a set of cycle synchronization physiological features.

4. The intelligent risk assessment system for gynecological surgery based on big data according to claim 3, characterized in that, The dominant indicator identification module includes: The feature trend classification submodule, based on the periodic synchronous physiological feature set, calls the estrogen values, progesterone trend trajectory, and body temperature fluctuation frequency band within the period segment to sequentially obtain estrogen time series samples, progesterone segment samples, and body temperature spectrum samples. According to the division of the period segment, the three samples are segmented trend extracted, the interaction trend response value is calculated, and interval clustering is performed based on the range and polarity direction of the interaction trend response value to obtain physiological feature trend classification label groups. The phase-dominant extraction submodule classifies the physiological feature trend into a label group, calls the segment label classified as continuously rising, filters estrogen, progesterone and body temperature feature segments that are in an upward state for two or more consecutive segments, and determines whether the cumulative value of persistence and amplitude is the peak value of the real-time period segment. If the condition of continuous amplitude superposition is met, the corresponding feature is set as the dominant feature of the period segment, and the period-dominant physiological feature sequence is obtained. The clinical event mapping submodule calls the cycle-dominant physiological feature sequence, compares it with the preoperative clinical event records of gynecological surgery within the corresponding cycle segment, matches event feature items within a similar time range by comparing the event occurrence time with the time period of the dominant feature, records the corresponding cycle identifier and dominant feature label, and obtains the cycle-dominant indicator mapping table.

5. The intelligent risk assessment system for gynecological surgery based on big data according to claim 4, characterized in that, The collaborative fluctuation detection module includes: The data stream receiving submodule calls the periodic dominant index mapping table to sequentially collect heart rate flow curves, blood pressure fluctuation frequency, transcutaneous blood oxygen trajectory and body temperature change segments from the real-time monitoring data during gynecological surgery. The original data is synchronized and aligned according to the corresponding time anchor points. By recalibrating the time series data according to a unified sampling period, a structured vital signs time series is generated. The rhythm consistency determination submodule extracts the fluctuation amplitude, cycle length and phase difference between any two pairs of vital signs data within adjacent windows based on the structured vital sign time series. According to the set rhythm phase threshold and amplitude difference critical coefficient, it performs sliding window determination on the time series data of multiple pairs of vital signs, records whether the trend direction is consistent and whether the synchronization response cycle is within the rhythm matching interval, and generates a vital sign rhythm consistency identifier sequence. The trend reversal identification submodule calls the vital sign rhythm consistency identifier sequence to locate the positions where trend consistency is missing and rhythm coordination is ineffective in multiple consecutive windows, continuously tracks the continuous extension of the reverse trend, marks the vital sign type and corresponding time anchor point within the time interval, and obtains the intraoperative lost vital sign interaction table.

6. The intelligent risk assessment system for gynecological surgery based on big data according to claim 5, characterized in that, The risk trend linkage module includes: The postoperative indicator extraction submodule calls the sign time anchor point in the intraoperative lost sign interaction table to obtain the postoperative fever curve, postoperative white blood cell response trajectory, and postoperative heart rate recovery segment in the postoperative monitoring record of gynecological surgery. It performs trend curve extraction and time series standardization processing on the three types of postoperative indicators respectively, unifies multiple data to an aligned time axis based on the postoperative start time, and generates a standardized postoperative indicator trend sequence. The time anchor matching submodule constructs a fixed time window for the time anchor based on the standardized postoperative indicator trend sequence. Within the window, it calculates the trend slope of three types of data: postoperative fever, white blood cell count, and heart rate. Indicators with a slope that is continuously greater than zero and lasts for more than half of the time window are marked with a trend. The time difference between the trend mark and the time anchor is used as a matching factor to obtain the postoperative indicator linkage node set. The linkage trajectory submodule calls the postoperative indicator linkage node set. For each set of successfully matched indicator data, it constructs the linkage trend trajectory of three types of indicators, namely postoperative fever, white blood cell count, and heart rate, according to the order of intraoperative time anchor points. The time series in the trajectory is processed by multidimensional merging, and the indicator type, trend direction, and time position in the linkage segment are integrated and arranged in sequence to form a risk trend evolution channel, generating a postoperative indicator risk linkage trajectory map.

7. The intelligent risk assessment system for gynecological surgery based on big data according to claim 1, characterized in that, The system also includes a hierarchical path adjustment module: The risk assessment path adjustment module calls the core indicator set in the real-time risk assessment path based on the postoperative indicator risk linkage trajectory map, extracts the vital signs in the path and matches them with the vital signs in the linkage trajectory, and determines whether the trend stabilization indicators in the linkage trajectory include the core indicators in the assessment path. If the match is incomplete, the assessment path structure is adjusted, the assessment priority of the matching items is optimized, and the risk response structure of the assessment level is generated. The risk response structure for the assessment level includes matching body collection, priority assessment path, and adjustment level nodes.

8. The intelligent risk assessment system for gynecological surgery based on big data according to claim 7, characterized in that, The hierarchical path adjustment module includes: The core indicator matching submodule calls the postoperative indicator risk linkage trajectory map, extracts the name of the vital signs, indicator weight and corresponding risk range of the path under the core indicator set in the real-time risk assessment level path, matches according to the indicator name, judges the trend stability of indicators whose trend direction remains unchanged in the linkage trajectory map, filters the vital signs whose trend duration exceeds the average assessment cycle, classifies the linkage indicators according to whether they are in the core indicator set, and obtains the trend stable core indicator matching table. The assessment structure reconstruction submodule, based on the trend stability core indicator matching table, adjusts the path structure for those with incomplete matching, rearranges the priority order of the matched indicators in the core indicator set, extracts the time position, trend length and amplitude fluctuation of the indicators in the postoperative indicator risk linkage trajectory diagram, constructs the indicator key ranking vector, embeds it into the original risk assessment level path structure, replaces the original priority arrangement, and generates the assessment level risk response structure.

9. The intelligent risk assessment system for gynecological surgery based on big data according to claim 8, characterized in that, The amplitude fluctuation is the difference between the peak and trough values ​​within a continuous trend segment in the postoperative indicator risk linkage trajectory diagram. The trend length is the length of the time period in which the continuous trend direction remains unchanged in the postoperative indicator risk linkage trajectory diagram, and only when the trend length is greater than the set threshold for the duration of the trough trend, the corresponding indicator in the indicator key ranking vector is given a positive weight. The threshold for the duration of the trough trend is the weighted average of the average trend duration of the core indicator's concentrated vital signs. The priority order is rearranged according to the weight values ​​of the indicators in the indicator criticality ranking vector from largest to smallest. In the ranking result, indicators with equal weights are arranged according to the order of their first appearance in the postoperative indicator risk linkage trajectory diagram.

Images