An obstetrics and gynecology diagnosis and treatment assistance system and method

Abstract

The present invention relates to the field of diagnostic technology, specifically to an obstetrics and gynecology diagnosis and treatment auxiliary system and method, the system includes: a parameter acquisition module, a feature offset recognition module, a feature reconstruction module, a collaborative trend recognition module, and a diagnosis and treatment suggestion generation module. In the present invention, by synchronously collecting body temperature, heart rate, pulse, blood pressure and image edge density and binding timestamps, a structured diagnosis and treatment data unit is formed, which effectively improves the time series comparison capability of multi-source information. Based on the joint judgment mechanism of numerical offset recognition and synchronization indicators, it is possible to locate potential risk points and construct feature labels, normalize and fuse physical signs, images, and laboratory data into three-dimensional vectors, and realize cross-dimensional feature linkage expression. By screening features with large fluctuations and long durations and judging trend consistency, the dynamic insight into changes in pathological states is improved, and combined with specific trend combination matching rules, the precise response to fetal risk states is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of diagnostic technologies, and particularly to an obstetrics and gynecology diagnosis and treatment assistance system and method. Background Art

[0002] The field of diagnostic technologies involves the process of identifying and determining the physiological state, biochemical indicators, or functional characteristics of the human body, and is widely used in the early detection of diseases, clinical evaluation, and health management. The core content of this technical field includes obtaining human-related information through means such as medical images, physical sign collection, bioelectrical signals, and physiological parameter measurement, and performing classification, comparison, and diagnostic judgment based on medical standards. The technical system covered by this field mainly consists of information collection, signal recognition, data analysis, and auxiliary judgment, and is widely used in the detection and auxiliary diagnosis scenarios of diseases in multiple systems such as cardiovascular, respiratory, nervous, endocrine, and reproductive systems.

[0003] Among them, the obstetrics and gynecology diagnosis and treatment assistance system refers to a comprehensive assistance plan designed for medical diagnosis activities related to the pregnancy and childbirth stages of women, involving multiple aspects such as pregnancy status monitoring, fetal health assessment, and childbirth risk warning, mainly through methods such as fetal heart signal collection and analysis, uterine contraction activity measurement, and reproductive tract physiological parameter monitoring. In clinical applications, this system often combines electronic fetal monitoring devices to obtain fetal heart rate and uterine contraction intensity data, and makes an auxiliary judgment on the status of fetal distress or the progress of labor by setting parameter thresholds or calculating model results, and is supplemented by reproductive health history data analysis and dynamic recording mechanisms to improve the pertinence and systematicness of diagnosis and treatment.

[0004] Existing technologies have limitations in data structure integration and trend analysis. The collected information is mostly single-category static indicators, lacking a linkage recognition mechanism between multi-dimensional features, and it is difficult to reflect the complex pathological evolution process. Without time-series binding and collaborative analysis conditions, the identification of abnormal nodes relies on a single threshold judgment, which is prone to inaccurate risk warning problems. For example, although the fetal heart rate does not exceed the boundary but changes with other indicators, the system is difficult to detect potential abnormalities, which is prone to judgment delay and intervention lag, affecting the efficiency of clinical treatment. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose an obstetrics and gynecology diagnosis and treatment assistance system and method.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: An obstetrics and gynecology diagnosis and treatment assistance system includes:

[0007] A parameter acquisition module obtains the body temperature, heart rate, pulse, and blood pressure of a patient during obstetric monitoring, collects and records the time stamp, extracts the edge density of the synchronous two-dimensional ultrasound image area, binds it with the physiological data and sets a dynamic identifier, and generates a set of bound diagnosis and treatment data units;

[0008] The feature offset recognition module calls the bound diagnosis and treatment data unit set to collect the fetal heart rate value and compares it with the fetal heart rate variation threshold. If it exceeds the threshold, it is marked as a warning node. At the same time, the corresponding pulse, blood pressure, and image edge density are extracted to judge the synchronous change, and an obstetric risk warning feature label group is generated.

[0009] Based on the fetal heart rate, edge density value, and estradiol value in the obstetric risk warning feature label group, the feature reconstruction module classifies them into the physical sign class, imaging class, and laboratory class feature groups respectively, and performs normalization processing according to the within-group acquisition standard deviation. If the three types of features appear simultaneously in the warning node, they are integrated into the form of a three-dimensional vector to generate an input body for feature fusion in obstetrics and gynecology.

[0010] The collaborative trend recognition module calls the three indicators in the input body for feature fusion in obstetrics and gynecology, screens out the one with the largest change amplitude and the longest duration, judges the direction consistency and the time window overlap ratio, and generates a multi-dimensional feature collaborative trend label.

[0011] As a further solution of the present invention, the bound diagnosis and treatment data unit set includes a dynamic identifier, physiological data image synchronization binding information, and a timestamp marker. The obstetric risk warning feature label group is specifically an abnormal recognition marker, a synchronous change parameter group, and a risk feature identifier. The input body for feature fusion in obstetrics and gynecology includes a three-dimensional vector of the physical sign class, a three-dimensional vector of the imaging class, and a three-dimensional vector of the laboratory class. The multi-dimensional feature collaborative trend label is specifically a change amplitude index, a duration parameter, a direction consistency marker, and a time window overlap ratio.

[0012] As a further solution of the present invention, the parameter acquisition module includes:

[0013] The physiological data acquisition sub-module collects the patient's body temperature, heart rate, pulse, and blood pressure during obstetric care, records the timestamp of each data item, classifies and stores them, calculates the data change rate and fluctuation amplitude, obtains the change intervals of multiple data items, and generates a physiological parameter change interval.

[0014] The image region extraction sub-module calls the synchronous ultrasound image, extracts the edge information of the fetal region, calculates the image gray-scale edge density, screens out the regions that meet the reference density, and obtains the synchronous edge density value.

[0015] The diagnosis and treatment unit generation sub-module calls the physiological parameter change interval and the synchronous edge density value, calculates the relationship between the physiological fluctuation and the image edge, extracts the bound data, and uses the formula:

[0016] ;

[0017] Performs arithmetic operations to obtain the interactive mapping difference degree value and generates a bound diagnosis and treatment data unit set.

[0018] Among them, represents the A set of physiological parameter fluctuation values, representing the group of image edge densities, representing the group of image edge gray - level base differences, representing the group of actual boundary fitting values of the image area, representing the number of timestamp groups participating in the matching, and M is the interactive mapping difference value.

[0019] As a further solution of the present invention, the feature offset recognition module includes:

[0020] The fetal heart rate warning recognition sub - module calls the fetal heart rate values concentrated in the bound diagnosis and treatment data unit, obtains the fetal heart rate fluctuation data within each monitoring cycle, and at the same time calculates the difference between the fluctuation range and the threshold based on the fetal heart rate variability threshold, marks the cycles exceeding the threshold, and obtains the warning mark node values;

[0021] The synchronous change feature extraction sub - module extracts the pulse, blood pressure and image edge density data at the corresponding time nodes according to the warning mark node values, calculates the change rate and the average change amount, and determines whether there is synchronous fluctuation to obtain the synchronous fluctuation analysis value;

[0022] The obstetric risk label generation sub - module is based on the synchronous fluctuation analysis value and uses the formula:

[0023] ;

[0024] Performs operations to obtain the composite fluctuation index of the change intensity of the fetal heart rate and the pulse under the influence of the image features, and combines the consistency of the synchronous change direction to generate an obstetric risk warning feature label group;

[0025] Among them, represents the change amplitude of the fetal heart rate, represents the change amplitude of the pulse, represents the group of blood pressure values, represents the group of image edge density values, represents the group of image edge density change values, is the reference value, is the number of time nodes, is the composite fluctuation index.

[0026] As a further solution of the present invention, the feature reconstruction module includes:

[0027] The sign classification sub-module obtains a sequence of continuous measurement values at multiple time nodes based on the fetal heart rate data in the obstetric risk warning feature label group, calculates the standard deviation and performs within-group standard deviation normalization, classifies it into the sign feature group, and generates the sign normalization value;

[0028] The image classification sub-module obtains the pixel distribution area of the edge density value within the image frame, extracts the maximum value and the average value, calculates its standard deviation and performs within-group standard deviation normalization, classifies it into the image feature group, and obtains the image normalization value;

[0029] The multi-source feature fusion sub-module calls the sign normalization value and the image normalization value, obtains the measurement data of the estradiol value, calculates the coefficient of variation and eliminates the abnormal points, using the formula:

[0030] ;

[0031] Integrate the three types of normalization values, determine whether they co-occur at the same warning node. If they co-occur, combine them in the form of a three-dimensional vector to generate the feature fusion input body;

[0032] Among them, represents the coefficient of variation after the fusion of the three types of features, represents the coefficient of variation of the sign normalization value, represents the mean value of the sign time series, represents the skewness of the sign time series, represents the kurtosis of the sign time series, represents the coefficient of variation of the image normalization value, represents the coefficient of variation of the estradiol normalization value, represents the absolute value operation to solve the difference between the coefficient of variation of the estradiol value and the sign mean value.

[0033] As a further solution of the present invention, the co-trend recognition module includes:

[0034] The amplitude screening sub-module calculates the sequence of change values of multiple indicators within the time window based on the three indicators in the obstetrics and gynecology feature fusion input body, compares the amplitudes of the maximum fluctuation intervals, screens the one with the largest amplitude value, and generates the maximum amplitude trend value;

[0035] The time correlation sub-module calls the change time window corresponding to the maximum amplitude trend value index, calculates the overlap ratio with the other two indicators in the obstetrics and gynecology feature fusion input body, screens the time period with a high overlap ratio, and generates the time window overlap interval value;

[0036] The trend determination submodule determines the direction consistency within the time window based on the change direction of the indicators contained in the time window overlap interval value. If the directions are consistent, it is recorded as positive; if there is a conflict, it is recorded as negative. The trend scalar is calculated by combining the change amplitude and the overlap ratio, using the formula:

[0037] ;

[0038] By calculating the trend synergy degree value and performing threshold judgment, a multi-dimensional feature collaborative trend mark is obtained;

[0039] Among them, represents the multi-dimensional feature collaborative trend mark, represents the continuous change amplitude of the th indicator, represents the time window overlap ratio of the th indicator and the target indicator, represents the consistency matching value of the change direction of the th indicator and the remaining indicators in the obstetrics and gynecology feature fusion input body, is the number of indicators participating in the trend judgment.

[0040] As a further solution of the present invention, the system further includes:

[0041] The diagnosis and treatment advice generation module matches the condition items in the advice list according to the combination of increased fetal heart rate, expanded marginal density, and decreased estradiol in the multi-dimensional feature collaborative trend mark, triggers the fetal intrauterine distress monitoring condition, and generates obstetrics and gynecology diagnosis and treatment assistance advice items;

[0042] The obstetrics and gynecology diagnosis and treatment assistance advice items include monitoring advice types, intervention advice types, and early warning level identifiers.

[0043] As a further solution of the present invention, the diagnosis and treatment advice generation module includes:

[0044] The collaborative trend analysis submodule extracts the difference between the fetal heart rate baseline value and the preset safe baseline range based on the multi-dimensional feature collaborative trend mark, calculates the deviation amplitude between the growth gradient of the marginal density per unit time and the critical expansion rate, fits the linear decline slope of the estradiol concentration change rate, and performs weighted distribution and superposition operations on the difference, deviation amplitude, and slope to generate a collaborative trend response value;

[0045] The monitoring condition matching submodule calls the fetal heart rate safety upper limit threshold, the marginal density expansion critical threshold, and the estradiol fluctuation lower limit threshold, compares the fetal heart rate difference in the collaborative trend response value with the safety upper limit, compares the marginal density deviation amplitude with the critical rate, and compares the estradiol slope with the fluctuation lower limit. According to the joint trigger logic of the simultaneous overlimit of the three parameters, a monitoring trigger determination result is generated;

[0046] Based on the monitoring trigger determination result, the intervention recommendation formulation sub-module calls the standard operation parameters of fetal heart rate monitoring and the maternal oxygen supply classification index, divides the monitoring frequency level according to the difference in fetal heart rate, integrates the estradiol slope to adjust the oxygen supply flow threshold, and defines the cesarean section priority parameter in combination with the deviation amplitude of the edge density, and generates the diagnosis and treatment recommendation items for fetal distress in utero.

[0047] An obstetrics and gynecology diagnosis and treatment assistance method, which is based on the above-mentioned obstetrics and gynecology diagnosis and treatment assistance system and includes the following steps:

[0048] S1: Obtain the physiological data including the patient's body temperature, heart rate, pulse, and blood pressure in obstetric monitoring, record the time stamp, extract the regional edge density of the synchronized two-dimensional ultrasound image, bind it with the physiological data and dynamically mark it, and generate a set of bound diagnosis and treatment data units;

[0049] S2: Compare the fetal heart rate data in the bound diagnosis and treatment data unit set with the fetal heart rate variability threshold. If it exceeds the threshold, mark it as a warning node, extract the synchronized pulse, blood pressure, and image edge density, and generate an obstetric risk warning feature label group;

[0050] S3: According to the fetal heart rate, edge density, and estradiol value in the obstetric risk warning feature label group, classify them into the physical sign class, imaging class, and laboratory class feature groups, perform standard deviation normalization. If the three types of features appear simultaneously at the warning node, integrate them into a three-dimensional vector to generate an obstetrics and gynecology feature fusion input body;

[0051] S4: Call the three indicators in the obstetrics and gynecology feature fusion input body, screen the indicator with the largest change amplitude and the longest duration, judge the direction consistency and the time window overlap ratio, and generate a multi-dimensional feature collaborative trend label;

[0052] S5: According to the combination of fetal heart rate increase, edge density expansion, and estradiol decrease in the multi-dimensional feature collaborative trend label, match the conditional items, trigger the fetal distress in utero monitoring condition, and generate the obstetrics and gynecology diagnosis and treatment assistance recommendation items.

[0053] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0054] In the present invention, by synchronously collecting the body temperature, heart rate, pulse, blood pressure, and image edge density and binding the time stamp, a structured diagnosis and treatment data unit is formed, effectively improving the time series comparison ability of multi-source information. Based on the numerical deviation recognition and synchronous index joint judgment mechanism, potential risk points can be located and feature labels can be constructed. Normalize and fuse the physical sign, imaging, and laboratory data into a three-dimensional vector to achieve cross-dimensional feature linkage expression. By screening features with large change amplitudes and long durations and judging the trend consistency, the dynamic insight ability into pathological state changes is improved. Combining specific trend combination matching rules enhances the accurate response to the fetal risk state. Brief Description of the Drawings

[0055] Figure 1 is the system flow chart of the present invention;

[0056] Figure 2 is the flow chart of the parameter acquisition module of the present invention;

[0057] Figure 3 is the flow chart of the feature offset recognition module of the present invention;

[0058] Figure 4 is the flow chart of the feature reconstruction module of the present invention;

[0059] Figure 5 is the flow chart of the collaborative trend recognition module of the present invention;

[0060] Figure 6 is the flow chart of the diagnosis and treatment advice generation module of the present invention. Detailed Description of the Preferred Embodiments

[0061] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present 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 only used to explain the present invention and are not used to limit the present invention.

[0062] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more unless otherwise specifically defined.

[0063] Example 1: Please refer to Figure 1 , the present invention provides a technical solution: An obstetrics and gynecology diagnosis and treatment assistance system includes:

[0064] The parameter acquisition module obtains the body temperature, heart rate, pulse and blood pressure of the patient in obstetric care, collects and records the time stamp, extracts the edge density of the synchronous two-dimensional ultrasound image area, binds it with the physiological data and sets a dynamic identifier, and generates a bound diagnosis and treatment data unit set;

[0065] The feature offset recognition module calls the bound diagnosis and treatment data unit set to centralize the fetal heart rate value and compare it with the fetal heart rate variability threshold. If it exceeds the threshold, it is marked as a warning node. At the same time, the corresponding pulse, blood pressure, and image edge density are extracted to judge synchronous changes, and an obstetric risk warning feature label group is generated.

[0066] Based on the fetal heart rate, edge density value, and estradiol value in the obstetric risk warning feature label group, the feature reconstruction module classifies them into the physical sign class, imaging class, and laboratory class feature groups respectively, and performs normalization processing according to the within-group acquisition standard deviation. If the three types of features appear simultaneously in the warning node, they are integrated into the form of a three-dimensional vector to generate an input body for obstetrics and gynecology feature fusion.

[0067] The collaborative trend recognition module calls the three indicators in the input body for obstetrics and gynecology feature fusion, screens the one with the largest change amplitude and the longest duration, judges the direction consistency and the time window overlap ratio, and generates a multi-dimensional feature collaborative trend label.

[0068] According to the combination of increased fetal heart rate, expanded edge density, and decreased estradiol in the multi-dimensional feature collaborative trend label, the diagnosis and treatment recommendation generation module matches the condition items in the recommendation list, triggers the monitoring conditions for fetal distress in utero, and generates obstetrics and gynecology diagnosis and treatment assistance recommendation items.

[0069] The bound diagnosis and treatment data unit set includes dynamic identification, physiological data image synchronous binding information, and timestamp marking. The obstetric risk warning feature label group specifically includes abnormal identification marks, synchronous change parameter groups, and risk feature identifications. The input body for obstetrics and gynecology feature fusion includes three-dimensional vectors of the physical sign class, three-dimensional vectors of the imaging class, and three-dimensional vectors of the laboratory class. The multi-dimensional feature collaborative trend label specifically includes change amplitude indicators, duration parameters, direction consistency marks, and time window overlap ratios. The obstetrics and gynecology diagnosis and treatment assistance recommendation items include monitoring recommendation types, intervention recommendation types, and warning level identifications.

[0070] Please refer to Figure 2 , the parameter acquisition module includes:

[0071] The physiological data acquisition sub-module acquires the patient's body temperature, heart rate, pulse, and blood pressure during obstetric monitoring, records the timestamp of each data item, classifies and stores them, calculates the data change rate and fluctuation amplitude, obtains the change intervals of multiple data items, and generates physiological parameter change intervals.

[0072] First, the physiological data acquisition sub-module uses sensors connected to specific parts of the pregnant woman's body to monitor and collect key physiological index data in real time. For example, an electronic thermometer is used to measure the armpit temperature, and the body temperature value is obtained as 37.1 °C. A cardiac monitor is used to record the heart rate, and the current value is 85 beats per minute. A pulse oximeter is used to measure the fingertip pulse, and the current value is 88 beats per minute. An electronic sphygmomanometer is used to measure the upper arm blood pressure, and the systolic blood pressure is recorded as 125 mmHg and the diastolic blood pressure as 75 mmHg. At the same time, the system precisely timestamps each piece of collected data.

[0073] Table 1 Example table of physiological data acquisition:

[0074] ;

[0075] As shown in Table 1, some physiological data continuously collected within a short period and their timestamps are presented. Then, these timestamped data are stored in corresponding databases or memory areas according to the index types (body temperature, heart rate, pulse, blood pressure). Then, the module retrieves continuous data points stored in the database over a past period (e.g., from 10:00:05 to 10:02:07), and calculates the change rate of each piece of data. For example, when calculating the heart rate change rate, the heart rate value of 87 beats per minute at 10:01:06 and the heart rate value of 85 beats per minute at 10:00:06 are extracted, the difference is calculated to obtain a change amount of 2 beats per minute, and then divided by the time interval of 1 minute to get a change rate of 2 beats per minute². Similarly, the change rates of other indicators are calculated, and the fluctuation range is calculated. For example, the maximum value (87 beats per minute) and the minimum value (85 beats per minute) of the heart rate within this time period are found, and the difference is calculated to obtain a fluctuation range of 2 beats per minute. Based on these calculation results, the change intervals of each physiological data within the monitoring period are determined. For example, the body temperature change interval is [37.0 °C, 37.2 °C], the heart rate change interval is [85 beats per minute, 87 beats per minute], the pulse change interval is [88 beats per minute, 90 beats per minute], the systolic blood pressure change interval is [125 mmHg, 128 mmHg], and the diastolic blood pressure change interval is [74 mmHg, 76 mmHg]. Finally, these intervals are integrated to generate the physiological parameter change interval.

[0076] The image region extraction sub-module calls the synchronous ultrasound image, extracts the edge information of the fetal region, calculates the grayscale edge density of the image, filters the regions that meet the benchmark density, and obtains the synchronous edge density value;

[0077] Call the ultrasound image frame obtained synchronously with the physiological data acquisition time point (such as 10:00:06), preprocess the image, such as grayscale conversion and median filtering for denoising, and then process the image using the Canny operator. This operator extracts all potential edge pixel point information in the image by calculating the image gradient, non-maximum suppression, and double-threshold connection to obtain a preliminary edge map. Then, calculate each pixel point in the image The gradient magnitude And direction , quantify the edge intensity according to the magnitude of the gradient value, and further calculate the gray edge density of a specific region (such as the position estimated by fetal biological parameters or the region of interest ROI manually framed). This density can be defined as the ratio of the number of pixels with a gradient magnitude greater than a certain threshold (such as ), for example, calculate the preliminary edge density of the fetal region to be 0.15, set a reference density threshold, which is obtained through statistical analysis of a large number of normal ultrasound images. For example, analyze 1000 normal fetal ultrasound images of the same gestational age, calculate the edge density of their fetal regions, obtain an average value of 0.11 and a standard deviation of 0.03, and set the reference threshold as the average value minus one standard deviation, that is , or set a fixed value according to clinical needs, such as setting it to 0.10, to distinguish between fuzzy regions and clear regions, and select the regions in the image with an edge density greater than or equal to 0.10 as valid regions. For example, if the previously calculated 0.15 is greater than 0.10, then this region is selected. Finally, obtain the edge density values of these selected regions at the synchronous time point as synchronous edge density values. For example, record the synchronous edge density value as 0.15.

[0078] The diagnosis and treatment unit generation sub-module calls the physiological parameter change interval and the synchronous edge density value, calculates the relationship between physiological fluctuations and image edges, extracts the bound data, and uses the formula:

[0079] ;

[0080] Calculate the interaction mapping difference value through operations and generate a set of bound diagnosis and treatment data units;

[0081] Among them, Represents the Group of physiological parameter fluctuation values, Represents the Group of image edge densities, Represents the Group of image edge gray base differences, Represents the Group of actual boundary fitting values of the image region, represents the number of timestamp groups participating in the matching, and M is the value of the interactive mapping difference degree.

[0082] Call the physiological parameter change interval, such as the heart rate change interval [85, 87] beats per minute, and calculate its fluctuation value, such as taking the interval amplitude , and call the synchronous edge density value obtained at the same time point , meanwhile, it is also necessary to obtain the image edge gray level base difference and the fitting value of the actual boundary of the image area , represents the difference between the average gray value of the selected edge pixels and the average gray value of the adjacent non-edge background area. For example, if the average edge gray value is calculated to be 150 and the average background gray value is 125, then , By performing shape matching on the extracted edge with the standard fetal anatomical boundary template (such as using the Hausdorff distance or Dice coefficient), and quantifying the matching degree. For example, if the matching degree is 0.8, then .

[0083] Suppose we select the data of the last 3 time stamps for matching, that is , and the data of each time stamp is as follows:

[0084] Time stamp 1 (q = 1): , , , ;

[0085] Time stamp 2 (q = 2): , , , (assuming the heart rate fluctuation in the next minute is 3, and the edge density etc. also change);

[0086] Time stamp 3 (q = 3): , , , (data in the next minute) Substitute these data into the formula:

[0087] ;

[0088] Perform the calculation, detailed calculation process:

[0089] Molecular calculation:

[0090] Item 1 (q = 1): ;

[0091] Item 2 (q = 2): ;

[0092] Item 3 (q = 3): ;

[0093] Sum of molecules = ;

[0094] Denominator calculation:

[0095] Item 1 (q = 1): ;

[0096] Item 2 (q = 2): ;

[0097] Item 3 (q = 3): ;

[0098] Denominator = ;

[0099] Calculate the M value: ;

[0100] The advantage of the formula is that by combining the volatility of physiological parameters ( ) with image features (edge density , gray level base difference , boundary fitting degree ), and considering their internal relationships and the cumulative effect in the time series, the degree of interactive mapping difference between multi-source information is quantified. The calculated value of the interactive mapping difference will be extracted and bound together with the original data such as to form a structured data unit containing timestamps, physiological fluctuations, image features and their correlation metrics, generating a set of bound diagnosis and treatment data units. This result indicates that there is a certain correlation between physiological fluctuations and image features during this time period. The calculated is a quantification index for subsequent module analysis.

[0101] Please refer to Figure 3 , the feature offset recognition module includes:

[0102] The fetal heart rate early warning recognition sub-module calls the fetal heart rate values collected in the bound diagnosis and treatment data unit, obtains the fetal heart rate fluctuation data within each monitoring period, and at the same time calculates the difference between the fluctuation range and the threshold based on the fetal heart rate variability threshold, marks the periods exceeding the threshold, and obtains the early warning marker node values;

[0103] First, the fetal heart rate warning recognition sub-module extracts the concentrated fetal heart rate values (Fetal Heart Rate, FHR) within a specific time period (e.g., the past 15 minutes). Suppose the extracted FHR sequence (unit: beats per minute) is: [135, 138, 140, 142, 139, 145, 148, 150, 152, 155, 153, 150, 147, 144, 140].

[0104] Table 2 Example of fetal heart rate (FHR) sequence and difference representation:

[0105] ;

[0106] As shown in Table 2, the FHR sequence, its adjacent differences, periodic fluctuations, and sums are listed, and it is marked whether the warning threshold is triggered. The fetal heart rate fluctuation data within each monitoring period (e.g., recorded once per minute) is obtained, that is, the difference sequence of FHR at adjacent time points: [3, 2, 2, -3, 6, 3, 2, 2, 3, -2, -3, -3, -3, -4]. At the same time, a fetal heart rate variability threshold needs to be set. This threshold is set with reference to the guidelines of the American College of Obstetricians and Gynecologists (ACOG) or based on statistical analysis of a large amount of historical data. For example, the normal short-term variability (STV) range is set to 5 - 25 beats per minute. Here, we focus on single-beat changes and short-term cumulative changes, and set a periodic variability threshold. For example, it is set that the sum of the absolute values of two consecutive FHR differences is greater than 10 beats per minute or the absolute value of a single change is greater than 8 beats per minute as the abnormal fluctuation threshold. Now, calculate the difference between the fluctuation range of each monitoring period (taking two consecutive FHR differences as a cycle unit, as shown in Table 2) and the threshold. For example, in cycle 1 (differences 3, 2), the sum of fluctuations is 5, less than 10, and the maximum absolute value of a single difference is 3, less than 8, not exceeding the threshold. Suppose the data shown in Table 2 at T6’ and T7’ appears in subsequent monitoring: the FHR sequence is […, 130, 145, 132, …], and the difference sequence is […, 15, -13]. At time point T6’, the absolute value of a single change is 15, greater than the threshold 8, and mark the time point (the moment when 145 appears) as the warning mark node. At time point T7’, the absolute value of a single change is 13, greater than the threshold 8, and the sum with the previous difference is |15| + |-13| = 28, greater than the threshold 10. Similarly, mark the time point (the moment when 132 appears) as the warning mark node. Suppose we identify that two nodes, time stamps T6’ and T7’, exceed the threshold, and obtain the warning mark node values T6’, T7’.

[0107] The synchronous change feature extraction sub-module extracts the pulse, blood pressure, and image edge density data at the corresponding time nodes according to the warning mark node values, calculates the change rate and the mean change amount, determines whether there is synchronous fluctuation, and obtains the synchronous fluctuation analysis value;

[0108] According to the identified warning mark node values T6’ and T7’, extract the corresponding maternal pulse values (for example, 95 beats per minute at T6’ and 92 beats per minute at T7’), blood pressure values (for example, systolic blood pressure of 135 mmHg at T6’ and 130 mmHg at T7’), and the previously calculated synchronous edge density values (for example, 0.19 at T6’ and 0.16 at T7’) from the bound diagnosis and treatment data unit set. Then calculate the change rates of these parameters near the warning nodes. For example, calculate that within a short time (such as 1 minute) before and after T6’, the pulse changes from 90 beats per minute to 95 beats per minute, with a rate of 5 beats per minute², the systolic blood pressure changes from 130 mmHg to 135 mmHg, with a rate of 5 mmHg per minute, and the edge density changes from 0.17 to 0.19, with a rate of 0.02 per minute. Also calculate the mean change amount of these parameters within a time window (such as 5 minutes) containing the warning node. For example, the mean pulse value changes from 88 beats per minute to 92 beats per minute within 5 minutes before and after the T6’ node, with a mean change amount of 4 beats per minute. By comparing whether the change directions of the fetal heart rate at the warning node (for example, the FHR increases by 15 at T6’) are consistent with those of other physiological parameters (the pulse increases by 5, the blood pressure increases by 5) and image features (the edge density increases by 0.02), determine whether there is synchronous fluctuation. For example, at the T6’ node, if the FHR, pulse, blood pressure, and edge density all show an upward trend, it is judged as synchronous fluctuation. Quantify this analysis result. For example, synchronous fluctuation is recorded as 1, asynchronous is recorded as 0, or a value between -1 and 1 is assigned according to the correlation coefficient. Assume that the synchronous fluctuation analysis value at the T6’ node is 0.9, and at the T7’ node, the FHR decreases by 13, the pulse decreases by 3, the blood pressure decreases by 5, and the edge density decreases by 0.03, which is also judged as synchronous fluctuation (synchronous decrease), and the synchronous fluctuation analysis value is 0.8, obtaining the synchronous fluctuation analysis value sequence [0.9, 0.8].

[0109] The obstetric risk label generation sub-module is based on the synchronous fluctuation analysis value and uses the formula:

[0110] ;

[0111] Perform operations to obtain the change intensity of the fetal heart rate and pulse, the composite fluctuation index under the influence of image features, and combine the consistency of the synchronous change direction to generate an obstetric risk warning feature label group;

[0112] Among them, represents the change amplitude of the fetal heart rate, Represents the pulse change amplitude, Indicates the group of blood pressure values, Indicates the group of image edge density values, Indicates the group of image edge density change values, Is the reference value, Is the number of time nodes, Is the composite fluctuation index.

[0113] Based on these synchronous fluctuation analysis values, select the data of one of the nodes (such as the riskier T6' node) for calculation, using the formula:

[0114] ;

[0115] Parameter assignment and calculation: : The change amplitude of fetal heart rate, referring to the single FHR change amount at the warning node T6', . : The pulse change amplitude, referring to the single pulse change amount at the warning node T6', . : The number of time nodes, take 2 points before and after the T6' node, a total of time nodes, : The group of blood pressure values (systolic pressure).

[0116] Assume The systolic pressure sequence of the points is [130, 132, 135, 133, 130] mmHg. : The group of image edge density values. Assume The edge density sequence of the points is [0.17, 0.18, 0.19, 0.18, 0.16].[[]END] : The group of image edge density change values (adjacent point differences). The sequence is [0.01, 0.01, -0.01, -0.02], to make up points, the first value is set to 0, and the sequence is [0, 0.01, 0.01, -0.01, -0.02].[[]END] : The image edge density reference value, with the setting basis the same as before, .

[0117] Detailed calculation process (taking the T6' node as an example, k = 5):

[0118] Calculate the denominator : ;

[0119] Calculate the second term : ;

[0120] The second term = ;

[0121] Calculate : ;

[0122] The advantage of the formula is that by integrating the amplitudes of the fluctuations of the heart rate and pulse in the numerator and combining the blood pressure and the edge density of the image in the denominator, the second term quantifies the deviation of the change in the image features, comprehensively evaluates the combined effects of various factors, and the composite fluctuation index of the change intensity of the fetal heart rate and pulse obtained by the operation under the influence of the image features , combined with the synchronous fluctuation analysis value obtained from the previous calculation (the T6' node is 0.9, indicating high synchrony and strong consistency in the change direction), after comprehensive judgment by the system, an obstetric risk warning feature label is generated, such as "significant acceleration of FHR accompanied by synchronous increase in blood pressure, pulse, and edge density, medium composite fluctuation index", forming a group of obstetric risk warning feature labels. This result indicates that there is a certain composite fluctuation risk at present, and its numerical value can be compared with the preset risk level threshold (for example, low risk < 0.1, medium risk 0.1 - 0.2, high risk > 0.2), and it is initially judged as medium risk

[0123] Please refer to Figure 4 , the feature reconstruction module includes:

[0124] The sign classification sub-module obtains a sequence of continuous measurement values at multiple time nodes based on the fetal heart rate data in the obstetric risk warning feature label group, calculates the standard deviation and performs within-group standard deviation normalization processing, classifies it into the sign class feature group, and generates the normalized value of the sign class;

[0125] First, the sign classification sub-module extracts the fetal heart rate data related to the signs from it, especially the sequence of continuous fetal heart rate measurement values within the time period near the warning node and the warning node itself. For example, for the previously identified T6' warning node, extract the FHR sequences at 10 time points before and after it: [138, 140, 142, 139, 130, 145, 132, 135, 138, 140] (unit: beats per minute), and calculate the standard deviation of this sequence , mean , variance , standard deviation beats per minute, and then perform within-group standard deviation normalization processing on this standard deviation, using Z-score standardization. Assume that the mean of the standard deviations of a group of historical FHR sequences is known as times per minute, with a standard deviation of times per minute (these values are obtained based on the statistical analysis of a large amount of historical monitoring data), then the normalized value is , and this normalized value is classified into the physical sign feature group to generate the normalized physical sign value, denoted as .

[0126] The image classification sub-module obtains the pixel distribution area of the edge density value within the image frame, extracts the maximum value and the average value, calculates its standard deviation and normalizes it using the within-group standard deviation, and classifies it into the image feature group to obtain the normalized image value;

[0127] Retrieve the previously calculated edge density value in the ultrasonic image frame synchronized with the physical sign data , and analyze the pixel spatial distribution of these edge density values within the image frame, determine the area where the high-density values are concentrated, extract the maximum value (for example, Max_Ed = 0.20 near T6') and the average value (for example, Avg_Ed = 0.18) of the edge density within this area, and then calculate the standard deviation of the edge density values in this area , for example, calculate to obtain , and also perform within-group standard deviation normalization processing. Assume that the known mean of the historical edge density standard deviation is , with a standard deviation of (based on historical image data statistics), then the normalized image value is , and this value is classified into the image feature group to obtain the normalized image value, denoted as .

[0128] The multi-source feature fusion sub-module calls the normalized physical sign value and the normalized image value, obtains the measurement data of the estradiol value, calculates the coefficient of variation and eliminates the outliers, using the formula:

[0129] ;

[0130] Integrate the three types of normalized values, and judge whether they co-occur at the same warning node. If they do, combine them in the form of a three-dimensional vector to generate the feature fusion input body;

[0131] Among them, represents the coefficient of variation after the fusion of the three types of features, represents the coefficient of variation of the normalized physical sign value, represents the mean of the physical sign time series, represents the skewness of the physical sign time series, represents the kurtosis of the physical sign time series, represents the coefficient of variation of the normalized image value, represents the coefficient of variation of the normalized estradiol value, Represents the absolute value operation to solve for the difference between the coefficient of variation of estradiol values and the mean of physical signs.

[0132] Call the normalized values of physical signs obtained previously and the normalized values of imaging , and additionally obtain the measurement data of the estradiol (Estradiol, E2) level of the parturient within the same time period. Estradiol is an important endocrine index reflecting placental function.

[0133] Table 3 Example of estradiol (E2) measurement sequence:

[0134] ;

[0135] As shown in Table 3, assume the obtained E2 measurement value sequence, where the last point is adjusted to 4300 to reflect the downward trend, and calculate the coefficient of variation (Coefficient of Variation, CV) of this sequence, the mean pg / mL;

[0136] Variance ;

[0137] Standard deviation pg / mL, then ;

[0138] Before calculation, check whether the data points 4400 and 4300 are outliers. Assume that according to the historical data distribution, they are still within a reasonable range and do not need to be excluded. Then normalize the E2 data and calculate the Z-score. Assume that the mean of the historical E2 coefficient of variation is , and the standard deviation is (based on historical data statistics), then the coefficient of variation of the normalized estradiol value (replace with the normalized CV value ).

[0139] ;

[0140] Then use the formula:

[0141] ;

[0142] Perform integrated calculation, parameter assignment and calculation: : Represented by , . : The mean of the physical sign time series (FHR series), previously calculated . : Skewness of the vital sign time series (FHR series [138,140,142,139,130,145,132,135,138,140]). Calculated . : Kurtosis of the vital sign time series (FHR series). Calculated (excess kurtosis). : Represented by . . : Represented by . .

[0143] Detailed calculation process:

[0144] Calculate the terms in the numerator: ;

[0145] Numerator term = ;

[0146] Calculate the terms in the denominator: ;

[0147] Denominator = ;

[0148] Calculate : ;

[0149] The advantage of the formula is that it integrates information from three different sources, namely vital signs, images, and endocrine, through a comprehensive formula. In particular, it considers the statistical characteristics of vital sign data and the variability of multi-source features, structurally integrating the complex relationships of multi-source features. This module integrates three types of normalized values , and determines whether they simultaneously appear at the same warning node (such as T6'). If the T6' node simultaneously triggers abnormal vital signs ( deviation from the baseline), abnormal images ( deviation from the baseline), and specific changes in endocrine indicators ( variability reflected), then these three normalized values are combined into a three-dimensional vector form , serving as the feature fusion input body at this time node, generating a set of feature fusion input bodies. The calculated is a comprehensive variation index, and this result indicates that after considering various factors such as vital signs, images, and endocrine, this time point shows a certain degree of comprehensive variability, which is used for subsequent trend analysis.

[0150] Please refer to Figure 5 , and the collaborative trend recognition module includes:

[0151] The amplitude screening sub-module calculates the change value sequence of multiple indicators within the time window based on three indicators in the input body of obstetrics and gynecology characteristics, compares the amplitudes of the maximum fluctuation intervals, selects the one with the largest amplitude value, and generates the maximum amplitude trend value;

[0152] Receive the input body of feature fusion, that is, the three-dimensional vector sequence generated in the previous step,

[0153] Table 4 Example of the input body sequence of feature fusion:

[0154] ;

[0155] As shown in Table 4, assume that we have the input body of feature fusion at 5 consecutive time points , where the three components of each vector represent the normalized values of the physical sign category , the imaging category and the estradiol category . First, the amplitude screening sub-module calculates the change value sequence (difference between adjacent points) of each indicator within the set time window (from time point 1 to time point 5) based on these three indicators. The change sequence of is [0.08, 0.07, -0.05, -0.05], The change sequence of is [0.075, 0.075, -0.05, -0.05], The change sequence of is [-0.065, 0.015, 0.02, 0.02]. Then, compare the amplitudes of the maximum fluctuation intervals (difference between the maximum value and the minimum value) of each indicator within the entire time window. For , the maximum value is 0.45, the minimum value is 0.30, and the amplitude , for , the maximum value is 0.70, the minimum value is 0.55, and the amplitude , for , the maximum value is -0.65, the minimum value is -0.715, and the amplitude . In this example, the amplitudes of the physical sign category and the imaging category

[0156] are tied for the largest (both are 0.15). One or both of them can be selected as the target indicator. Here, the imaging category indicator

[0157] is selected as the target indicator, recorded, and the maximum amplitude trend value ), and calculate the overlap ratio between this time window and the time windows of significant changes in the other two indicators in the feature fusion input body ( and ). The time window of significant change can be determined by setting a threshold (for example, the absolute value of the change is greater than 0.06). For , the significant change is [0.08, 0.07] (points 1 - 3). For , the significant change is [-0.065] (points 1 - 2). The significant change in the target indicator is [0.075, 0.075] (points 1 - 3). Then and have a significant window overlap at points 1 - 3, and the overlap ratio is (relative to the length of their respective significant windows). and have a significant window overlap at points 1 - 2, and the overlap ratio is (relative to the length of their respective significant windows) or 2 / 3 (relative to the significant window length of ) when calculated according to the total window. Here, a simplified method is adopted. It is considered that the overlap ratio is high within the main change interval, and (assumed to be obtained according to a certain calculation rule) is taken. The time periods with a higher overlap ratio are screened out. For example, if the threshold is set to 0.5, then and are both selected to generate the time window overlap interval value.

[0158] The trend determination sub - module determines the direction consistency within the time window according to the change directions of the indicators contained in the time window overlap interval value. If the directions are consistent, it is recorded as positive, and if there is a conflict, it is recorded as negative. The trend scalar is calculated by combining the change amplitude and the overlap ratio, using the formula:

[0159] ;

[0160] By calculating the trend synergy degree value and performing threshold judgment, a multi - dimensional feature collaborative trend mark is obtained;

[0161] Among them, represents the multi - dimensional feature collaborative trend mark, represents the continuous change amplitude of the th indicator, represents the time window overlap ratio between the th indicator and the target indicator, represents the consistency matching value between the th indicator and the change directions of the other indicators in the obstetrics and gynecology feature fusion input body, is the number of indicators participating in the trend judgment.

[0162] Based on the overlapping interval values of the selected time windows (e.g., at points 1 - 3, at points 1 - 2), judge the change direction of the indicators ( ). Within the time points 1 - 3, changes to [0.075, 0.075] (positive), changes to [0.08, 0.07] (positive), with the same direction. , within the time points 1 - 2, changes to [0.075] (positive), changes to [-0.065] (negative), with the opposite direction. , combined with the change amplitudes of each indicator (using the total change amplitude within the overlapping window), , ;

[0163] and the overlapping ratio of the time window ( ), calculate the trend scalar, using the formula:

[0164] ;

[0165] Parameter assignment and calculation ( , the indicators are and ): , , , , ;

[0166] Calculate the numerator :

[0167] Item 1 ( ): ;

[0168] Item 2 ( ): ;

[0169] Sum of numerator = ;

[0170] Calculate the denominator :

[0171] Item 1 ( ): ;

[0172] Item 2 ( ): ;

[0173] Denominator = ;

[0174] Calculation : ;

[0175] The advantage of the formula is that it integrates the variation range and time overlap degree of each feature relative to the target feature through the numerator, and the denominator is adjusted using direction consistency, quantifying the co - movement trend between multi - dimensional features. The trend synergy degree value obtained through calculation , compares this degree value with a preset threshold. For example, if the synergy trend threshold is set to 0.1, then currently , indicating that the synergy trend is not significant, and the multi - dimensional feature synergy trend is marked as "weak synergy". This result indicates that although each indicator fluctuates and there is a certain time synchronization between some indicators, the overall trend strength of their co - variation, calculated according to the current formula and parameters, is at a relatively low level

[0176] Please refer to Figure 6 , the diagnosis and treatment advice generation module includes:

[0177] The synergy trend analysis sub - module, based on the multi - dimensional feature synergy trend mark, extracts the difference between the fetal heart rate baseline value and the preset safe baseline range, calculates the deviation amplitude between the growth gradient of the edge density per unit time and the critical expansion rate, fits the linear decline slope of the estradiol concentration change rate, and performs weighted assignment and superposition operation on the difference, deviation amplitude, and slope to generate a synergy trend response value;

[0178] Receiving the multi - dimensional feature synergy trend mark ("weak synergy", synergy degree value ), first, the synergy trend analysis sub - module, based on this mark, further extracts specific indicators directly related to clinical judgment for analysis. It extracts the baseline value of the previously calculated fetal heart rate (FHR) time series [138, 140, 142, 139, 130, 145, 132, 135, 138, 140]. The baseline value is calculated to be approximately 137.9 beats per minute by performing a 10 - point moving average (or other filtering methods) on this sequence, and obtains the preset safe baseline range of fetal heart rate (110 - 160 beats per minute), calculates the difference between the current baseline value and the boundary of the safe range, , , takes the difference from the nearest boundary, that is, 22.1 beats per minute. At the same time, calculates the growth gradient of the previously obtained imaging - type indicator (edge density ) per unit time. For example, in the rising stage (points 1 - 3), the average gradient is approximately / time point, and compare it with a preset critical expansion rate of edge density. The basis for setting this critical rate is, for example, that when the edge density grows too fast, it may indicate tissue abnormality. According to clinical statistics, the critical value is set to 0.05 / time point, and calculate the deviation amplitude between the current gradient and the critical rate, that is / time point. Then, fit the change rate of the previously obtained sequence of estradiol (E2) concentration values ([4500, 4650, 4800, 4700, 4950, 5100, 5200, 5050, 4400, 4300] in Table 3). Perform a linear fit on the last three points [5050, 4400, 4300], and the slope is approximately pg / mL / time point, indicating a downward trend. Assign weights to these three key analysis results: the difference in fetal heart rate baseline (22.1 beats per minute), the deviation amplitude of edge density (0.025 / time point), and the change slope of estradiol (-375 pg / mL / point). The basis for setting the weights is, for example, the weight of FHR baseline stability , the weight of edge density change , the weight of estradiol trend , These weights are determined based on the risk prediction ability of the indicators. First, the risk scores of different unit values need to be transformed. For example, the FHR difference of 22.1 is within the safe range, and the risk score is 0.1. The density deviation of 0.025 is lower than the critical value, and the risk score is 0.2. The E2 slope of -375 indicates a significant decrease, and the risk score is 0.7. Then the combined trend response value = , generating a combined trend response value of 0.31

[0179] The monitoring condition matching sub-module calls the upper safety limit threshold of fetal heart rate, the critical threshold of edge density expansion, and the lower threshold of estradiol fluctuation. Compare the fetal heart rate difference in the combined trend response value with the upper safety limit, the deviation amplitude of edge density with the critical rate, and the estradiol slope with the lower threshold of fluctuation. Generate a monitoring trigger determination result according to the combined trigger logic that all three parameters exceed the limit at the same time;

[0180] Call multiple preset thresholds, including the upper safety threshold of fetal heart rate (160 beats per minute), the lower safety threshold of fetal heart rate (110 beats per minute), the critical threshold for marginal density expansion (gradient > 0.05 per time point), the lower threshold for estradiol fluctuation (e.g., E2 concentration < 4000 pg / mL or the decline rate exceeds -500 pg / mL per point). Compare the original index values or risk scores decomposed from the calculated co - trend response value with these thresholds. The current FHR baseline value (137.9) is within the range of [110, 160] and is not exceeded. The marginal density growth gradient (0.075) exceeds the critical rate of 0.05, triggering Condition 1. The lowest value of estradiol, 4300, is not lower than the lower limit of 4000, but the slope - 375 does not exceed - 500, so Condition 2 is not triggered. According to the preset combined trigger logic, for example, "marginal density gradient exceeds the limit and (FHR baseline exceeds the limit or E2 level / slope exceeds the limit)" for judgment. In this case, the marginal density gradient exceeds the limit, but both the FHR baseline and E2 do not exceed the limit (according to this logic), so the combined trigger condition is not met, and the monitoring trigger determination result is "Some indicators are abnormal, and the combined trigger condition is not reached."

[0181] The sub - module for formulating intervention suggestions calls the standard operation parameters of fetal heart rate monitoring and the maternal oxygen supply classification indicators based on the monitoring trigger determination result, divides the monitoring frequency levels according to the difference in fetal heart rate, integrates the estradiol slope to adjust the oxygen supply flow threshold, and defines the cesarean section priority parameters in combination with the marginal density deviation amplitude to generate the diagnosis and treatment suggestion items for fetal distress in utero.

[0182] Based on the monitoring trigger determination result ("Some indicators are abnormal, and the combined trigger condition is not reached"), and calling the standard operation parameters of fetal heart rate monitoring and the maternal oxygen supply classification indicators stored in the knowledge base. If the determination result is "triggered", suggestions will be formulated according to the specific exceeded indicators and degrees. However, since the combined trigger condition is not reached in this case, but there is an accelerated growth of marginal density and a downward trend of E2, the system will generate a suggestion item of "An accelerated growth of marginal density and a downward trend of estradiol are detected. Although the emergency intervention indication is not reached, it is recommended to strengthen the monitoring frequency to NST once per hour and closely monitor the subsequent changes", rather than direct intervention measures. If the co - trend response value (0.31) is within a certain early warning range (for example, 0.3 - 0.5 represents that attention is needed), the suggestion will also be strengthened, generating the diagnosis and treatment suggestion items for fetal distress in utero.

[0183] A method for assisting in the diagnosis and treatment of obstetrics and gynecology. The method for assisting in the diagnosis and treatment of obstetrics and gynecology is executed based on the above - mentioned obstetrics and gynecology diagnosis and treatment assistance system, and includes the following steps:

[0184] S1: Obtain the physiological data including the patient's body temperature, heart rate, pulse, and blood pressure in obstetric monitoring, record the time stamp, extract the regional marginal density of the synchronous two - dimensional ultrasound image, bind it with the physiological data and dynamically identify it to generate a set of bound diagnosis and treatment data units;

[0185] S2: Compare the fetal heart rate data in the bound diagnosis and treatment data unit set with the fetal heart rate variability threshold. If it exceeds the threshold, mark it as a warning node, extract the synchronous pulse, blood pressure, and image edge density, and generate an obstetric risk warning feature label group;

[0186] S3: According to the fetal heart rate, edge density, and estradiol value in the obstetric risk warning feature label group, classify them into the physical sign class, imaging class, and laboratory class feature groups, perform standard deviation normalization. If the three types of features appear simultaneously in the warning node, integrate them into a three-dimensional vector to generate an obstetric and gynecological feature fusion input body;

[0187] S4: Call the three indicators in the obstetric and gynecological feature fusion input body, screen the indicator with the largest change amplitude and the longest duration, judge the direction consistency and the time window overlap ratio, and generate a multi-dimensional feature collaborative trend label;

[0188] S5: According to the combination of fetal heart rate increase, edge density expansion, and estradiol decrease in the multi-dimensional feature collaborative trend label, match the condition items, trigger the fetal intrauterine distress monitoring condition, and generate an obstetric and gynecological diagnosis and treatment assistance advice item.

[0189] The above is only a preferred embodiment of the present invention, and it does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An obstetrics and gynecology diagnosis and treatment assistance system, characterized in that, The system includes: The parameter acquisition module obtains the patient's body temperature, heart rate, pulse and blood pressure during obstetric monitoring, collects and records the timestamp, extracts the edge density of the synchronous two-dimensional ultrasound image area, binds it with the physiological data and sets a dynamic identifier, and generates a bound diagnosis and treatment data unit set; The feature deviation recognition module calls the fetal heart rate value in the bound diagnosis and treatment data unit set and compares it with the fetal heart rate variability threshold. If it exceeds, it is marked as a warning node. At the same time, it extracts the corresponding pulse, blood pressure and image edge density to judge the synchronous change, and generates an obstetric risk warning feature label group; The feature reconstruction module is based on the fetal heart rate, edge density value and estradiol value in the obstetric risk warning feature label group, classifies them into the physical sign class, imaging class and laboratory class feature groups respectively, performs normalization processing according to the standard deviation collected within the group. If the three types of features appear in the warning node at the same time, they are integrated into the form of a three-dimensional vector to generate an input body for the fusion of obstetrics and gynecology features; The collaborative trend recognition module calls the three indicators in the input body for the fusion of obstetrics and gynecology features, screens the one with the largest change amplitude and the longest duration, judges the direction consistency and the time window overlap ratio, and generates a multi-dimensional feature collaborative trend mark; The collaborative trend recognition module includes: The amplitude screening sub-module calculates the change value sequence of multiple indicators within the time window based on the three indicators in the input body for the fusion of obstetrics and gynecology features, compares the amplitudes of the maximum fluctuation intervals, screens the one with the largest amplitude value, and generates the maximum amplitude trend value; The time association sub-module calls the change time window corresponding to the maximum amplitude trend value indicator, calculates the overlap ratio with the other two indicators in the input body for the fusion of obstetrics and gynecology features, screens the time period with a high overlap ratio, and generates the time window overlap interval value; The trend determination sub-module judges the direction consistency within the time window according to the change direction of the indicators contained in the time window overlap interval value. If the directions are consistent, it is recorded as positive, and if there is a conflict, it is negative. Combine the change amplitude and the overlap ratio to calculate the trend scalar, using the formula: ； By calculating the trend synergy degree value and performing threshold judgment, a multi-dimensional feature collaborative trend mark is obtained; Among them, represents the multi-dimensional feature collaborative trend marker, represents the continuous change range of the th indicator, represents the time window overlap ratio of the th indicator and the target indicator, represents the consistency matching value of the change direction between the th indicator and the remaining indicators in the fusion input body of obstetrics and gynecology features, is the number of indicators participating in trend judgment.

2. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 1, wherein The bound diagnosis and treatment data unit set includes a dynamic identifier, the synchronous binding information of physiological data and images, and a timestamp mark. The obstetric risk warning feature label group is specifically an abnormal recognition mark, a synchronous change parameter group, and a risk feature identifier. The input body for the fusion of obstetrics and gynecology features includes a three-dimensional vector of the physical sign class, a three-dimensional vector of the imaging class, and a three-dimensional vector of the laboratory class. The multi-dimensional feature collaborative trend mark is specifically a change amplitude indicator, a duration parameter, a direction consistency mark, and a time window overlap ratio.

3. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 2, characterized in that, The parameter acquisition module includes: The physiological data acquisition sub-module collects the patient's body temperature, heart rate, pulse and blood pressure during obstetric monitoring, records the timestamp of each data item, classifies and stores them, calculates the data change rate and the fluctuation amplitude, obtains the change intervals of multiple data items, and generates the physiological parameter change interval; The image area extraction sub-module calls the synchronous ultrasound image, extracts the edge information of the fetal area, calculates the image gray edge density, screens the areas that meet the reference density, and obtains the synchronous edge density value; The diagnosis and treatment unit generation sub-module calls the physiological parameter change range and the synchronization edge density value, calculates the relationship between physiological fluctuations and image edges, extracts bound data, and uses the formula: ； Operate to obtain the interactive mapping difference degree value and generate a set of bound diagnosis and treatment data units; Among them, represents the group of physiological parameter fluctuation values, represents the group of image edge densities, represents the group of image edge gray level base differences, represents the group of actual boundary fitting values of the image area, represents the number of timestamp groups participating in the matching, and M is the interactive mapping difference value.

4. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 3, characterized in that, The feature offset recognition module includes: The fetal heart rate early warning recognition sub-module calls the fetal heart rate values in the set of bound diagnosis and treatment data units, obtains the fetal heart rate fluctuation data within each monitoring period, and at the same time calculates the difference between the fluctuation range and the threshold based on the fetal heart rate variability threshold, marks the periods exceeding the threshold, and obtains the early warning mark node value; The synchronous change feature extraction sub-module extracts the pulse, blood pressure, and image edge density data at the corresponding time nodes according to the early warning mark node value, calculates the change rate and the mean change amount, and determines whether there is synchronous fluctuation to obtain the synchronous fluctuation analysis value; The obstetric risk label generation sub-module is based on the synchronous fluctuation analysis value and uses the formula: ； Operate to obtain the change intensity of the fetal heart rate and pulse, the composite fluctuation index under the influence of image features, and combine the consistency of the synchronous change direction to generate an obstetric risk early warning feature label group; Among them, represents the change range of fetal heart rate, represents the change range of pulse, represents the th group of blood pressure values, represents the th group of image edge density values, represents the th group of image edge density change values, is the reference value, is the number of time nodes, is the composite fluctuation index.

5. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 4, wherein The feature reconstruction module includes: The physical sign classification sub-module obtains the continuous measurement value sequence at multiple time nodes based on the fetal heart rate data in the obstetric risk early warning feature label group, calculates the standard deviation and performs within-group standard deviation normalization processing, classifies it into the physical sign feature group, and generates the physical sign class normalization value; The image classification sub-module obtains the pixel distribution area of the edge density value in the image frame, extracts the maximum value and the average value, calculates its standard deviation and uses within-group standard deviation normalization, and classifies it into the image feature group to obtain the image class normalization value; The multi-source feature fusion sub-module calls the physical sign class normalization value and the image class normalization value, obtains the measurement data of the estradiol value, calculates the coefficient of variation and eliminates the abnormal points, and uses the formula: ； Integrate the three types of normalization values, determine whether they co-occur at the same warning node, and if they do, combine them in the form of a three-dimensional vector to generate a feature fusion input body; Among them, represents the coefficient of variation after the fusion of three types of features, represents the coefficient of variation of the normalized values of the physical signs category, represents the mean of the time series of the physical signs category, represents the skewness of the time series of the physical signs category, represents the kurtosis of the time series of the physical signs category, represents the coefficient of variation of the normalized values of the imaging category, represents the coefficient of variation of the normalized value of estradiol, represents the absolute value operation to solve the difference between the coefficient of variation of the estradiol value and the mean of the physical signs.

6. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 5, characterized in that, The system further includes: The diagnosis and treatment recommendation generation module matches the condition items in the recommendation list according to the combination of the increase in fetal heart rate, the expansion of the edge density, and the decrease in estradiol in the multi-dimensional feature collaborative trend marking, triggers the fetal intrauterine distress monitoring condition, and generates the obstetrics and gynecology diagnosis and treatment assistance recommendation items; The obstetrics and gynecology diagnosis and treatment assistance recommendation items include the monitoring recommendation type, the intervention recommendation type, and the warning level identifier.

7. The obstetrics and gynecology diagnosis and treatment assistance system according to claim 6, wherein The diagnosis and treatment recommendation generation module includes: The collaborative trend analysis sub-module extracts the difference between the fetal heart rate baseline value and the preset safe baseline range based on the multi-dimensional feature collaborative trend marking, calculates the deviation amplitude between the growth gradient of the edge density per unit time and the critical expansion rate, fits the linear decline slope of the estradiol concentration change rate, assigns weights to the difference, the deviation amplitude, and the slope, and performs superposition operation to generate the collaborative trend response value; The monitoring condition matching sub-module calls the upper safety limit threshold of fetal heart rate, the critical threshold for edge density expansion, and the lower threshold for estradiol fluctuation. It compares the fetal heart rate difference in the collaborative trend response value with the upper safety limit, the deviation amplitude of the edge density with the critical rate, and the estradiol slope with the lower fluctuation limit. According to the joint trigger logic of all three parameters exceeding the limit simultaneously, it generates a monitoring trigger determination result; Based on the monitoring trigger determination result, the intervention recommendation formulation sub-module calls the standard operation parameters of fetal heart rate monitoring and the maternal oxygen supply classification index. It divides the monitoring frequency level according to the fetal heart rate difference, integrates the estradiol slope to adjust the oxygen supply flow threshold, and defines the cesarean section priority parameter in combination with the deviation amplitude of the edge density, generating items of diagnosis and treatment recommendations for fetal distress in utero.

8. An auxiliary method for diagnosis and treatment in obstetrics and gynecology, characterized in that, Execute according to the obstetrics and gynecology diagnosis and treatment assistance system described in any one of claims 1-7, including the following steps: S1: Obtain the physiological data including the patient's body temperature, heart rate, pulse, and blood pressure in obstetric monitoring and record the time stamp. Extract the regional edge density of the synchronized two-dimensional ultrasound image, bind it to the physiological data and dynamically mark it, generating a set of bound diagnosis and treatment data units; S2: Compare the fetal heart rate data in the set of bound diagnosis and treatment data units with the fetal heart rate variability threshold. If it exceeds the threshold, mark it as a warning node, extract the synchronized pulse, blood pressure, and image edge density, generating a set of obstetric risk warning feature labels; S3: According to the fetal heart rate, edge density, and estradiol value in the set of obstetric risk warning feature labels, classify them into the physical sign class, imaging class, and laboratory class feature groups, perform standard deviation normalization. If the three types of features appear simultaneously at the warning node, integrate them into a three-dimensional vector, generating an input body for feature fusion in obstetrics and gynecology; S4: Call the three indicators in the input body for feature fusion in obstetrics and gynecology, screen the indicator with the largest change amplitude and the longest duration, judge the direction consistency and the time window overlap ratio, generating a multi-dimensional feature collaborative trend label; Based on the three indicators in the input body for feature fusion in obstetrics and gynecology, calculate the change value sequence of multiple indicators within the time window, compare the amplitude of the largest fluctuation interval, and screen the one with the largest amplitude value, generating the maximum amplitude trend value; Call the change time window of the indicator corresponding to the maximum amplitude trend value, calculate the overlap ratio with the other two indicators in the input body for feature fusion in obstetrics and gynecology, screen the time period with a high overlap ratio, generating the time window overlap interval value; According to the change direction of the indicators contained in the time window overlap interval value, judge the direction consistency within the time window. If the directions are consistent, record it as positive, and if there is a conflict, it is negative. Combine the change amplitude and the overlap ratio to calculate the trend scalar, using the formula: ； By calculating the trend synergy degree value and performing threshold judgment, obtain the multi-dimensional feature collaborative trend label; Among them, represents the multi-dimensional feature collaborative trend marker, represents the continuous change amplitude of the th index, represents the time window overlap ratio between the th index and the target index, represents the consistency matching value of the change direction between the th index and the remaining indices in the fusion input body of obstetrics and gynecology features, is the number of indices participating in the trend judgment; S5: According to the combination of fetal heart rate increase, edge density expansion, and estradiol decrease in the multi-dimensional feature collaborative trend label, match the condition items, trigger the monitoring conditions for fetal distress in utero, generating items of obstetrics and gynecology diagnosis and treatment assistance recommendations.

Images