Movatterモバイル変換


[0]ホーム

URL:


CN120727304A - Intelligent prediction method for gestational diabetes risk based on multimodal data fusion - Google Patents

Intelligent prediction method for gestational diabetes risk based on multimodal data fusion

Info

Publication number
CN120727304A
CN120727304ACN202511250827.2ACN202511250827ACN120727304ACN 120727304 ACN120727304 ACN 120727304ACN 202511250827 ACN202511250827 ACN 202511250827ACN 120727304 ACN120727304 ACN 120727304A
Authority
CN
China
Prior art keywords
risk
pregnancy
detection
time
time period
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202511250827.2A
Other languages
Chinese (zh)
Inventor
吴旭
毛雅婷
朱莹莹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nantong Maternity and Child Health Care Hospital
Original Assignee
Nantong Maternity and Child Health Care Hospital
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nantong Maternity and Child Health Care HospitalfiledCriticalNantong Maternity and Child Health Care Hospital
Priority to CN202511250827.2ApriorityCriticalpatent/CN120727304A/en
Publication of CN120727304ApublicationCriticalpatent/CN120727304A/en
Pendinglegal-statusCriticalCurrent

Links

Landscapes

Abstract

Translated fromChinese

本发明公开了基于多模态数据融合的妊娠期糖尿病风险智能预测方法,具体涉及数据处理技术领域,包括S1:设置预测窗口、S2:风险检测数据获取、S3:多模态时间序列矩阵构建、S4:风险分析、S5:构建妊娠风险变化曲线、S6:可视化展示。本发明通过生理风险指标和生化风险指标进行检测,并基于风险检测数据进行多模态时间序列矩阵构建,满足数据采集准确性,形成时间段与多指标的矩阵结构,由此将妊娠风险与检测时间段进行关联分析,从中构建妊娠风险变化曲线,直观展示风险随孕期的动态演变,并通过拐点检测可精确定位风险发生时间,有助于风险发生时间的重点检测,为后续风险变化趋势、波动幅度奠定结构化基础。

The present invention discloses an intelligent prediction method for gestational diabetes risk based on multimodal data fusion, specifically relating to the field of data processing technology, including S1: setting a prediction window, S2: acquiring risk detection data, S3: constructing a multimodal time series matrix, S4: risk analysis, S5: constructing a pregnancy risk change curve, and S6: visual display. The present invention uses physiological risk indicators and biochemical risk indicators for detection, and constructs a multimodal time series matrix based on the risk detection data to meet data collection accuracy, forming a matrix structure of time periods and multiple indicators. This allows correlation analysis between pregnancy risk and the detection time period, from which a pregnancy risk change curve is constructed, intuitively displaying the dynamic evolution of risk over the gestational period, and accurately locating the time of risk occurrence through inflection point detection, which facilitates focused detection of the risk occurrence time and lays a structured foundation for subsequent risk change trends and fluctuation amplitudes.

Description

Gestational diabetes risk intelligent prediction method based on multi-mode data fusion
Technical Field
The invention relates to the technical field of data processing, in particular to an intelligent gestational diabetes risk prediction method based on multi-mode data fusion.
Background
Gestational diabetes is diabetes which is first discovered or developed in gestation, and has great harm to a mother, for example, the pregnant woman may have embryo dysplasia or death, gestational hypertension diseases, infection, amniotic fluid excess and other complications, a fetus may have huge fetus, limited fetal growth, abortion and premature delivery, fetal distress, fetal death, intrauterine death and other conditions, and meanwhile, the probability of postpartum type 2 diabetes of the gestational diabetes patient is increased, and the incidence rate of long-term cardiovascular system diseases is also high.
The existing gestational diabetes risk prediction method is mainly based on static risk factors such as age, BMI, family history and the like to predict, an empirical or semi-quantitative evaluation system is constructed to judge the occurrence probability of gestational diabetes, but the method still has some defects in actual use, firstly, the existing risk factor evaluation prediction method is mainly based on static risk factors such as age, BMI, family history and the like to evaluate, the prediction accuracy is low, dynamic risk evaluation cannot be performed, individual differences cannot be fully considered, multiple data evaluation is needed based on the situation, but the existing response prediction accuracy is mostly limited to single biochemical index monitoring, the fasting blood glucose and postprandial blood glucose are mainly monitored to predict, the monitoring result is easily influenced by various factors, the overall state of the system is lack of evaluation, early warning time is late, and early intervention is difficult to realize;
Secondly, the existing method mainly depends on disposable data before pregnancy or during early pregnancy when the gestational diabetes is detected, dynamic change indexes in the pregnancy cannot be tracked, the pregnancy is a process of severe physiological state change, particularly, the large secretion of placenta hormones in late pregnancy leads to the aggravation of insulin resistance, in this case, dynamic risk detection is required during the gestation, and the risk inflection point in the dynamic process cannot be captured.
Disclosure of Invention
In view of this, the embodiment of the invention provides a gestational diabetes risk intelligent prediction method based on multi-mode data fusion, which performs physiological risk analysis and biochemical risk analysis respectively in the gestational diabetes detection process, performs gestational risk index calculation, maximally guarantees the combination of data detection and risk prediction, and performs the construction of a gestational risk change curve after performing gestational risk index calculation, thereby determining the risk occurrence time in the gestational period, and effectively solving the problems raised in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions:
S1, setting a prediction window, wherein the prediction window comprises an early pregnancy window, a middle pregnancy window and a late pregnancy window, and the early pregnancy window, the middle pregnancy window and the late pregnancy window are respectively marked as T1, T2 and T3, and the time of the prediction window is set according to the medical diagnosis and treatment standard of diabetes;
s2, acquiring risk detection data, namely extracting current pregnancy detection time from a prediction window, wherein the pregnancy detection time is in units of gestational weeks and is marked as 1,2 according to the gestational weeks, wherein the j is m, and j represents the j-th pregnancy detection time, so that risk detection data corresponding to each pregnancy detection time are acquired, the risk detection data comprise physiological risk indexes and biochemical risk indexes, and further pregnancy detection time periods corresponding to the risk detection data are acquired;
S3, constructing a multi-mode time sequence matrix, namely generating a time sequence corresponding to the risk detection data according to a preset sequence based on a pregnancy detection time period corresponding to the risk detection data, and constructing the multi-mode time sequence matrix according to the time sequence;
S4, risk analysis, namely respectively carrying out physiological risk analysis and biochemical risk analysis on the multi-mode time sequence matrix corresponding to each pregnancy detection time period, and analyzing pregnancy risk indexes of each pregnancy detection time period according to the physiological risk analysis and the biochemical risk analysis;
S5, constructing a pregnancy risk variation curve, namely performing association analysis on pregnancy risk indexes corresponding to each pregnancy detection time period and the pregnancy detection time period, so as to construct the pregnancy risk variation curve, and further determining risk occurrence time;
And S6, visually displaying, namely sending the risk occurrence time to a user terminal according to a preset summarization mode, wherein the preset summarization mode comprises report summarization, picture summarization and chart summarization, and a manager carries out key detection of gestational diabetes on the risk occurrence time through summarization contents, and the summarization contents comprise a multi-mode time sequence matrix, a gestational risk condition and a gestational risk change curve.
The invention has the technical effects and advantages that:
1. According to the invention, the wearable equipment and the pregnancy test terminal are used for detecting physiological risk indexes and biochemical risk indexes of gestational diabetes mellitus, and the multi-mode time sequence matrix is constructed based on risk detection data, so that multi-dimensional data acquisition is performed, the two-way influence assessment of the gestation period on the risks is facilitated, comprehensive and reliable data support is provided for determining the occurrence time of the risks, the unilaterality of a single index is avoided, on the other hand, the gestation test time is marked according to the gestation Zhou Shunxu, a time sequence is formed, the data acquisition accuracy can be met to the greatest extent, and a structural basis is laid for the follow-up risk variation trend and fluctuation range;
2. According to the invention, the physiological risk index and the biochemical risk index are strictly aligned on the time axis through the pregnancy detection time period, the problem of inconsistent detection frequencies of different modes is solved, and a matrix structure of the time period and multiple indexes is formed, so that the pregnancy risk and the detection time period are subjected to correlation analysis, a pregnancy risk change curve is constructed, the dynamic evolution of the risk along with the pregnancy period is visually displayed, the risk occurrence time can be accurately positioned through inflection point detection, the key detection of the risk occurrence time is facilitated, the gestational diabetes risk is reduced to the greatest extent, and a powerful detection reference is provided for the diabetes judgment of doctors.
Drawings
Fig. 1 is a schematic diagram of the overall structure of the present invention.
Fig. 2 is a flow chart of risk analysis according to the present invention.
Fig. 3 is a flowchart of the risk occurrence time determining step of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The intelligent prediction method for gestational diabetes risk based on multi-mode data fusion as shown in the accompanying figures 1-3 comprises a wearable device, a labor detection terminal, a user terminal and a control center, wherein the wearable device and the labor detection terminal are connected with the user terminal through Bluetooth.
In a more specific application of the invention, the wearable device is used for monitoring physical sign states, in particular weight, height, resting heart rate and electrocardiosignals, wherein the wearable device can be an intelligent bracelet and is used for monitoring in combination with a heart rate monitoring device, and the heart rate monitoring device can be an electrocardio monitor for continuously monitoring the resting heart rate and the electrocardiosignals, for example, the weight and the height of a pregnant woman are continuously recorded through the intelligent bracelet.
The obstetric examination terminal is used for collecting insulin conditions of the pregnant women, and specifically can be an RFID reader, an insulin detector and an obstetric examination database, wherein the insulin conditions are that fasting blood glucose and insulin concentration of the pregnant women are monitored, monitoring results are stored in the obstetric examination database, and the insulin conditions of the pregnant women are automatically identified through the RFID reader.
The control center is used for analyzing and controlling related parameters based on the data of the equipment in the gestational diabetes detection process.
The specific embodiment of the invention comprises the following steps:
S1, setting a prediction window, wherein the prediction window comprises an early pregnancy window, a middle pregnancy window and a late pregnancy window, and is respectively marked as T1, T2 and T3, the time of the prediction window is set according to the medical diagnosis and treatment standard of diabetes, and the time of the prediction window is exemplified by 6-12 weeks of pregnancy, 24-28 weeks of pregnancy, and 36-40 weeks of pregnancy as T2.
It should be explained that the purpose of the setting of the prediction window is to scientifically define the key time range of data acquisition based on large physiological change of pregnancy, ensure timeliness, pertinence and standardization of multi-mode data, ensure that the early pregnancy window is the acquisition time of first time of pregnancy detection and is used for evaluating the risk factors of pre-pregnancy diabetes, the middle pregnancy window is the middle pregnancy, at the moment, insulin resistance obviously increases due to the fact that insulin resistance secreted by placenta reaches a peak value at the stage, therefore, the stage is the high-stage of gestational diabetes, the late pregnancy window is increased in risk of gestational diabetes due to fetal development, the pregnancy physiological index is changed along with pregnancy Zhou Dongtai, the detection of the non-key window is possibly limited in value due to gentle change, and the focus key sensitive window can highlight the index shock related to gestational diabetes.
And S2, acquiring risk detection data, namely extracting current pregnancy detection time from a prediction window, wherein the pregnancy detection time is in units of gestational weeks and is marked as 1,2 according to the gestational weeks, wherein the j is m, and j represents the j-th pregnancy detection time, so that risk detection data corresponding to each pregnancy detection time are acquired, the risk detection data comprise physiological risk indexes and biochemical risk indexes, and further a pregnancy detection time period corresponding to the risk detection data is acquired.
In this embodiment, it should be specifically described that the specific process of obtaining the physiological risk indicator includes:
Extracting electrocardiosignals through wearable equipment, and identifying R wave peak values according to the electrocardiosignals, so as to respectively calculate adjacent R wave intervals;
Calculating R wave root mean square based on adjacent R wave interval difference values, and further calculating all R wave interval standard deviations through the R wave root mean square;
Extracting weight and height, and obtaining a body weight index through a body weight index calculation formula, and further obtaining a weight increase rate through comparing the body weight index of the adjacent pregnancy detection time with the time interval of the adjacent pregnancy detection time;
The standard deviation of the R wave interval and the weight gain rate are integrated as physiological risk indexes.
It should be explained that the standard deviation of R wave interval and the weight gain rate are the core physiological indexes reflecting the functions and metabolic loads of the maternal autonomic nerves, the standard deviation of R wave interval is the standard deviation of heartbeat period, the greater the standard deviation of R wave interval is, the more balanced the activity of the autonomic nervous system is, the heart regulating ability is strong, the insulin secretion and the glucose metabolism are more stable, the smaller the standard deviation of R wave interval is, the autonomic nerves are disregulated, the insulin resistance is aggravated, the fasting blood sugar is increased, the risk of gestational diabetes is increased, the weight gain rate is the change of body mass index of the adjacent pregnancy detection time divided by the time interval, and when the rate is too fast, the fat cells are excessively proliferated, and the insulin signal path is directly inhibited, thereby increasing the incidence risk.
In this embodiment, it should be specifically described that the specific biochemical risk indicator acquiring process includes:
Identifying insulin condition of pregnant women through the labor detection terminal, and extracting fasting blood glucose and insulin concentration of the pregnant women;
Abnormal state marking is carried out on the fasting blood glucose of the pregnant women, and the abnormal state marking is specifically classified by a diagnosis standard:
,
Wherein, theMarking the result of the abnormal state of the pregnant woman at each pregnancy detection time,Indicating the fasting blood glucose of the pregnant woman at each pregnancy test time,AndRespectively obtaining fasting blood glucose standard values of various abnormal states according to the medical diagnosis and treatment standard of diabetes;
adding the abnormal state marking results of the pregnancy detection times, if the added results are not equal to 0, judging that the abnormal condition exists in the fasting blood glucose corresponding to the pregnancy detection times, otherwise, judging that the normal fasting blood glucose corresponding to the pregnancy detection times exists;
extracting abnormal fasting blood sugar corresponding to each pregnancy detection time, subtracting the fasting blood sugar corresponding to the adjacent pregnancy detection time in sequence, and obtaining an absolute value, and comparing the absolute value with the maximum value of the fasting blood sugar and the absolute value, thereby obtaining a blood sugar fluctuation coefficient;
Comparing the extracted insulin concentrations, subtracting insulin concentrations corresponding to adjacent pregnancy detection times respectively, and comparing the extracted insulin concentrations with adjacent pregnancy detection time intervals, thereby obtaining insulin secretion rates;
The blood glucose fluctuation coefficient and the insulin secretion rate are integrated as biochemical risk indexes.
It should be explained that, the blood glucose fluctuation coefficient and the insulin secretion rate are used as biochemical risk indexes, because the blood glucose fluctuation coefficient is the fluctuation range of blood glucose along with the gestation week, the insulin secretion rate is the dynamic compensation of insulin secretion, the higher the blood glucose fluctuation coefficient is, the higher the insulin secretion rate is, the short-term regulation capacity of insulin on blood glucose is reduced, which indicates that the higher the secretion compensation insulin resistance is increased by cells, the higher the biochemical risk is, and meanwhile, the change of the biochemical risk index in gestation period can be reflected by carrying out adjacent gestation detection time treatment on fasting blood glucose and insulin concentration corresponding to each gestation detection time, rather than carrying out single-time-point state detection to cause the condition that the blood glucose fluctuation is large but not overdetermined.
It should be further noted that, because the detection frequencies of the physiological risk indicator and the biochemical risk indicator are different, by extracting the current pregnancy detection time from the prediction window, different mode data can be forcedly aligned to the same time range, so as to avoid fusion errors caused by time dislocation, and meanwhile, after the physiological risk indicator and the biochemical risk indicator are acquired, different pregnancy detection times are compared to generate new detection times, which are respectively marked as 1,2, the first, the second, the third, the fourth, the fifth, the seventh, the eighth, the ninth, and the eighth, wherein j-1 represents a j-1 th pregnancy detection time period.
And S3, constructing a multi-mode time sequence matrix, namely generating a time sequence corresponding to the risk detection data according to a preset sequence based on a pregnancy detection time period corresponding to the risk detection data, and constructing the multi-mode time sequence matrix according to the time sequence.
In this embodiment, it is to be specifically described that the time sequence corresponding to the risk detection data includes a physiological risk time sequence and a biochemical risk time sequence, where the physiological risk time sequence arranges physiological risk indexes according to a preset sequence, and specifically indicates that: Wherein each time period corresponds to a physiological risk feature comprising an R-wave interval standard deviation and a weight gain rate, and the biochemical risk time sequence is generated by the same method, and is specifically expressed as follows: And further, aligning different modal data through pregnancy detection time periods, filling the missing detection time periods by adopting time sequence interpolation, and constructing a multi-modal time sequence matrix based on the physiological risk time sequence and the biochemical risk time sequence, wherein the multi-modal time sequence matrix is specifically expressed as follows:
,
Wherein, theRepresenting a multi-modal time series matrix, each row corresponding to a pregnancy detection period, each column corresponding to a modal feature,Representing the physiological risk characteristics corresponding to the m-1 th pregnancy detection period in the physiological risk time sequence,Representing the biochemical risk characteristics corresponding to the m-1 pregnancy detection time period in the biochemical risk time sequence.
It should be explained that, the time-series interpolation is a data processing method for solving the problem of data missing, and the accuracy of data analysis can be improved by the time-series interpolation, so as to improve the prediction result.
It should be further noted that, a time sequence corresponding to the risk detection data is generated according to a preset sequence, where the preset sequence is a sequence of pregnancy detection time periods corresponding to the risk detection data.
And S4, risk analysis, namely respectively carrying out physiological risk analysis and biochemical risk analysis on the multi-mode time sequence matrix corresponding to each pregnancy detection time period, and analyzing the pregnancy risk index of each pregnancy detection time period according to the physiological risk analysis and the biochemical risk analysis.
In the embodiment, the physiological risk analysis is specifically described as follows, physiological risk characteristics are extracted based on a multi-mode time sequence matrix corresponding to each pregnancy detection time period, wherein the physiological risk characteristics comprise an R wave interval standard deviation and a weight growth rate;
Comparing and calculating the standard deviation of the R wave interval and the weight increase rate to obtain the physiological abnormality risk coefficient corresponding to each pregnancy detection time period, wherein the physiological abnormality risk coefficient is specifically expressed as:
,
Wherein, theRepresents the physiological abnormality risk coefficient corresponding to each pregnancy detection time,Represents the standard deviation of the R-wave interval,Represents the standard deviation of the ideal R-wave interval,Indicating the rate of weight gain.
It should be explained that the physiological abnormality risk condition is greatly influenced by the heart rate variability, and is also influenced by the obesity degree of pregnant women, the heart rate variability is described by using the R wave interval standard deviation, if the R wave interval standard deviation is close to the ideal R wave interval standard deviation, the heart rate variability is large in the pregnancy detection time period, the obesity degree is described by using the weight increase rate, and if the weight increase rate is large, the physiological abnormality risk coefficient corresponding to the pregnancy detection time period is large.
The biochemical risk analysis is specifically described as extracting biochemical risk characteristics including blood sugar fluctuation coefficient and insulin secretion rate based on a multi-mode time sequence matrix corresponding to each pregnancy detection time period;
Comparing and calculating the blood sugar fluctuation coefficient and the insulin secretion rate to obtain a biochemical abnormality risk coefficient corresponding to each pregnancy detection time period, wherein the biochemical abnormality risk coefficient is specifically expressed as:
,
Wherein, theRepresenting the corresponding biochemical abnormality risk coefficient of each pregnancy detection time period,Respectively represent the blood sugar fluctuation coefficient and the insulin secretion rate in biochemical risk characteristics,The corresponding standard blood sugar fluctuation coefficient and insulin secretion rate are respectively expressed, and are known from the medical diagnosis and treatment standard of diabetes.
It should be explained that, the biochemical abnormality risk condition is analyzed through the blood sugar fluctuation coefficient and the insulin secretion rate, when the insulin secretion rate is slower, the blood sugar regulation capability is poorer in the pregnancy detection time period, at the moment, the blood sugar fluctuation is larger, and the larger the blood sugar fluctuation coefficient is, the larger the biochemical abnormality risk coefficient corresponding to the pregnancy detection time period is.
It should be further noted that, the pregnancy risk index is added with weights based on the physiological abnormal risk coefficient and the biochemical abnormal risk coefficient corresponding to each pregnancy detection period, specifically expressed as:
,
Wherein, theRepresents the pregnancy risk index corresponding to each pregnancy test period,AndRespectively represent the physiological abnormality risk coefficient and the biochemical abnormality risk coefficient of the m-1 pregnancy detection time period,AndThe weight coefficients corresponding to the physiological abnormal risk coefficient and the biochemical abnormal risk coefficient are respectively represented, and are set through the prediction windows corresponding to the pregnancy detection time periods, and if the prediction window corresponding to the pregnancy detection time period is T1, the weight coefficients corresponding to the physiological abnormal risk coefficient and the biochemical abnormal risk coefficient in the pregnancy detection time period are respectively 0.7 and 0.3.
And S5, constructing a pregnancy risk variation curve, namely performing association analysis on pregnancy risk indexes corresponding to each pregnancy detection time period and the pregnancy detection time period, thereby constructing the pregnancy risk variation curve and further determining risk occurrence time.
In the embodiment, a specific construction method of a pregnancy risk variation curve needs to be specifically explained, namely, a two-dimensional coordinate system is constructed by taking pregnancy detection time as a horizontal axis and taking a pregnancy risk index as a vertical axis, a plurality of points are marked in the constructed two-dimensional coordinate system aiming at the pregnancy detection time and the pregnancy risk index which are detected each time in each pregnancy detection time period to form the pregnancy risk variation curve in the pregnancy detection time period, and then the risk occurrence time is determined, wherein the specific determination is that inflection points are marked on the pregnancy risk variation curve in the pregnancy detection time period;
Dividing the pregnancy risk change curve into an ascending section, a descending section and a flat section based on marked inflection points, respectively counting the duty ratio of the ascending section and the amplitude of the ascending section, and sequentially numbering the ascending sections corresponding to the pregnancy risk change curve from front to back according to the positions of the ascending sections;
calculating the risk occurrence probability of each rising segment based on the rising segment duty ratio and the rising segment amplitude, and specifically associating the rising segment duty ratio and the rising segment amplitude through a probability calculation formula, wherein the probability calculation formula specifically comprises the following steps:
,
Wherein, theRepresenting the probability of risk occurrence for each rising segment,AndThe rising segment duty cycle and the rising segment amplitude are represented respectively,AndRespectively representing risk fitting parameters, and estimating the fitting parameters by a maximum likelihood method;
and comparing the risk occurrence probabilities of the pregnancy risk change curves corresponding to the ascending segments in the pregnancy detection time periods, and screening the pregnancy detection time period corresponding to the maximum risk occurrence probability from the risk occurrence probabilities as risk occurrence time.
It should be further noted that the rising segment duty ratio and the rising segment amplitude are counted as follows:
Extracting pregnancy detection time of the rising section through a pregnancy risk change curve, and calculating to obtain the rising section duty ratio compared with the total pregnancy detection time;
And acquiring the slope of each ascending segment in the pregnancy risk change curve, and carrying out mean value calculation to obtain the amplitude of the ascending segment.
It should be understood that the basis for taking the rising period duty ratio and the rising period amplitude of the pregnancy risk variation curve in the pregnancy detection time period as the average risk variation probability is that the indexes comprehensively consider the variation trend of different stages in the pregnancy detection time, but not just pay attention to a certain time point, and the overall dynamic variation of the pregnancy risk can be more comprehensively estimated by considering the rising period duty ratio and the rising period amplitude, so that the gestational diabetes risk can be more accurately predicted.
And S6, visually displaying, namely sending the risk occurrence time to a user terminal according to a preset summarization mode, wherein the preset summarization mode comprises report summarization, picture summarization and chart summarization, and a manager carries out key detection of gestational diabetes on the risk occurrence time through summarization contents, and the summarization contents comprise a multi-mode time sequence matrix, a gestational risk condition and a gestational risk change curve.
In the drawings of the disclosed embodiments, only the structures related to the embodiments of the present disclosure are referred to, and other structures can refer to the common design, so that the same embodiment and different embodiments of the present disclosure can be combined with each other without conflict;
finally, the foregoing description of the preferred embodiment of the invention is provided for the purpose of illustration only, and is not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

Claims (9)

Translated fromChinese
1.基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于,包括:1. An intelligent prediction method for gestational diabetes risk based on multimodal data fusion, characterized by comprising:S1:设置预测窗口,其中预测窗口包括早孕窗口、中孕窗口以及晚孕窗口,并将其分别标记为T1、T2以及T3;S1: Set the prediction window, which includes the early pregnancy window, the middle pregnancy window and the late pregnancy window, and mark them as T1, T2 and T3 respectively;S2:风险检测数据获取:从预测窗口中提取当前妊娠检测时间,分别标记为1,2,…,j,…,m,其中j表示第j妊娠检测时间,由此采集各妊娠检测时间对应的风险检测数据,其中风险检测数据包括生理风险指标、生化风险指标,进而获取风险检测数据对应的妊娠检测时间段;S2: Acquisition of risk detection data: Extract the current pregnancy detection time from the prediction window and mark it as 1, 2, ..., j, ..., m, where j represents the jth pregnancy detection time. The risk detection data corresponding to each pregnancy detection time is collected, where the risk detection data includes physiological risk indicators and biochemical risk indicators, and the pregnancy detection time period corresponding to the risk detection data is obtained.S3:多模态时间序列矩阵构建:基于风险检测数据对应的妊娠检测时间段按照预设的顺序生成风险检测数据对应的时间序列,进而根据时间序列构建多模态时间序列矩阵;S3: Multimodal time series matrix construction: Generate the time series corresponding to the risk detection data according to the preset order based on the pregnancy detection time period corresponding to the risk detection data, and then construct a multimodal time series matrix based on the time series;S4:风险分析:对各妊娠检测时间段对应的多模态时间序列矩阵分别进行生理风险分析、生化风险分析,并据此分析各妊娠检测时间段的妊娠风险指数;S4: Risk analysis: Physiological risk analysis and biochemical risk analysis are performed on the multimodal time series matrix corresponding to each pregnancy detection period, and the pregnancy risk index of each pregnancy detection period is analyzed accordingly;S5:构建妊娠风险变化曲线:对各妊娠检测时间段对应的妊娠风险指数与妊娠检测时间段进行关联分析,由此构建妊娠风险变化曲线,进而确定风险发生时间;S5: Constructing a pregnancy risk change curve: Conducting correlation analysis between the pregnancy risk index corresponding to each pregnancy testing time period and the pregnancy testing time period, thereby constructing a pregnancy risk change curve and determining the time of risk occurrence;S6:可视化展示:按照预设的总结方式将风险发生时间发送至用户终端,管理人员通过总结内容对风险发生时间进行妊娠期糖尿病的重点检测。S6: Visual display: The risk occurrence time is sent to the user terminal according to the preset summary method. The management personnel conduct key detection of gestational diabetes at the risk occurrence time based on the summary content.2.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述生理风险指标具体采集过程包括:2. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the specific process of collecting the physiological risk indicators includes:通过可穿戴设备提取心电信号,并根据心电信号识别R波峰值,由此分别计算相邻R波间隔;The wearable device extracts the ECG signal, identifies the R-wave peak based on the ECG signal, and calculates the intervals between adjacent R waves.基于相邻R波间隔差值计算R波均方根,进而通过R波均方根计算所有R波间隔标准差;The R-wave root mean square is calculated based on the difference between adjacent R-wave intervals, and then the standard deviation of all R-wave intervals is calculated using the R-wave root mean square;提取妊娠检测时间对应的体重和身高,并通过体重指数计算公式获取体重指数,进而通过相邻妊娠检测时间的体重指数与相邻妊娠检测时间的时间间隔相比,得到体重增长速率;The weight and height corresponding to the pregnancy test time were extracted, and the body mass index was obtained using the body mass index calculation formula. The weight gain rate was then obtained by comparing the body mass index of adjacent pregnancy test times with the time interval between adjacent pregnancy test times.将R波间隔标准差和体重增长速率进行整合作为生理风险指标。The standard deviation of R-wave intervals and weight gain rate were integrated as physiological risk indicators.3.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述生化风险指标具体采集过程包括:3. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the specific process of collecting the biochemical risk indicators includes:通过产检终端识别孕妇的胰岛素情况,从中提取其空腹血糖和胰岛素浓度;Identify the insulin status of pregnant women through the prenatal examination terminal and extract their fasting blood sugar and insulin concentrations;对孕妇的空腹血糖进行异常状态标记,具体通过诊断标准进行分类:Abnormal fasting blood sugar levels in pregnant women are marked and classified according to diagnostic criteria: ,其中,为孕妇在各妊娠检测时间的异常状态标记结果,表示孕妇在各妊娠检测时间的空腹血糖,分别为各异常状态的空腹血糖标准值,具体依据糖尿病医学诊疗标准获取;in, Mark the abnormal status results of pregnant women at each pregnancy test time. It indicates the fasting blood sugar of pregnant women at each pregnancy test time. and They are the standard values of fasting blood glucose for each abnormal state, which are obtained according to the medical diagnosis and treatment standards for diabetes;对各妊娠检测时间的异常状态标记结果进行相加,若相加结果不等于0,则判断各妊娠检测时间对应空腹血糖存在异常,反之,则判断各妊娠检测时间对应空腹血糖正常;The abnormal status mark results of each pregnancy test time are added together. If the added result is not equal to 0, it is judged that the fasting blood glucose corresponding to each pregnancy test time is abnormal. Otherwise, it is judged that the fasting blood glucose corresponding to each pregnancy test time is normal.提取存在异常的各妊娠检测时间对应空腹血糖,依次对相邻妊娠检测时间对应空腹血糖进行相减后取绝对值,同时与两者最大值相比,由此获取血糖波动系数;Extract the fasting blood glucose corresponding to each abnormal pregnancy test time, subtract the fasting blood glucose corresponding to adjacent pregnancy test times, take the absolute value, and compare it with the maximum value of the two to obtain the blood glucose fluctuation coefficient;对提取的胰岛素浓度进行对比,将相邻妊娠检测时间对应胰岛素浓度分别相减,同时与相邻妊娠检测时间间隔相比,由此获取胰岛素分泌速率;The extracted insulin concentrations were compared, and the insulin concentrations corresponding to adjacent pregnancy test times were subtracted from each other, and compared with the time intervals between adjacent pregnancy tests to obtain the insulin secretion rate;将血糖波动系数和胰岛素分泌速率进行整合作为生化风险指标。The blood glucose fluctuation coefficient and insulin secretion rate were integrated as biochemical risk indicators.4.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述预设的顺序为风险检测数据对应妊娠检测时间段先后顺序,风险检测数据对应的时间序列包括生理风险时间序列和生化风险时间序列,其中生理风险时间序列按预设的顺序排列生理风险指标,具体表示为:,其中每个时间段对应生理风险特征包括R波间隔标准差和体重增长速率,同理生成生化风险时间序列,具体表示为:,其中每个时间段对应生化风险特征包括血糖波动系数和胰岛素分泌速率;通过妊娠检测时间段对齐不同模态数据,缺失检测的时间段采用时序插值进行填充,并基于生理风险时间序列和生化风险时间序列构建多模态时间序列矩阵,具体表示为:4. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the preset order is the chronological order of the risk detection data corresponding to the pregnancy detection time period, and the time series corresponding to the risk detection data includes a physiological risk time series and a biochemical risk time series, wherein the physiological risk time series arranges the physiological risk indicators in a preset order, specifically expressed as follows: , where each time period corresponds to physiological risk characteristics including the standard deviation of the R-wave interval and the weight growth rate. Similarly, the biochemical risk time series is generated, which is specifically expressed as: , where each time period corresponds to biochemical risk characteristics including blood glucose fluctuation coefficient and insulin secretion rate; different modal data are aligned by pregnancy detection time period, missing detection time period is filled by time series interpolation, and a multimodal time series matrix is constructed based on physiological risk time series and biochemical risk time series, which is specifically expressed as: ,其中,表示多模态时间序列矩阵,每行对应一个妊娠检测时间段,每列对应一个模态特征,表示生理风险时间序列中对应第m-1个妊娠检测时间段的生理风险特征,表示生化风险时间序列中对应第m-1个妊娠检测时间段的生化风险特征。in, Represents a multimodal time series matrix, where each row corresponds to a pregnancy detection time period and each column corresponds to a modal feature. represents the physiological risk characteristics corresponding to the m-1th pregnancy detection time period in the physiological risk time series, Represents the biochemical risk characteristics corresponding to the m-1th pregnancy detection time period in the biochemical risk time series.5.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述生理风险分析具体如下:基于各妊娠检测时间段对应的多模态时间序列矩阵提取生理风险特征,包括R波间隔标准差和体重增长速率;5. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the physiological risk analysis is specifically as follows: physiological risk features are extracted based on the multimodal time series matrix corresponding to each pregnancy detection time period, including the standard deviation of the R-wave interval and the weight gain rate;将R波间隔标准差和体重增长速率进行对比计算,得到各妊娠检测时间段对应生理异常风险系数。The standard deviation of the R-wave interval and the weight gain rate were compared and calculated to obtain the risk coefficient of physiological abnormality corresponding to each pregnancy detection time period.6.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述生化风险分析具体如下:基于各妊娠检测时间段对应的多模态时间序列矩阵提取生化风险特征,包括血糖波动系数和胰岛素分泌速率;6. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the biochemical risk analysis is specifically performed as follows: biochemical risk features, including blood glucose fluctuation coefficient and insulin secretion rate, are extracted based on the multimodal time series matrix corresponding to each pregnancy detection time period;将血糖波动系数和胰岛素分泌速率进行对比计算,得到各妊娠检测时间段对应生化异常风险系数,具体表示为:The blood glucose fluctuation coefficient and insulin secretion rate were compared and calculated to obtain the biochemical abnormality risk coefficient corresponding to each pregnancy testing period, which is specifically expressed as: ,其中,表示各妊娠检测时间段对应生化异常风险系数,分别表示生化风险特征中血糖波动系数和胰岛素分泌速率,分别表示相应标准血糖波动系数和胰岛素分泌速率。in, Indicates the biochemical abnormality risk coefficient corresponding to each pregnancy detection time period, They represent the blood sugar fluctuation coefficient and insulin secretion rate in the biochemical risk characteristics, They represent the corresponding standard blood glucose fluctuation coefficient and insulin secretion rate respectively.7.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述妊娠风险指数基于各妊娠检测时间段对应生理异常风险系数和生化异常风险系数进行权重相加,具体表示为:7. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 1, wherein the pregnancy risk index is weighted summation of the physiological abnormality risk coefficient and the biochemical abnormality risk coefficient corresponding to each pregnancy detection time period, specifically expressed as: ,其中,表示各妊娠检测时间段对应的妊娠风险指数,分别表示第m-1个妊娠检测时间段的生理异常风险系数和生化异常风险系数,z1和z2分别表示生理异常风险系数和生化异常风险系数对应的权重系数。in, Indicates the pregnancy risk index corresponding to each pregnancy detection time period, and They represent the physiological abnormality risk coefficient and biochemical abnormality risk coefficient of the m-1th pregnancy detection time period, respectively. z1 and z2 represent the weight coefficients corresponding to the physiological abnormality risk coefficient and biochemical abnormality risk coefficient, respectively.8.根据权利要求1所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述妊娠风险变化曲线具体构建方法如下:以妊娠检测时间为横轴,以妊娠风险指数为纵轴构建二维坐标系,针对各妊娠检测时间段内每次进行检测的妊娠检测时间和妊娠风险指数在所构建的二维坐标系内标注若干点,形成妊娠检测时间段内的妊娠风险变化曲线,进而确定风险发生时间。8. The method for intelligently predicting gestational diabetes risk based on multimodal data fusion according to claim 1 is characterized in that: the pregnancy risk change curve is specifically constructed as follows: a two-dimensional coordinate system is constructed with the pregnancy detection time as the horizontal axis and the pregnancy risk index as the vertical axis. Several points are marked in the constructed two-dimensional coordinate system for each pregnancy detection time and pregnancy risk index during each pregnancy detection time period to form a pregnancy risk change curve within the pregnancy detection time period, thereby determining the time when the risk occurs.9.根据权利要求8所述的基于多模态数据融合的妊娠期糖尿病风险智能预测方法,其特征在于:所述风险发生时间具体确定如下:在妊娠检测时间段内的妊娠风险变化曲线上标注出拐点;9. The method for intelligent prediction of gestational diabetes risk based on multimodal data fusion according to claim 8, wherein the risk occurrence time is specifically determined as follows: an inflection point is marked on the pregnancy risk change curve within the pregnancy detection time period;基于标注的拐点将妊娠风险变化曲线划分为上升段、下降段以及平段,由此分别对上升段占比和上升段幅度进行统计,同时将妊娠风险变化曲线对应上升段按照所处位置由前到后进行顺序编号;Based on the marked inflection points, the pregnancy risk change curve is divided into rising segments, falling segments, and flat segments. The proportion and amplitude of the rising segments are then statistically analyzed. At the same time, the rising segments of the pregnancy risk change curve are numbered sequentially from front to back according to their position.基于上升段占比和上升段幅度计算各上升段的风险发生概率,具体通过概率计算公式对上升段占比和上升段幅度进行关联;Calculate the risk probability of each rising segment based on the proportion of rising segments and the amplitude of rising segments. Specifically, associate the proportion of rising segments and the amplitude of rising segments through the probability calculation formula;将妊娠检测时间段内的妊娠风险变化曲线对应各上升段的风险发生概率进行对比,从中筛选出最大风险发生概率对应的妊娠检测时间段作为风险发生时间。The risk occurrence probabilities of each rising segment of the pregnancy risk change curve within the pregnancy detection time period are compared, and the pregnancy detection time period corresponding to the maximum risk occurrence probability is selected as the risk occurrence time.
CN202511250827.2A2025-09-032025-09-03 Intelligent prediction method for gestational diabetes risk based on multimodal data fusionPendingCN120727304A (en)

Priority Applications (1)

Application NumberPriority DateFiling DateTitle
CN202511250827.2ACN120727304A (en)2025-09-032025-09-03 Intelligent prediction method for gestational diabetes risk based on multimodal data fusion

Applications Claiming Priority (1)

Application NumberPriority DateFiling DateTitle
CN202511250827.2ACN120727304A (en)2025-09-032025-09-03 Intelligent prediction method for gestational diabetes risk based on multimodal data fusion

Publications (1)

Publication NumberPublication Date
CN120727304Atrue CN120727304A (en)2025-09-30

Family

ID=97155401

Family Applications (1)

Application NumberTitlePriority DateFiling Date
CN202511250827.2APendingCN120727304A (en)2025-09-032025-09-03 Intelligent prediction method for gestational diabetes risk based on multimodal data fusion

Country Status (1)

CountryLink
CN (1)CN120727304A (en)

Similar Documents

PublicationPublication DateTitle
US12023178B2 (en)Method of detecting sepsis using vital signs, including systolic blood pressure, hematology parameters, and combinations thereof
EP0370085B1 (en)Cardiac death probability determining device
CN117954116B (en)Breathing severe patient state monitoring and early warning method based on artificial intelligence
CN114533121B (en) Cerebral perfusion state prediction device, method, device and model training device
CN119028512B (en)Obstetrical real-time nursing optimization method based on big data
CN113506640A (en) Cerebral perfusion state classification device, method, device and model training device
US20210361217A1 (en)Ecg-based age and sex estimation
WO2022246882A1 (en)System and method for diagnosing polycystic ovary syndrome
US8838221B2 (en)Automatic categorization of fetal heart rate information
Zaman et al.Estimating reliability of signal quality of physiological data from data statistics itself for real-time wearables
Liu et al.Feasibility of antenatal ambulatory fetal electrocardiography: a systematic review
CN112289435B (en) A screening system for gestational diabetes mellitus based on machine learning and physical examination data
Guo et al.Non-invasive monitoring and evaluating cardiac function of pregnant women based on a relative value method
Narain et al.Antenatal assessment of fetal well-being
CN120727304A (en) Intelligent prediction method for gestational diabetes risk based on multimodal data fusion
CN116246777B (en)Gynaecology and obstetrics information analysis method based on community yield test data stream
CN101264012B (en) A blood pressure measurement data processing device
CN114795170A (en)Equipment for evaluating intracranial state based on intracranial pressure signal
Gu et al.A study of maternal hemodynamic change during healthy pregnancy and women with gestation hypertension
Mappa et al.Maternal cardiovascular function in the prediction of fetal distress in labor: a prospective cohort study
CN116759085A (en)Pregnancy period circumference data management system based on intelligent terminal
Fullerton et al.Measurement of aortic augmentation index in pregnant women with raised blood pressure and subsequent outcomes: a preliminary prospective cohort study
EP4368115A1 (en)Cardiotocographic scanning session duration
CN118658644B (en)Disease state change trend early warning system
Korneeva et al.A System for Remote Monitoring of Pregnant Women's Health State and Pregnancy Complications Prediction.

Legal Events

DateCodeTitleDescription
PB01Publication

[8]ページ先頭

©2009-2025 Movatter.jp