A multi-parameter dynamic monitoring system and method for a patient at risk of preterm labor

Abstract

This invention discloses a multi-parameter dynamic monitoring system and method for patients undergoing pregnancy maintenance, relating to the field of dynamic monitoring technology during pregnancy. The system includes: binding the patient's identity to an external communication terminal and server, writing fixed monitoring time windows and retesting rules, and establishing individualized equivalent conversion relationships based on serum control collection and interstitial fluid sampling during fasting windows to obtain individual baseline statistics for monitoring indicators; quantitatively sampling interstitial fluid during each monitoring time window to obtain detection values ​​for glucose, lipid profiles, inflammatory factors, and mediating proxy indicators of the maternal-fetal interface, and generating monitoring data by combining quality scores and retesting rules; standardizing the effective data to construct proxy features for metabolic stress, inflammatory amplification, and maternal-fetal interface functional imbalance, and performing sequential constraint discrimination based on lag time windows to form a causal path score; generating risk levels by combining clinical events and pushing them to medical personnel terminals to achieve dynamic assessment of pregnancy risks.

Description

Technical Field

[0001] This invention relates to the field of pregnancy dynamic monitoring technology, and in particular to a multi-parameter dynamic monitoring system and method for patients undergoing pregnancy maintenance. Background Technology

[0002] Perinatal management for patients undergoing pregnancy maintenance typically involves outpatient follow-up, ultrasound assessment, and laboratory testing. Observations of biochemical indicators such as inflammatory factors and glucose and lipid metabolism are usually obtained through fasting or specific time-point blood sampling, and are used in conjunction with symptom and sign records to assess changes in pregnancy status. With the development of wearable physiological monitoring, remote follow-up, and data platforms, multi-parameter recording methods for pregnancy are gradually emerging in clinical practice. These methods archive patients' daily symptoms, medication information, and multi-time-point physiological and biochemical test data on the same platform to support continuous management and follow-up decisions.

[0003] Based on the monitoring and analysis framework of this invention, the conventional methods described above still have two aspects that need further improvement when implemented in engineering: First, multiple parameters often come from different sampling points and different detection systems, making it difficult to construct the sequential relationship between metabolic changes and inflammatory changes on a unified time axis and form a calculable lag constraint discrimination, resulting in an unstable quantitative expression of chain processes such as "metabolic stress - inflammation amplification - maternal-fetal interface functional imbalance"; Second, home or multi-time point sampling is easily affected by the sampling channel status, readout drift and missing windows. If there is a lack of quantifiable data quality scoring and retesting rules, the repeatability of dynamic feature construction and cross-day comparison is difficult to guarantee. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance, which solves the problems of existing technologies where multiple source parameters are difficult to form a quantifiable lag causal path on a unified time axis, and where there is a lack of unified criteria for monitoring data quality control and missing data retesting.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance, comprising: establishing a monitoring subject file and deploying a data acquisition carrier; binding the patient's identity with an external communication terminal and server, writing daily monitoring time windows and retesting rules, and establishing an individualized equivalent conversion relationship through serum control collection and interstitial fluid sampling during fasting windows to obtain individual baseline statistics for each monitoring indicator; quantitatively sampling interstitial fluid at each fixed time window to obtain detection values ​​of glucose, lipid profile, inflammatory factors, and mediating proxy indicators of the maternal-fetal interface, calculating a quality score, and retesting according to retesting rules when the quality is insufficient to generate monitoring data; standardizing the effective monitoring data and constructing metabolic stress characteristics, inflammatory amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics, and performing constraint discrimination based on lag time windows to obtain a causal path score; collecting the patient's clinical events on the same day and performing consistency verification, merging the causal path score and clinical events to generate a risk level and treatment record package, and pushing it to the medical personnel's terminal.

[0007] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the step of establishing a monitoring object file and deploying the data acquisition carrier includes: generating a unique patient identification code for the patient on the server side and establishing a monitoring object file corresponding to the patient's basic information; binding the patient identification code to the terminal identifier of the external communication terminal; binding the patient identification code to the carrier identifier of the data acquisition carrier and establishing a one-to-one correspondence between the data acquisition carrier and the monitored patient; and writing a daily monitoring time window table and retesting rules to the data acquisition carrier so that the data acquisition carrier automatically enters the sampling and testing process when the corresponding time window arrives.

[0008] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the establishment of individualized equivalent conversion relationships includes: collecting serum samples from patients within a fasting monitoring time window and obtaining corresponding serum test values; within the same serum collection time window, performing interstitial fluid sampling on a collection carrier and obtaining interstitial fluid test values; pairing serum test values ​​and interstitial fluid test values ​​according to timestamps to form paired data sets; and calculating and generating individualized equivalent conversion parameters for each monitoring indicator based on at least two sets of paired data from different dates.

[0009] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the acquisition of individual baseline statistics for each monitoring indicator includes: after establishing individualized equivalent conversion relationships, setting multiple consecutive natural days as the baseline collection period; the collection carrier performing interstitial fluid sampling according to the daily monitoring time window during the baseline collection period; screening the acquired equivalent serum values ​​through quality scoring; and calculating the individual baseline mean and individual baseline standard deviation for each monitoring indicator based on the screened data.

[0010] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the calculation of the quality score includes: when a fixed monitoring time window arrives, the acquisition carrier performs quantitative interstitial fluid sampling; during the sampling process, the impedance characteristic value and sampling driving characteristic value of the sampling channel are acquired; the impedance characteristic value and sampling driving characteristic value are compared with the corresponding baseline statistics to calculate the impedance normalization deviation and driving normalization deviation; and based on the impedance normalization deviation and driving normalization deviation, a quality score characterizing the state of the sampling channel is generated.

[0011] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the retesting according to the retesting rules includes: comparing the quality score of the current monitoring time window with a quality score threshold; generating a retesting trigger command when the quality score is lower than the quality score threshold; performing retesting sampling on the acquisition carrier within a delay period after the end of the current monitoring time window; and using the data obtained from the retesting sampling as the valid monitoring data for the current monitoring time window.

[0012] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the standardization of effective monitoring data and the construction of metabolic stress features, inflammation amplification features, and maternal-fetal interface functional imbalance proxy features include: obtaining the individual baseline mean and individual baseline standard deviation corresponding to each monitoring indicator; standardizing the equivalent serum values ​​in the effective monitoring data; constructing metabolic stress features based on the standardized glucose-related data; constructing inflammation amplification features based on the standardized inflammatory factor data; and constructing maternal-fetal interface functional imbalance proxy features through maternal-fetal interface mediating proxy indicators.

[0013] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance described in this invention, the step of constraint discrimination based on a lag time window includes: determining whether metabolic stress characteristics have reached the metabolic triggering condition within the time window; determining whether inflammatory amplification characteristics have reached the inflammatory triggering condition within the lag time window after metabolic triggering; determining whether maternal-fetal interface functional imbalance proxy characteristics have reached the abnormal condition within a further lag time window; and generating a feature causal path score when the constraints are determined to be met sequentially in chronological order.

[0014] As a preferred embodiment of the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance as described in this invention, the generation of risk level and treatment record package includes: collecting the patient's clinical event information for the day and uploading it to a server; performing consistency verification on the clinical event information and obtaining the verification result; jointly processing the verified clinical event information and causal path score to generate a risk level; and encapsulating the risk level, causal path score, and corresponding monitoring data into a treatment record package and pushing it to the medical personnel's terminal.

[0015] Secondly, this invention provides a multi-parameter dynamic monitoring system for patients undergoing pregnancy maintenance, comprising a file establishment and calibration module, a sampling quality control module, a causal analysis module, and a risk management module. The file establishment and calibration module is responsible for establishing a file for the monitored patient and deploying the collection carrier, binding the patient's identity to the external communication terminal and server, writing daily monitoring time windows and retesting rules, and establishing individualized equivalent conversion relationships through serum control collection and interstitial fluid sampling during fasting windows to obtain individual baseline statistics for each monitoring indicator. The sampling quality control module is responsible for quantitatively sampling interstitial fluid at fixed time windows to obtain... The system collects the detection values ​​of glucose, blood lipid profile, inflammatory factors, and maternal-fetal interface mediator proxy indicators, calculates the quality score, and performs retesting according to retesting rules when the quality is insufficient, generating monitoring data. The causal analysis module is responsible for standardizing the effective monitoring data and constructing metabolic stress characteristics, inflammatory amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics, and performs constraint discrimination based on the lag time window to obtain the causal path score. The risk management module is responsible for collecting the patient's clinical events on the same day and performing consistency verification, merging the causal path score and clinical events to generate a risk level and management record package, and pushing it to the medical personnel terminal.

[0016] The beneficial effects of this invention are as follows: By constructing metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics from effective monitoring records, and by using the sequential constraint discrimination rules of lag time windows, the chain process from metabolic stress to inflammation amplification and then to maternal-fetal interface functional imbalance is quantitatively expressed, transforming pregnancy risk assessment from single-point indicator judgment to dynamic path discrimination based on causal order; by combining a quality scoring mechanism and retest rules during the monitoring process, objective control of monitoring data quality is achieved, reducing the uncertain impact of occasional sampling anomalies or missing windows on the dynamic feature construction and risk discrimination results. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance.

[0019] Figure 2 This is a schematic diagram of a multi-parameter dynamic monitoring system used for patients seeking to maintain pregnancy.

[0020] Figure 3 This is a flowchart for sampling quality control.

[0021] Figure 4 A flowchart generated for causal path scoring.

[0022] Figure 5 The figure shows the results of the three types of dynamic characteristic curves and the lag time window constraint.

[0023] Figure 6 The graph shows the impact of the quality scoring threshold on the performance of missing windows and chain discrimination. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Reference Figures 1-6 This is one embodiment of the present invention, which provides a multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance, comprising the following steps: S1. Establish monitoring subject files and complete the deployment of collection carriers; bind patient identities with in vitro communication terminals and servers, write them into daily monitoring time windows and retest rules, and establish individualized equivalent conversion relationships through serum control collection and interstitial fluid sampling during fasting windows to obtain individual baseline statistics for each monitoring indicator.

[0028] Furthermore, medical staff create monitoring subject files on the server side and generate unique patient identifiers, which are recorded as patient identification codes. The patient identification code consists of a fixed-length string and is stored in association with fields such as the patient's name, age, gestational week, type of adverse pregnancy history, current medication list, and measurement start date.

[0029] The server generates a binding certificate based on the patient's identification code; the binding certificate includes at least the patient's identification code, a one-time random number, an expiration date, and encrypted verification information; and displays the binding certificate on the external communication terminal in the form of a QR code or a pairing code.

[0030] After the external communication terminal reads the binding certificate, it generates a terminal identifier locally and writes the terminal identifier and the patient's identity code into the local secure storage area. At the same time, it sends a binding request to the server. After the server verifies the validity period of the binding certificate and the encrypted verification information, it writes the relationship between the patient's identity code and the terminal identifier into the server-side binding table, forming the first binding relationship.

[0031] Furthermore, under sterile conditions, the collection carrier is placed in a predetermined subcutaneous position on the outer side of the patient's upper arm, and fixation and wound treatment are completed; after implantation, the deployment timestamp is recorded and written into the patient's file.

[0032] The external communication terminal enters pairing mode and establishes a wireless communication connection with the acquisition carrier; the acquisition carrier returns the carrier identifier and the current clock value; the carrier identifier is a unique identifier fixed at the factory.

[0033] The external communication terminal uploads the carrier identifier, terminal identifier, and patient identification code to the server; after the server verifies that the patient identification code has been bound to the terminal identifier, it associates the carrier identifier with the patient identification code, forming a second binding relationship.

[0034] The external communication terminal performs clock calibration on the acquisition carrier, adjusting the acquisition carrier's clock to match the server's time based on the terminal's current standard time; and writes a clock calibration completion mark and calibration deviation value.

[0035] Furthermore, the server sends a fixed monitoring time window table to the external communication terminal; the fixed monitoring time window table includes a set of time points for sampling that need to be performed on the same day, which are usually set by default to fasting window, post-breakfast window, afternoon window and evening window; the set of sampling time points are usually fixed times or short time deviations allowed before and after the fixed times.

[0036] The external communication terminal writes a fixed monitoring time window table into the data acquisition carrier, so that the data acquisition carrier automatically enters the sampling and detection process when the corresponding time window is reached each day.

[0037] Simultaneously, the retesting rules should be included; these rules should at least include a quality score threshold, a retesting delay duration, and an upper limit on the number of retests.

[0038] Furthermore, in this embodiment, the retesting rule is fixed as follows: when the quality score of the sampling result of the time window is lower than the quality score threshold, a retest is automatically performed after a fixed delay after the current time window ends, and the maximum number of retests is one.

[0039] It should be noted that the quality score threshold is determined by statistical analysis of the quality score distribution of multiple normal sampling data during the baseline acquisition period. Sampling results below the quality score threshold deviate from the baseline in impedance characteristics and are thus judged as having insufficient sampling quality. The value range is usually [0.6, 0.85].

[0040] The data acquisition device saves the fixed monitoring time window and retesting rules and sends back a successful receipt; the server archives the receipt and marks the patient as having entered the formal monitoring state.

[0041] Furthermore, during the fasting window on the day the monitoring begins, medical staff collect venous blood from patients, and the samples are sent for testing to obtain serum test values. These serum test values ​​include at least the serum test values ​​corresponding to glucose, inflammatory factors, lipid profile, and maternal-fetal interface mediator indicators.

[0042] It should be noted that the maternal-fetal interface mediator index refers to biochemical indicators used to characterize the interaction between the maternal decidua and the placental trophoblast, but not to directly measure the structure of the maternal-fetal interface itself. The maternal-fetal interface mediator index indirectly quantifies whether the function of the maternal-fetal interface is in a stable state by reflecting the balance of placental angiogenesis and perfusion.

[0043] In this embodiment, the mediating proxy index of the maternal-fetal interface is selected as angiogenesis-related factor index, specifically including soluble Flt-1 and placental growth factor, and the ratio of the two is used to construct a proxy variable for the functional imbalance of the maternal-fetal interface.

[0044] Within the same fasting window after venous blood collection, the in vitro communication terminal sends a synchronous sampling command to the collection carrier. Upon receiving the command, the collection carrier immediately performs a quantitative sampling of interstitial fluid and multi-index detection to obtain the interstitial fluid detection value corresponding to the serum sample timestamp.

[0045] The server pairs the serum test values ​​and interstitial fluid test values ​​within the fasting window according to the same timestamp to form the first set of paired data.

[0046] After the monitoring is started, at a fixed number of days, such as the seventh day, venous blood collection and synchronous sampling are repeated at the fasting window to form a second set of paired data.

[0047] It should be noted that both sets of paired data were written into the patient's file as the calibration basis for individualized equivalence conversion.

[0048] Furthermore, for each monitoring indicator, an individualized equivalent conversion relationship is established to convert the interstitial fluid detection value into the equivalent serum value. The individualized equivalent conversion relationship adopts a first-order linear mapping, where the proportionality coefficient is used to reflect the difference in the concentration ratio between interstitial fluid and serum between individuals, and the bias coefficient is used to reflect the fixed offset.

[0049] Specifically, for any monitoring indicator, the interstitial fluid test value obtained from a single sampling is recorded as the interstitial fluid value, and the serum test value obtained from clinical testing is recorded as the serum value.

[0050] The individualized proportion coefficient and bias coefficient of patients, converted to equivalent serum values, are expressed as follows: ; in, Indicators The mapping coefficient of interstitial fluid-serum ratio, Indicators Interstitial fluid serum bias coefficient. Indicates the collection medium in Japan Detection time window The measured values ​​of interstitial fluid parameters, This represents the equivalent serum value obtained by converting the interstitial fluid value.

[0051] Furthermore, given two sets of paired data, the individualized scaling factor and bias factor are solved by determining a straight line using two points, specifically: when When the line is determined by two points, it is obtained as follows: ; ; in, This indicates that the second set of collection carriers was in Japan. Detection time window The measured values ​​of interstitial fluid parameters, This indicates that the first set of collection carriers was in Japan. Detection time window The measured values ​​of interstitial fluid parameters, This indicates that the second set of collection carriers was in Japan. Detection time window The measured serum values, This indicates that the first set of collection carriers was in Japan. Detection time window The measured serum values.

[0052] It should be noted that when the interstitial fluid values ​​of the second group are equal to those of the first group, resulting in a denominator of zero, the individualization ratio coefficient is set to 1, and the bias coefficient is set to the serum value of the first group minus the interstitial fluid value of the first group.

[0053] The server writes the individualized proportional coefficient and bias coefficient corresponding to each indicator into the patient parameter table.

[0054] Furthermore, after establishing the equivalent conversion relationship, the baseline collection period begins. The baseline collection period is a fixed number of consecutive days. For example, the carrier is automatically sampled and tested according to a fixed monitoring time window for three days. The server will convert each record into an equivalent serum value and then archive it.

[0055] For each indicator, the set of all equivalent serum values ​​that meet the quality score threshold during the baseline collection period is denoted as the baseline sample set.

[0056] The server calculates the individual baseline mean and individual baseline standard deviation for each indicator's baseline sample set, which are used to describe the individual level and dispersion of patients under conditions without acute fluctuations.

[0057] It should be noted that when the number of available samples for an indicator is insufficient during the baseline collection period, making it impossible to calculate the standard deviation, the baseline standard deviation of the indicator will be used as the minimum positive threshold.

[0058] The baseline mean and baseline standard deviation of each indicator were written into the patient baseline table.

[0059] Furthermore, during the baseline acquisition period, in addition to biochemical indicators, the acquisition carrier simultaneously acquires reference signals to reflect the sampling channel status and electrochemical interface stability; among which, the reference signals include impedance characteristic values ​​and sampling drive current characteristic values.

[0060] The server also calculates the baseline mean and baseline standard deviation for the reference signal and writes them into the quality reference table.

[0061] It should also be noted that by setting a fixed monitoring time window within the same monitoring object and quantitatively sampling interstitial fluid, and recording multiple physiological and biochemical parameters with a unified timestamp, the alignment and continuous acquisition of multi-source monitoring parameters on a unified time axis are achieved. This enables metabolic indicators, inflammatory indicators, and maternal-fetal interface proxy indicators to be analyzed in a comparable manner within the same time series, thereby improving the temporal consistency and engineering feasibility of dynamic monitoring results during pregnancy.

[0062] S2. Quantitatively sample interstitial fluid at fixed time windows to obtain the detection values ​​of glucose, blood lipid profile, inflammatory factors, and maternal-fetal interface mediator indicators, calculate the quality score, and retest according to the retest rules when the quality is insufficient to generate monitoring data.

[0063] Furthermore, under the control of a fixed monitoring time window, the sampling carrier enters the sampling preparation state at the start of any fixed monitoring time window; the sampling preparation state includes at least self-test power supply, electrode polarization stability judgment, microchannel pre-filling judgment, and clock verification.

[0064] When the acquisition carrier is in the sampling preparation state, it reads the internal clock and generates a timestamp for this sampling. The timestamp includes at least the date, time, and fixed monitoring time window number. The acquisition carrier writes the timestamp into the sampling buffer for this sampling.

[0065] Furthermore, the sampling carrier activates the quantitative sampling mechanism to extract a fixed volume of interstitial fluid from the subcutaneous interstitial fluid entry channel; wherein, the volume is a fixed value and is solidified at the time of manufacture or in the parameters written into the carrier, and is recorded as the single sampling volume.

[0066] The collection carrier introduces interstitial fluid into the detection chamber and performs a fixed period of mixing and reaction waiting. The waiting period is denoted as the binding waiting time, and the value of the binding waiting time is fixed. During the waiting period, the collection carrier keeps the detection chamber in a closed or controlled flow state to ensure detection consistency.

[0067] After the acquisition carrier completes the loading of the detection cavity, it records the sampling drive characteristic values ​​as input for subsequent quality scoring; among them, the sampling drive characteristic values ​​include the peak value of the sampling drive current and the sampling drive duration.

[0068] Furthermore, the acquisition carrier sequentially acquires multiple indicator detection values.

[0069] In this embodiment, the order in which the multi-indicator detection values ​​are obtained is as follows: the first category is glucose detection values; the second category is lipid profile detection values, which include total cholesterol detection values ​​and triglyceride detection values; the third category is inflammatory factor detection values, which include interleukin-6 detection values, tumor necrosis factor-α detection values, and C-reactive protein detection values; and the fourth category is maternal-fetal interface mediation proxy indicator detection values, which include soluble Flt-1 detection values ​​and placental growth factor detection values.

[0070] During the readout of each indicator, the acquisition carrier applies excitation to the indicator channel and acquires the raw electrical signal; wherein the raw electrical signal includes at least one or a combination of current, voltage or impedance; and writes the raw electrical signal into the current sampling buffer.

[0071] The data acquisition carrier generates a raw signal value for each indicator and binds the raw signal value with the indicator name and timestamp to form a multi-indicator signal set for this sampling.

[0072] Furthermore, the acquisition carrier converts the obtained multi-index signal set into interstitial fluid detection values ​​of the indexes according to calibration; wherein, the calibration conversion adopts a first-order linear conversion, expressed as: ; in, Indicators Interstitial fluid test values, Indicators Signal, Represents the sensitivity coefficient. This represents the zero-point correction constant.

[0073] It should be noted that the sensitivity coefficient is determined by applying multiple sets of standard samples of known concentrations sequentially to the corresponding detection channel during the calibration phase, acquiring stable response signals, and performing a linear fit between the signal and concentration. The value is typically within a certain range. The zero-point correction constant is determined by using a zero-concentration control sample as input during the calibration phase, acquiring the corresponding baseline response signal, and then linearly converting the baseline signal to determine the corresponding offset.

[0074] The data collection carrier constructs a data recording package for this sampling; the data recording package includes at least the patient identification code, carrier identifier, terminal identifier, timestamp, multi-index interstitial fluid detection values, multi-index signals, and sampling driving feature values.

[0075] The data acquisition carrier sends the data recording package to the external communication terminal, which then forwards the data recording package to the server; the server then stores the data recording package in the database according to the patient's identification code.

[0076] Furthermore, after each sampling, the acquisition carrier performs impedance detection to obtain impedance characteristic values; the impedance characteristic values ​​and the sampling driving characteristic values ​​are used together as inputs for quality scoring.

[0077] During the baseline acquisition period, the server has established the baseline mean and baseline standard deviation of impedance characteristic values ​​and sampling drive characteristic values.

[0078] The impedance characteristic value and the sampled drive characteristic value are calculated separately to obtain the impedance normalization error and the drive normalization error.

[0079] Furthermore, the impedance normalization deviation and the drive normalization deviation are calculated by standardizing the deviations relative to the corresponding baseline statistics.

[0080] Specifically, the server acquires the baseline mean and standard deviation of impedance characteristic values ​​established during the baseline acquisition period, as well as the baseline mean and standard deviation of sampling driving characteristic values; subtracts the impedance baseline mean from the impedance characteristic values ​​acquired in the current monitoring time window and divides it by the impedance baseline standard deviation to obtain the impedance normalization deviation; at the same time, subtracts the driving baseline mean from the current sampling driving characteristic values ​​and divides it by the driving baseline standard deviation to obtain the driving normalization deviation.

[0081] Impedance normalization deviation and drive normalization deviation are mapped to quality scores. Specifically, each normalization deviation is exponentially decayed and multiplied to obtain a quality score between 0 and 1. The larger the normalization deviation, the lower the quality score.

[0082] Quality rating, expressed as: ; in, In Japan Detection time window Quality rating In Japan Detection time window The impedance normalization deviation, In Japan Detection time window The driving normalization bias.

[0083] The server compares the quality score with the quality score threshold. When the quality score is lower than the quality score threshold, the server sends a retest command to the external communication terminal. The external communication terminal then forwards the retest command to the acquisition carrier.

[0084] After receiving the retest instruction, the data acquisition carrier performs a retest sampling once after the delay period written in the retest rules, after the same fixed monitoring time window has ended. The retest sampling completely repeats the sampling process and generates a retest data record package.

[0085] The server also calculates a quality score for the retest data record package; when the retest quality score reaches the quality score threshold, the retest data record package is marked as a valid record, and the original record is marked as an invalid record and not included in subsequent calculations.

[0086] When the retest quality score is still lower than the quality score threshold, the data of the fixed monitoring time window is marked as a missing window, archived but not included in subsequent calculations, and a fixed self-test sampling instruction is added in the fasting window of the next day.

[0087] Furthermore, the server generates a final valid data record for each fixed monitoring time window; if a valid retest record exists, the retest record is used as the final valid record; otherwise, the original record is used as the final valid record; if both are invalid, the time window is recorded as missing.

[0088] It should also be noted that by establishing an individualized equivalent conversion relationship based on fasting window serum control collection and interstitial fluid synchronous sampling, and forming corresponding individual baseline statistics, the interstitial fluid detection results are traceably mapped to equivalent serum indicators. This enables the monitoring data obtained by different patients at different physiological levels to have a unified basis for comparison, which is conducive to improving the stability and interpretability of dynamic monitoring results in clinical applications.

[0089] It should also be noted that by introducing a quality scoring mechanism and combining it with fixed retesting rules during the monitoring process, the sampling channel status and readout stability are quantitatively evaluated and retesting is automatically triggered, thus achieving objective control over the quality of monitoring data and reducing the uncertain impact of occasional sampling anomalies or missing windows on the dynamic feature construction and risk judgment results.

[0090] S3. Standardize the effective monitoring data and construct metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics. Then, perform constraint discrimination based on the lag time window to obtain the causal path score.

[0091] Furthermore, the server reads all data records within the evaluation period from the time series data table and filters them according to the patient's identification code. Specifically, the server uses the quality score as the criterion to screen valid records, and records with a quality score not lower than the quality score threshold and belonging to the final valid records are recorded as valid records; records with missing windows or insufficient quality and no qualified retest are recorded as invalid records.

[0092] The server aligns valid records according to fixed monitoring time window numbers. In this embodiment, taking the default settings (fasting window, post-breakfast window, afternoon window, and evening window) as an example, a time series structure of four windows within the day is formed.

[0093] For any natural day Construct the time window sequence for the current day ,in, Indicates a fasting window. Indicates the window after breakfast. Indicates the afternoon window. This indicates the evening window; and the equivalent serum values ​​for each time window of the current day are written into the intraday structure according to the time window number.

[0094] If a missing window exists in a calendar day, the server marks the missing window number in the calendar day structure and executes the missing window handling rules during subsequent feature construction.

[0095] Furthermore, the server reads the individual baseline mean and individual baseline standard deviation of each indicator.

[0096] For any indicator at any given natural day and time window, calculate the standardized value of the equivalent serum value, which is expressed as: ; in, Indicators exist Standardized value at that point, and They represent the monitoring indicators respectively. Individual baseline mean and standard deviation.

[0097] Furthermore, metabolic stress is defined as a combination of postprandial glucose abnormalities and intraday glucose area under the curve abnormalities, and is synthesized into a metabolic stress index with fixed weights. The fixed weights are determined by performing correlation analysis and stability assessment on the standardized values ​​of postprandial glucose and intraday glucose area under the curve in baseline and previous follow-up samples, and selecting the weight proportions with stable contribution ranking and the lowest fluctuation coefficients in multiple batches of samples as fixed weights.

[0098] Specifically, the server reads the natural day. The equivalent serum glucose values ​​corresponding to the four fixed monitoring time windows are recorded as the natural days. The glucose sequence.

[0099] The server uses the fixed time interval corresponding to each fixed monitoring time window as the integration step size and adopts the trapezoidal integration method to calculate the area of ​​the glucose curve within the day. That is, the glucose values ​​of two adjacent time windows are averaged and multiplied by the time interval between the two time windows, and then the area of ​​the curve is obtained by accumulating the adjacent time windows throughout the day.

[0100] The server calculates the post-meal increment, which is the glucose value in the post-breakfast window minus the glucose value in the fasting window.

[0101] The server reads the standardized value of glucose in the post-breakfast window, the standardized value of the intraday glucose curve area, and the standardized value of the postprandial increment, respectively. The baseline mean and baseline standard deviation of the standardized value of the intraday glucose curve area and the standardized value of the postprandial increment are obtained from the curve area and postprandial increment sequences calculated by the same method during the baseline acquisition period.

[0102] The server synthesizes a metabolic stress index with fixed weights, which is obtained by linearly weighting and summing the standardized values ​​of glucose in the post-breakfast window, the standardized values ​​of the area under the curve, and the standardized values ​​of the postprandial increment.

[0103] It should be noted that the weights in the linear weighting of the metabolic stress index are determined by offline statistical analysis of historical monitoring data from the baseline and risk-triggered periods. The relative contribution of each metabolic characteristic to the occurrence of subsequent inflammatory amplification events is ranked and fixed. The value range of each weight is usually [0,1] and the sum of each weight is 1.

[0104] If there is a missing window that affects the calculation of the metabolic stress index on a given day, such as a missing fasting window or a missing post-breakfast window, the metabolic stress index will be processed according to the missing window rules.

[0105] Furthermore, inflammation amplification is defined as abnormal synthesis of inflammatory factors in the post-breakfast window and abnormal increase in inflammatory factors relative to the previous day.

[0106] Specifically, the server reads the natural day. The standardized values ​​of three inflammatory factors in the post-breakfast window were: interleukin-6, tumor necrosis factor-α, and C-reactive protein.

[0107] The server synthesizes the three standardized values ​​into an inflammation index with fixed weights, that is, by linearly weighting and summing the three standardized inflammation values ​​according to their weights to obtain the inflammation index.

[0108] It should be noted that the weights in the linear weighting of the inflammatory index are determined by offline statistical analysis of historical monitoring data from the baseline period and the risk triggering period, and normalized according to the relative contribution of each inflammatory factor to the changes in the proxy indicators of maternal-fetal interface functional imbalance. The value range of each weight is usually [0,1] and the sum of each weight is 1.

[0109] Server for natural days and the previous day The difference between the inflammation index and the inflammatory index is used to obtain the inflammation increment.

[0110] If the inflammation index from the previous day is unavailable, the current day's inflammation increment is set to 0 and marked as missing previous day reference.

[0111] Furthermore, the ratio of soluble Flt-1 to placental growth factor was used as a proxy to mutate the maternal-fetal interface functional imbalance, and the logarithm of the ratio was used to form a stable characteristic.

[0112] Specifically, the server reads the natural day. The equivalent serum values ​​of two maternal-fetal interface mediators in the post-breakfast window were obtained, namely, the equivalent serum value of soluble Flt-1 and the equivalent serum value of placental growth factor.

[0113] The server calculates the ratio and obtains the maternal-fetal interface ratio.

[0114] The server takes the natural logarithm of the maternal-fetal interface ratio to obtain the maternal-fetal interface index.

[0115] The server standardizes the maternal-fetal interface index and obtains the standardized value of the maternal-fetal interface.

[0116] Furthermore, to ensure that the monitoring can still be carried out even if patients experience missing windows during actual home monitoring, rules for handling missing windows are specified.

[0117] Specifically, when the fasting window or post-breakfast window required for calculating the metabolic stress index is missing, the metabolic stress index for that day is fixed as undiscriminable, and the metabolic triggering condition is fixed as invalid in the causal chain discrimination.

[0118] If any adjacent window required for calculating the area of ​​the curve is missing, the area of ​​the curve for that day will not be calculated, and the standardized value of the area of ​​the curve will be fixed at 0; however, the postprandial increment can still be calculated and participate in the synthesis of the metabolic stress index, provided that both the fasting window and the post-breakfast window exist.

[0119] If any of the three inflammatory factors required for the inflammation index are missing, the inflammation index for that day will be set to unavailable; the inflammation triggering conditions will be determined to be invalid.

[0120] When maternal-fetal interface indicators are missing, the standardized value of the maternal-fetal interface for that day is fixed as unavailable; the triggering conditions for the maternal-fetal interface are fixed as invalid.

[0121] When the equivalent serum value of any monitoring indicator exceeds the allowable range of the project, the current record of the monitoring indicator is marked as invalid and treated as a missing window.

[0122] For example, the engineering allowable range for equivalent serum glucose can be set to 2.0 mmol / L to 20.0 mmol / L. When the detected equivalent serum glucose is below 2.0 mmol / L or above 20.0 mmol / L, it is judged to be outside the engineering allowable range and marked as an invalid record.

[0123] Furthermore, the causal path is defined as a three-stage discrimination that must satisfy the sequence and lag time window; the three-stage discrimination includes the occurrence of metabolic stress and inflammation amplification within the lag window, and the occurrence of maternal-fetal interface proxy abnormality within a later lag window, and finally outputs the causal path score.

[0124] Specifically, the server compares the daily metabolic stress index with the metabolic threshold. When the metabolic stress index is greater than or equal to the metabolic threshold, the metabolic stress triggering condition is considered to be met for that day.

[0125] The server compares the next day's inflammation increment with the inflammation increment threshold. When the next day's inflammation increment is greater than or equal to the inflammation increment threshold, it is considered that the inflammation amplification occurred within the hysteresis window after metabolic stress.

[0126] The server takes the maximum value of the maternal-fetal interface standardized value within the lag window after the metabolic trigger day and compares it with the maternal-fetal interface threshold. If the maternal-fetal interface standardized value exceeds the maternal-fetal interface threshold on any day within the lag window, it is considered that the maternal-fetal interface functional imbalance proxy abnormality occurred after the inflammation amplification.

[0127] The server multiplies the three conditions to obtain a causal path score; when the three conditions of metabolic triggering, inflammation following, and maternal-fetal interface following are all true, the causal path score is 1, otherwise it is 0.

[0128] It should be noted that the metabolic threshold is determined by offline statistical analysis of the distribution of metabolic stress index during the baseline and risk-triggered periods, selecting quantiles that can distinguish between the stable state and the subsequent occurrence of inflammatory amplification events, and fixing them. The value range is usually [1,3]. The inflammatory increment threshold is determined by offline statistical modeling of the changes in the inflammatory index over consecutive natural days, selecting the smallest effective increment of the change in the proxy index associated with maternal-fetal interface functional imbalance within the lag time window as the fixed threshold, and the value range is usually [0.3,1.0]. The maternal-fetal interface threshold is determined by statistical analysis of the fluctuation range of the standardized values ​​of the maternal-fetal interface proxy index during the baseline period, selecting the standardized level that exceeds the upper limit of individual normal fluctuation and is associated with the occurrence of clinical adverse events, and fixing it. The value range is [0.8,2.0].

[0129] It should also be noted that by standardizing the effective monitoring records and constructing metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics, and introducing a pre-set lag time window constraint discrimination rule on the server side, the quantitative expression of the chain process of "metabolic stress - inflammation amplification - maternal-fetal interface functional imbalance" is realized, which transforms the pregnancy risk assessment from single-point indicator judgment to dynamic path discrimination based on causal order.

[0130] In this embodiment, in order to verify the benefits of constructing metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics from effective monitoring records, and to realize the quantitative expression of the chain process from metabolic stress to inflammation amplification to maternal-fetal interface functional imbalance through the sequential constraint discrimination rules of lag time windows, the pregnancy risk assessment is transformed from single-point indicator judgment to dynamic path discrimination based on causal order.

[0131] After quality control of monitoring records from a single subject over 28 consecutive days at four time windows daily (fasting, post-breakfast, afternoon, and evening), three types of dynamic characteristic curves were constructed to obtain data such as... Figure 5 The figure shows the results of the three types of dynamic characteristic curves and the lag time window constraint. Among them, the blue curve represents the metabolic stress feature, the orange curve represents the inflammation amplification feature, and the green curve represents the maternal-fetal interface functional imbalance proxy feature. The three dashed lines in the figure are the thresholds of each feature, which are calculated from the normal distribution of the control population, and the normal mean + 2σ is used to determine the triggering time when the feature enters the significantly abnormal interval. The light blue shaded area is the lag time window. The first segment corresponds to the allowable lag interval of metabolic stress → inflammation amplification (when the metabolic stress feature is triggered, the inflammation amplification feature is triggered within the allowable time window of 2 days starting from the 1st day thereafter). The second segment corresponds to the allowable lag interval of inflammation amplification → maternal-fetal interface (when the inflammation amplification feature is triggered, the maternal-fetal interface functional imbalance proxy feature is triggered within the allowable time window of 2 days starting from the 3rd day thereafter). Figure 5The results show that the three types of dynamic features can present a chain sequence of metabolic stress, inflammation amplification, and maternal-fetal interface functional imbalance in time series, which transforms risk assessment from single-point threshold judgment to dynamic path discrimination based on causal sequence constraints. Under the effect of quality control and retesting, the interference of occasional sampling anomalies and missing windows on feature construction is suppressed, making the matching of trigger time and lag window more stable and reducing the impact of uncertainty.

[0132] In this embodiment, the aim is to verify the beneficial effect of combining a quality scoring mechanism with retesting rules during the monitoring process to achieve objective control over the quality of monitoring data and reduce the uncertain impact of occasional sampling anomalies or missing windows on the dynamic feature construction and risk discrimination results.

[0133] A group of subjects underwent quality control within the same 28-day monitoring period and four time windows within the same day. Some subjects had built-in chain-like truth processes representing surrogate features of metabolic stress, amplified inflammation, and maternal-fetal interface dysfunction; the other group served as a normal control. Measurement noise, occasional sampling anomalies, and potential missing data were injected at the sampling level to make the data more closely resemble actual monitoring. During the experiment, parameters other than the quality score threshold were kept at their normal mean values. Quality score thresholds were scanned, and a quality scoring mechanism was activated at each threshold, combined with retest rules, to output three objective indicators: missing window rate, proportion of valid monitoring records, and F1 score for chain-like process identification under the lag time window constraint rules. The results were obtained as follows: Figure 6 The graph shown illustrates the impact of the quality scoring threshold on the performance of missing windows and chain discrimination. Figure 6 The study verified that the quality scoring mechanism and retesting rules can achieve objective control over the quality of monitoring data by adjusting the quality scoring threshold. While reducing the inclusion of occasional anomalies, it quantitatively presents the changing trends of missing windows and valid records. The chain recognition F1 shows that both too loose and too strict quality scoring thresholds are unfavorable, indicating that a balance is formed between anomaly suppression and missing window control through quality control and retesting. This reduces the uncertain impact of occasional sampling anomalies or missing windows on dynamic feature construction and risk discrimination results, and improves the stability of dynamic path discrimination based on causal order.

[0134] S4. Collect the patient's clinical events on the same day and perform consistency verification. Combine the causal path score and clinical events to generate a risk level and treatment record package, and push it to the medical staff's terminal.

[0135] Furthermore, the server sends a clinical event reporting prompt to the external communication terminal at a fixed time each day, such as a fixed time in the evening, prompting the patient to complete the clinical event reporting for the day.

[0136] The external communication terminal presents clinical event fields in a fixed enumeration manner on the prompt interface; among them, the clinical event fields include at least: vaginal bleeding level field, lower abdominal pain level field, and whether an emergency or hospitalization occurred field.

[0137] Among them, the vaginal bleeding level field has a four-level enumeration, including none, spotting, requiring a sanitary napkin but not soaked, and significantly increased bleeding or blood clots; the lower abdominal pain level field has a four-level enumeration, including none, mild discomfort, intermittent pain, and continuous pain or affecting activity; the emergency room or hospitalization field has a two-level enumeration, including no and yes.

[0138] After the patient completes the selection on the external communication terminal, the external communication terminal generates a clinical event data packet and uploads it to the server. The clinical event data packet includes at least the patient identification code, date, values ​​of three fields (clinical event), terminal identifier, and reporting timestamp.

[0139] After receiving the clinical event data packet, the server writes the receiving timestamp of the clinical event on the server side and enters it into the clinical event table, so that it can be associated with the causal path score of the same day on the same date dimension.

[0140] Furthermore, the server retrieves the values ​​of the three fields of the clinical event for the day and executes the consistency verification rules.

[0141] Among them, the consistency verification rules include that when the "whether an emergency or hospitalization has occurred" field is "yes", at least one of the "vaginal bleeding level" field and "lower abdominal pain level" field should reach moderate or above enumeration (i.e., needing a sanitary napkin but not soaking it, significantly increased bleeding or blood clots, intermittent pain, and any of the following that are continuous or affect activity).

[0142] If the "whether an emergency room or hospitalization occurred" field is "yes", and both the bleeding level and abdominal pain level are below moderate, then the condition is considered inconsistent; otherwise, the condition is considered consistent.

[0143] When the consistency verification rule determines that there is an inconsistency, the server sends a verification failure message to the external communication terminal, requiring the patient to reselect the bleeding level and abdominal pain level; the external communication terminal forces the patient to resubmit before the event is entered into the database for the day.

[0144] When the consistency verification rule determines that the clinical events of that day are consistent, the server marks them as having passed the verification and proceeds to the risk level merging calculation.

[0145] Furthermore, the server obtains the causal path score for the day, records it as the causal path score for the day, and obtains the values ​​of the three fields of the clinical events that have passed the consistency verification for the day.

[0146] If the causal path score is unavailable on a given day, such as when a missing window causes the causal path score to be invalid and there is insufficient labeled data, the server will set the causal path score to 0 and mark the data as insufficient.

[0147] Risk level is recorded as The value is limited to a four-level enumeration, including level 0, level 1, level 2, and level 3, and the risk level is obtained according to the risk level generation rules A1-A4: A1. When the causal path score is 0 and both the bleeding level and the abdominal pain level are mild or below, the risk level is 0.

[0148] A2. When the causal path score is 0 but any abnormal triggering sign appears (the server records any one of the three conditions: metabolic triggering, inflammation following, and maternal-fetal interface following as true), the risk level is level 1.

[0149] A3. When the causal pathway score is 1 and the clinical event does not reach the moderate level, the risk level is 2.

[0150] A4. When the causal pathway score is 1 and the clinical event reaches a moderate or higher level, or an emergency or hospitalization occurs, the risk level is 3.

[0151] The server writes the daily risk level into the risk level table and archives the key fields on which the generation is based.

[0152] Furthermore, the server generates a treatment record package for each natural day; the treatment record package includes at least the patient identification code, gestational week, date, risk level of the day, causal path score and three-segment condition markers of the day, metabolic stress index, inflammation index, inflammation increment, maternal-fetal interface standardized value, three fields of clinical events and consistency verification results, and data integrity markers (missing window, retest substitution situation).

[0153] The server pushes the treatment record package to the medical personnel's terminal; the medical personnel's terminal presents it in the form of a form, which includes at least a risk level prompt area, a list of key features, a list of clinical events, and a treatment entry area.

[0154] Medical personnel enter treatment details in the treatment entry area. The treatment details use a combination of fixed enumeration fields and restricted text fields. The fixed enumeration fields include at least the treatment type field, treatment timestamp field, treatment person identifier field, and treatment description field.

[0155] Among them, the treatment type field is limited to four levels of enumeration, including no adjustment, medication adjustment, additional examination, and recommendation to seek medical treatment or hospitalization; the treatment timestamp field is automatically generated by the terminal; the treatment person identifier field is generated by binding the terminal account; the treatment description field has a limited number of characters and cannot be empty.

[0156] After medical personnel submit the treatment details via their terminal, the server receives and writes them into the treatment record table. The server then merges and solidifies the treatment details and the daily treatment record package according to the date and patient identification code, forming a complete treatment file for the day.

[0157] It should also be noted that by combining a quality scoring mechanism with fixed retesting rules during the monitoring process, the sampling channel status and readout stability are quantitatively evaluated and retesting is automatically triggered, thus achieving objective control over the quality of monitoring data and reducing the uncertain impact of occasional sampling anomalies or missing windows on the dynamic feature construction and risk judgment results.

[0158] This embodiment also provides a multi-parameter dynamic monitoring system for patients undergoing pregnancy maintenance, including: a record-keeping and calibration module, a sampling quality control module, a causal analysis module, and a risk management module.

[0159] The monitoring system comprises several modules: **File Establishment and Calibration Module:** This module establishes files for monitored subjects and deploys the data collection carrier. It binds patient identities to external communication terminals and servers, writes daily monitoring time windows and retesting rules, and establishes individualized equivalent conversion relationships through fasting serum control collection and interstitial fluid sampling to obtain individual baseline statistics for each monitoring indicator. **Sampling Quality Control Module:** This module performs quantitative sampling of interstitial fluid at fixed time windows, obtaining detection values ​​for glucose, lipid profiles, inflammatory factors, and maternal-fetal interface mediator indicators. It calculates quality scores and performs retesting according to retesting rules when quality is insufficient, generating monitoring data. **Causal Analysis Module:** This module standardizes effective monitoring data and constructs metabolic stress characteristics, inflammatory amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics. It performs constraint discrimination based on lag time windows to obtain causal path scores. **Risk Management Module:** This module collects patients' daily clinical events and performs consistency verification. It merges causal path scores and clinical events to generate risk levels and management record packages, which are then pushed to medical personnel terminals.

[0160] In summary, this invention constructs metabolic stress features, inflammation amplification features, and maternal-fetal interface functional imbalance proxy features from effective monitoring records. By using a lag time window-based constraint discrimination rule, it achieves a quantitative expression of the chain process from metabolic stress to inflammation amplification and then to maternal-fetal interface functional imbalance. This transforms pregnancy risk assessment from single-point indicator judgment to dynamic path discrimination based on causal order. By combining a quality scoring mechanism and a retest rule during the monitoring process, it achieves objective control over the quality of monitoring data, reducing the uncertain impact of occasional sampling anomalies or missing windows on the dynamic feature construction and risk discrimination results.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for multi-parameter dynamic monitoring in patients undergoing pregnancy maintenance, characterized in that: include, Establish files for monitoring targets and complete the deployment of data collection devices; The patient's identity is bound to the in vitro communication terminal and server, written into the daily monitoring time window and retest rules, and an individualized equivalent conversion relationship is established through serum control collection and interstitial fluid sampling during the fasting window to obtain the individual baseline statistics of each monitoring indicator. Quantitative sampling of interstitial fluid was performed at fixed time windows to obtain the detection values ​​of glucose, blood lipid profile, inflammatory factors and maternal-fetal interface mediator indicators, calculate the quality score, and retest according to the retest rules when the quality was insufficient to generate monitoring data. The effective monitoring data were standardized and metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics were constructed. Constraint discrimination was performed based on the lag time window to obtain causal path scores. The system collects patients' clinical events on the same day and performs consistency verification. It then merges the causal path score and clinical events to generate a risk level and treatment record package, which is then pushed to the medical staff's terminal.

2. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 1, characterized in that: The establishment of monitoring object files and the deployment of data collection carriers include, On the server side, a unique patient identification code is generated for each patient, and a monitoring object file corresponding to the patient's basic information is established. Bind the patient's identification code to the terminal identifier of the external communication terminal; The patient's identification code is bound to the carrier identifier of the data collection carrier, and a one-to-one correspondence between the data collection carrier and the monitored patient is established. Write the daily monitoring time window table and retest rules into the data collection carrier so that the data collection carrier automatically enters the sampling and testing process when the corresponding time window arrives.

3. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 2, characterized in that: The establishment of individualized equivalent conversion relationships includes... Serum samples were collected from patients during the fasting monitoring time window, and the corresponding serum test values ​​were obtained. Within the same time window as serum collection, the collection carrier performs interstitial fluid sampling to obtain interstitial fluid detection values; Serum test values ​​and interstitial fluid test values ​​are paired according to timestamps to form paired data sets; Based on paired data from at least two different dates, individualized equivalent conversion parameters for each monitoring indicator are calculated and generated.

4. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 3, characterized in that: The individual baseline statistics for each monitoring indicator include, After establishing the individualized equivalent conversion relationship, a continuous number of natural days were set as the baseline data collection period; During the baseline acquisition period, interstitial fluid samples were collected from the collection medium according to the daily monitoring time window. The collected equivalent serum values ​​were screened using a quality scoring system. The individual baseline mean and individual baseline standard deviation of each monitoring indicator were calculated based on the screened data.

5. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 4, characterized in that: The calculation of the quality score includes, When the fixed monitoring time window arrives, the collection carrier performs quantitative interstitial fluid sampling; During the sampling process, the impedance characteristic value and sampling drive characteristic value of the sampling channel are acquired; The impedance characteristic value and the sampled drive characteristic value are compared with the corresponding baseline statistics to calculate the impedance normalization deviation and the drive normalization deviation. A quality score characterizing the sampling channel state is generated based on impedance normalization deviation and drive normalization deviation.

6. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 5, characterized in that: The retesting according to the retesting rules includes... Compare the quality score of the current monitoring time window with the quality score threshold; When the quality score is lower than the quality score threshold, a retest trigger instruction is generated; Within the delay period following the end of the current monitoring time window, the data acquisition carrier will perform re-testing and sampling. The data obtained from the retest sampling will be used as the valid monitoring data for the current monitoring time window.

7. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 6, characterized in that: The standardization of effective monitoring data and the construction of metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics include... Obtain the individual baseline mean and individual baseline standard deviation for each monitoring indicator; Standardize the equivalent serum values ​​in the valid monitoring data; Metabolic stress characteristics were constructed based on standardized glucose-related data; Inflammation amplification features were constructed based on standardized inflammatory factor data. We construct the functional imbalance proxy characteristics of the maternal-fetal interface using mediator proxy indicators.

8. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 7, characterized in that: The constraint discrimination based on the lag time window includes... Determine whether metabolic stress characteristics meet metabolic triggering conditions within the time window; Determine whether the inflammatory amplification features meet the inflammatory triggering conditions within the hysteresis window after metabolic triggering. Within a further lag time window, determine whether the maternal-fetal interface functional imbalance proxy characteristics meet the abnormal conditions. When the decision constraints are met sequentially in chronological order, a feature causal path score is generated.

9. The multi-parameter dynamic monitoring method for patients requiring pregnancy maintenance as described in claim 8, characterized in that: The risk level and response record package includes, Collect the patient's clinical event information for the day and upload it to the server; Perform consistency checks on clinical event information and obtain the check results; The validated clinical event information and causal path scores are combined and processed to generate a risk level; The risk level, causal path score, and corresponding monitoring data are packaged into a treatment record package and pushed to the medical personnel's terminal.

10. A multi-parameter dynamic monitoring system for patients undergoing pregnancy maintenance, based on the multi-parameter dynamic monitoring method for patients undergoing pregnancy maintenance according to any one of claims 1 to 9, characterized in that: This includes a filing and calibration module, a sampling quality control module, a causal analysis module, and a risk management module; The file establishment and calibration module is responsible for establishing monitoring subject files and completing the deployment of collection carriers. It binds patient identity with in vitro communication terminals and servers, writes daily monitoring time windows and retest rules, and establishes individualized equivalent conversion relationships through serum control collection and interstitial fluid sampling during the fasting window to obtain individual baseline statistics for each monitoring indicator. The sampling quality control module is responsible for quantitatively sampling interstitial fluid at fixed time windows, obtaining detection values ​​of glucose, blood lipid profile, inflammatory factors and maternal-fetal interface mediator indicators, calculating quality scores, and retesting according to retesting rules when the quality is insufficient, and generating monitoring data. The causal analysis module is responsible for standardizing the effective monitoring data and constructing metabolic stress characteristics, inflammation amplification characteristics, and maternal-fetal interface functional imbalance proxy characteristics, and performing constraint discrimination based on the lag time window to obtain the causal path score. The risk management module is responsible for collecting the patient's clinical events on the same day and performing consistency verification. It combines the causal path score and clinical events to generate a risk level and management record package, which is then pushed to the medical personnel's terminal.

Images