Predelivery monitoring device, system and method

Abstract

The invention relates to the technical field of medical assistance, in particular to a predelivery monitoring device, system and method, and the method comprises the steps: collecting the physical sign information of a puerpera and the physical sign information of a fetus, and carrying out the preprocessing; a production analysis model is constructed and adopted, and the uterine contraction condition of the puerpera is analyzed and predicted according to the physical sign information of the puerpera; according to the predicted uterine contraction condition, the production time is analyzed and predicted; performing comparative analysis according to the predicted uterine contraction condition and the actually collected uterine contraction condition in the preset time period, and optimizing the production analysis model according to an analysis result; according to the predicted parturition time of the puerpera, suggestion information is generated and pushed; whether the puerpera sign information and the fetus sign information are abnormal or not is judged, and if yes, early warning is conducted. According to the scheme, the state of the puerpera can be analyzed according to the monitored physical sign information, the parturition time is predicted, the information utilization rate is improved, the puerpera and medical staff are assisted in knowing the condition of the puerpera, corresponding measures can be conveniently taken in time, and parturition safety of the puerpera is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of medical assistance technology, and particularly to a labor monitoring device, system and method. Background Art

[0002] Labor monitoring refers to the process of conducting a series of examinations and monitoring on the health status of the mother and fetus when the pregnant woman is approaching the delivery period. These monitors help to promptly detect potential problems that may affect the health of the mother and baby and take necessary medical measures.

[0003] Common labor monitoring items include: fetal heart monitoring, which monitors the fetal heart rate through external or internal devices to evaluate the fetal status in the uterus; uterine contraction monitoring, which records the frequency, intensity and duration of uterine contractions to help doctors understand the progress of labor; cervical examination, which evaluates the size (dilation), softness, position and effacement of the cervix to determine the progress of labor; ultrasound examination, which is used to evaluate important indicators such as fetal size, position and amniotic fluid volume; blood and urine tests, which monitor the health status of the mother, such as anemia, signs of infection, etc.

[0004] Monitoring of fetal heart, uterine contractions, etc. can be carried out when the parturient is at home. Earlier monitoring is beneficial for further grasping the status of the parturient and ensuring the safety of the parturient during childbirth. However, the existing monitoring methods mainly judge whether there are abnormalities after monitoring the corresponding physical sign information, and the information utilization rate is low; a large amount of monitored information is not reasonably applied to enable the parturient to more accurately understand her own situation and take corresponding accurate measures.

[0005] Therefore, there is an urgent need for a labor monitoring device, system and method that can analyze the status of the parturient according to the monitored physical sign information, predict the delivery time, improve the information utilization rate, assist the parturient and medical staff to understand the situation of the parturient, facilitate timely taking corresponding measures, and ensure the safety of the parturient during childbirth. Summary of the Invention

[0006] One of the purposes of the present invention is to provide a labor monitoring method that can analyze the status of the parturient according to the monitored physical sign information, predict the delivery time, improve the information utilization rate, assist the parturient and medical staff to understand the situation of the parturient, facilitate timely taking corresponding measures, and ensure the safety of the parturient during childbirth.

[0007] The first basic solution provided by the present invention: A labor monitoring method includes:

[0008] Information monitoring step: Collecting the physical sign information of the parturient and the physical sign information of the fetus;

[0009] Data processing step: Preprocessing the physical sign information of the parturient and the physical sign information of the fetus;

[0010] Model construction steps: Construct a production analysis model for analyzing and predicting the uterine contraction situation of parturients based on parturient physical sign information;

[0011] Model training steps: Use historical parturient physical sign information and production time to construct a data set and train the production analysis model;

[0012] Data analysis steps: Use the production analysis model to analyze and predict the uterine contraction situation of parturients based on parturient physical sign information; Analyze and predict the production time according to the predicted uterine contraction situation;

[0013] Model optimization steps: Compare and analyze the predicted uterine contraction situation with the actually collected uterine contraction situation within the preset time period, and optimize the production analysis model according to the analysis results;

[0014] Production suggestion steps: Generate suggestion information according to the predicted production time of parturients and push it;

[0015] Monitoring and analysis steps: Judge whether there are abnormalities in parturient physical sign information and fetal physical sign information. If so, give an early warning.

[0016] Furthermore, the parturient physical sign information includes one or more of: body temperature, heart rate, blood pressure, uterine contraction situation; among which, the uterine contraction situation includes one or more of: start time of each uterine contraction, end time, uterine contraction frequency, pain level;

[0017] Fetal physical sign information includes: fetal heart rate.

[0018] Furthermore, the preprocessing includes: data cleaning and data normalization;

[0019] Data cleaning includes: handling missing values, removing noise data, handling outliers;

[0020] Data normalization scales the data to a specific range, such as [0, 1], using Min - Max scaling:

[0021]

[0022] where x is the original data, x min and x max are respectively the minimum and maximum values of this feature in the data set, and x new is the normalized data value.

[0023] Furthermore, the production analysis model uses a neural network model to analyze and predict the uterine contraction situation of parturients based on heart rate, blood pressure, and uterine contraction situation;

[0024] The model construction steps include:

[0025] Construct the input time series: Construct the input time series [(t1, p1, h1, b1), (t2, p2, h2, b2), …, (t n , p n , h n , b n )], where represents the time point, where t represents the time interval set according to the monitoring requirements, p represents the pain level of uterine contractions at the corresponding time point. If it is zero, it means there are no uterine contractions at the current time point. h represents the heart rate at the corresponding time point, and b represents the blood pressure at the corresponding time point;

[0026] Construct the LSTM model architecture: In the created sequential model, add K LSTM layers, with R LSTM units in each layer, and return the sequence for the next layer to continue processing the time series information; Add a Dropout layer after each LSTM layer, and the last Dropout layer is connected to a fully connected layer to output the prediction result; Compile the model using the Adam optimizer and select the mean squared error as the loss function.

[0027] Furthermore, the model training steps include:

[0028] Perform sequence partitioning, split the input time series into input and output pairs, select the pain level, heart rate, and blood pressure of uterine contractions corresponding to consecutive i time points as the input, and use them to predict the pain level, heart rate, and blood pressure of uterine contractions at the i + a to i + b time points, where b > a;

[0029] Among them, the construction of the input-output pair adopts historical maternal physical sign information, including: one or more of the maternal physical sign information of the delivered mother and the maternal physical sign information of the current expectant mother;

[0030] Use the input-output pair for training and obtain the LSTM model with the loss function meeting the preset requirements as the production analysis model.

[0031] Furthermore, the data analysis steps include:

[0032] S501. Initialize the judgment count k = 1;

[0033] S502. Use the production analysis model to analyze and predict the uterine contraction situation of the mother within the preset time period according to the maternal physical sign information, that is, the pain level within the preset time period;

[0034] S503. According to the output result of the production analysis model, analyze and calculate the duration of uterine contractions and the intermittent duration;

[0035] S504. Determine whether the duration of uterine contractions falls within the preset duration range and whether the intermittent duration falls within the preset intermittent duration range and is greater than the second preset interval duration. If so, execute S505; if not, execute S506;

[0036] S505. Extract the time period during which the duration of uterine contractions is greater than the minimum preset duration within the preset duration range and the intermittent duration is less than the maximum preset intermittent duration within the preset intermittent duration range as the predicted delivery time, and execute S5014;

[0037] S506. Analyze the relationship between the duration of uterine contractions and the preset duration range, and the relationship between the intermittent duration and the preset intermittent duration range;

[0038] If the duration of uterine contractions is greater than the preset duration range and the intermittent duration is less than the preset intermittent duration range, execute S507;

[0039] If the duration of uterine contractions is less than the preset duration range and the intermittent duration is greater than the preset intermittent duration range, execute S508;

[0040] If the duration of uterine contractions falls within the preset duration range and the intermittent duration is greater than the preset intermittent duration range, execute S5011;

[0041] If the duration of uterine contractions is less than the preset duration range and the intermittent duration falls within the preset intermittent duration range, execute S5012;

[0042] If the duration of uterine contractions falls within the preset duration range and the intermittent duration is less than the preset intermittent duration range, execute S507;

[0043] If the duration of uterine contractions is greater than the preset duration range and the intermittent duration falls within the preset intermittent duration range, execute S507;

[0044] If the duration of uterine contractions is greater than the preset duration range and the intermittent duration is greater than the preset intermittent duration range, execute S507;

[0045] If the duration of uterine contractions is less than the preset duration range and the intermittent duration is less than the preset intermittent duration range, execute S507;

[0046] S507. Move the minimum time point of the preset time period forward by the preset time step, that is, closer to the current time point, and execute S509;

[0047] S508. Move the maximum time point of the preset time period backward by the preset time step, that is, farther from the current time point, and execute S509;

[0048] S509. Update k = k + 1;

[0049] S5010. Determine whether k is equal to the preset number of times. If so, execute; if not, execute S502.

[0050] S5011. Extract the time period during which the duration of uterine contractions is greater than the minimum preset duration within the preset duration range as the predicted delivery time, and execute S5014.

[0051] S5012. Extract the time period during which the intermittent duration is less than the maximum preset intermittent duration within the preset intermittent duration range as the predicted delivery time, and execute S5014.

[0052] S5013. Output the uterine contraction conditions of the parturient within all preset time periods predicted, and give special reminders about the parturient's conditions.

[0053] S5014. Output the preset delivery time.

[0054] Furthermore, the model optimization steps include:

[0055] Calculate the loss function value of the predicted uterine contraction conditions and the actually collected uterine contraction conditions within this preset time period as the first loss function value.

[0056] Determine whether the first loss function value is greater than the preset loss function value. If so, collect the parturient's physical sign information before the current time point, construct a data set, retrain the production analysis model, and calculate the loss function value after training as the second loss function value. If the second loss function value is less than or equal to the preset loss function value, the data analysis steps adopt the production analysis model after retraining; if not, use the optimization model to optimize the weights in the production analysis model, and use the loss function as the fitness evaluation function to obtain the optimal weight scheme and update the production analysis model.

[0057] Furthermore, the production suggestion steps include:

[0058] Calculate the time interval length between the current time and the predicted delivery time. If the time interval length is less than the preset time interval length, generate suggestion information to go to the hospital immediately or call 120 and push it.

[0059] If the time interval length is greater than the preset time interval length, extract the production preparation suggestion information and the predicted delivery time as the suggestion information and push it.

[0060] The second object of the present invention is to provide a monitoring device for parturients in labor, which can analyze the parturient's state according to the monitored physical sign information, predict the delivery time, improve the information utilization rate, assist the parturient and medical staff in understanding the parturient's condition, and facilitate taking corresponding measures in a timely manner to ensure the safety of the parturient during childbirth.

[0061] The present invention provides Basic Solution 2: A device for performing the above-mentioned prenatal monitoring method.

[0062] The third object of the present invention is to provide a prenatal monitoring that can analyze the status of parturients according to the monitored physical sign information, predict the delivery time, improve the information utilization rate, assist parturients and medical staff in understanding the status of parturients, facilitate timely taking corresponding measures, and ensure the safety of parturient delivery.

[0063] The present invention provides Basic Solution 3: A prenatal monitoring system for performing the above-mentioned prenatal monitoring method, including:

[0064] An information monitoring module for collecting parturient physical sign information and fetal physical sign information;

[0065] A central control module for preprocessing parturient physical sign information and fetal physical sign information;

[0066] It is also used to construct a production analysis model for analyzing and predicting the uterine contraction situation of parturients according to parturient physical sign information;

[0067] It is also used to construct a data set with historical parturient physical sign information and delivery time to train the production analysis model;

[0068] It is also used to use the production analysis model to analyze and predict the uterine contraction situation of parturients according to parturient physical sign information; and analyze and predict the delivery time according to the predicted uterine contraction situation;

[0069] It is also used to compare and analyze the predicted uterine contraction situation with the actually collected uterine contraction situation within the preset time period, and optimize the production analysis model according to the analysis results;

[0070] It is also used to generate and push suggestion information according to the predicted delivery time of parturients;

[0071] It is also used to determine whether there are abnormalities in parturient physical sign information and fetal physical sign information. If so, trigger the warning module for warning;

[0072] A warning module for performing data anomaly warning.

[0073] Beneficial effects: This solution collects and monitors parturient physical sign information and fetal physical sign information for parturients in labor. For obtaining parturient physical sign information, a production analysis model is constructed for analysis, predicting the uterine contraction situation of parturients, and then analyzing and predicting the delivery time. According to the predicted delivery time, corresponding suggestion information can be generated and pushed to the terminals of parturients and medical staff; at the same time, abnormal analysis is also performed on the collected and monitored physical sign information. If there are abnormalities, warnings will be given in a timely manner, thereby improving the information utilization rate, better assisting parturients and medical staff in understanding the status of parturients, facilitating timely taking corresponding measures, and ensuring the safety of parturient delivery.

[0074] Among them, the production analysis model will be optimized according to the analysis result of comparing the predicted uterine contraction situation with the actually collected uterine contraction situation within the preset time period, so as to improve the accuracy of the prediction of the production analysis model and the accuracy of predicting the production time. At the same time, during the data analysis process, according to the predicted uterine contraction situation, the production time is analyzed. It is not a single judgment on whether the duration of uterine contraction and the intermittent duration are within the range of the labor state, but considering the possible atypical uterine contractions during the parturient's delivery. For different situations of different parturients, multi-level judgments are made, and the prediction time period is optimized, so that the prediction time period can better cover the parturient's production time, thereby obtaining a more accurate prediction result, ensuring the accuracy of the prediction, and further improving the guarantee strength for the parturient's production safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a schematic flowchart of an embodiment of a method for monitoring parturition of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0076] The following is a more detailed description through specific embodiments:

[0077] Embodiment 1

[0078] This embodiment is basically as shown in the appendix Figure 1 shown: A method for monitoring parturition, including:

[0079] Information monitoring step: Collecting the physical signs information of the parturient and the physical signs information of the fetus;

[0080] Among them, the physical signs information of the parturient includes, but is not limited to: body temperature, heart rate, blood pressure, uterine contraction situation; among them, the uterine contraction situation includes: the start time, end time, uterine contraction frequency, and pain level of each uterine contraction;

[0081] The physical signs information of the fetus includes, but is not limited to: fetal heart rate;

[0082] The collection method can use existing monitoring devices to collect data through a terminal;

[0083] Data processing step: Preprocessing the physical signs information of the parturient and the physical signs information of the fetus;

[0084] Among them, the preprocessing includes: data cleaning and data normalization;

[0085] Data cleaning includes processing missing values, removing noise data, and processing outliers. For example: using the mean, median or mode to fill in the missing values; for outliers, deletion or correction can be selected according to the specific situation;

[0086] Data normalization scales the data to a specific range, such as [0, 1]. In this embodiment, Min-Max scaling is adopted:

[0087]

[0088] where x is the original data, x min and x max are the minimum and maximum values of this feature in the dataset respectively, and x new is the normalized data value. The normalization process maintains the relationship between the original data but unifies the scales of all features into the same interval. In the subsequent neural network, the normalization of the input data can accelerate the training speed and help avoid the problems of gradient disappearance or explosion.

[0089] Model construction steps: Construct a production analysis model for analyzing and predicting the uterine contractions of parturients based on the physical signs information of parturients;

[0090] Among them, the production analysis model adopts a neural network model. In this embodiment, an LSTM model is adopted to analyze and predict the uterine contractions of parturients based on heart rate, blood pressure, and uterine contraction conditions;

[0091] The specific model construction process is as follows:

[0092] Construct the input time series: Construct the input time series [(t1, p1, h1, b1), (t2, p2, h2, b2), …, (t n , p n , h n , b n )], where represents the time point, where t represents the time interval set according to the monitoring requirements. Considering that this solution is for the monitoring and prediction of uterine contractions, all time nodes are set to one time point every 10s. p represents the pain level of uterine contractions at the corresponding time point. If it is zero, it means there is no uterine contraction at the current time point. h represents the heart rate at the corresponding time point, and b represents the blood pressure at the corresponding time point; The actual format input to the model is (number of samples, time steps, number of features), corresponding to (number of samples, 10, 3), that is, the number of feature numbers is 3, including: the pain level of uterine contractions, heart rate, and blood pressure;

[0093] Construct the LSTM model architecture: In the created sequential model, add K LSTM layers, with the number of LSTM units in each layer being R, and return the sequence for the next layer to continue processing the time series information; Add a Dropout layer after each LSTM layer to prevent overfitting. The last Dropout layer is connected to a fully connected layer to output the prediction results, such as the pain level, heart rate, and blood pressure at the i + a to i + b time points, [(t i+a , p i+a , h i+a , bi+a ),(t i+a+1 ,p i+a+1 ,h i+a+1 ,b i+a+1 ),…,(t i+b ,p i+b ,h i+b ,b i+b )]; Compile the model using the Adam optimizer and select the mean squared error as the loss function;

[0094] Model training steps: Construct a dataset using the historical maternal physical sign information and the delivery time, and train the delivery analysis model;

[0095] Specifically, perform sequence partitioning. Since the data has been normalized in the data processing step, sequence partitioning can be directly performed. Split the input time series into input (X) and output (y) pairs. Select the pain level, heart rate, and blood pressure corresponding to consecutive i time points of uterine contractions as the input to predict the pain level, heart rate, and blood pressure of uterine contractions at the i+a to i+b time points, where b > a;

[0096] Among them, for the construction of the input-output pairs, use the historical maternal physical sign information, including: one or more of the maternal physical sign information of the delivered parturients and the maternal physical sign information of the current parturients in labor;

[0097] Use the input-output pairs for training and obtain an LSTM model whose loss function meets the preset requirements as the delivery analysis model; Use the delivery analysis model to predict the pain level, heart rate, and blood pressure in a preset future time period according to the input time series and perform denormalization processing to obtain the actual predicted values (pain level, heart rate, and blood pressure).

[0098] Data analysis steps: Use the delivery analysis model to analyze and predict the uterine contraction situation of the parturient according to the maternal physical sign information; Analyze and predict the delivery time according to the predicted uterine contraction situation;

[0099] Specifically, S501, Initialize the judgment count k = 1;

[0100] S502, Use the delivery analysis model to analyze and predict the uterine contraction situation of the parturient within a preset time period, that is, the pain level within the preset time period;

[0101] S503, According to the output result of the delivery analysis model (the pain level in the preset time period), analyze and calculate the duration of uterine contractions and the intermittent duration;

[0102] S504. Determine whether the duration of uterine contractions belongs to the preset duration range and whether the interval duration belongs to the preset interval duration range and is greater than the second preset interval duration. If so, execute S505; if not, execute S506;

[0103] S505. Extract the time period during which the duration of uterine contractions is greater than the minimum preset duration within the preset duration range and the interval duration is less than the maximum preset duration within the preset interval duration range as the predicted delivery time, and execute S5014;

[0104] S506. Analyze the relationship between the duration of uterine contractions and the preset duration range, and the relationship between the interval duration and the preset interval duration range;

[0105] If the duration of uterine contractions is greater than the preset duration range and the interval duration is less than the preset interval duration range, execute S507;

[0106] If the duration of uterine contractions is less than the preset duration range and the interval duration is greater than the preset interval duration range, execute S508;

[0107] If the duration of uterine contractions belongs to the preset duration range and the interval duration is greater than the preset interval duration range, execute S5011;

[0108] If the duration of uterine contractions is less than the preset duration range and the interval duration belongs to the preset interval duration range, execute S5012;

[0109] If the duration of uterine contractions belongs to the preset duration range and the interval duration is less than the preset interval duration range, execute S507;

[0110] If the duration of uterine contractions is greater than the preset duration range and the interval duration belongs to the preset interval duration range, execute S507;

[0111] If the duration of uterine contractions is greater than the preset duration range and the interval duration is greater than the preset interval duration range, execute S507;

[0112] If the duration of uterine contractions is less than the preset duration range and the interval duration is less than the preset interval duration range, execute S507;

[0113] S507. Move the preset time period forward by a preset time step, that is, closer to the current time point, and execute S509;

[0114] S508. Move the preset time period forward by a preset time step, that is, farther from the current time point, and execute S509;

[0115] S509. Update k = k + 1;

[0116] S5010. Determine whether k is equal to the preset number of times. If so, execute; if not, execute S502;

[0117] S5011. Extract the time period during which the duration of uterine contractions is greater than the minimum preset duration within the preset duration range as the predicted delivery time, and execute S5014;

[0118] S5012. Extract the time period during which the intermittent duration is less than the maximum preset intermittent duration within the preset intermittent duration range as the predicted delivery time, and execute S5014;

[0119] S5013. Output the uterine contraction conditions of the parturient within all preset time periods, and give special reminders about the parturient's conditions;

[0120] S5014. Output the preset delivery time. In this embodiment, considering the actual production situation, the preset duration range is set to 25 - 90 seconds, and the intermittent duration range is 0.5 - 7 minutes, that is, 30 - 420 seconds. There are a large number of individual differences in the production process, and the uterine contraction conditions of not every parturient are typical. Therefore, in this solution, it is not a single judgment on whether the duration of uterine contractions and the intermittent duration belong to the range of the labor state, but a multi-level judgment for different situations of different parturients, and the predicted time period is optimized, so that the predicted time period can better cover the parturient's delivery time, thereby obtaining a more accurate prediction result and ensuring the accuracy of the prediction.

[0121] Model optimization step: Compare and analyze the predicted uterine contraction conditions with the actually collected uterine contraction conditions within this preset time period, and optimize the production analysis model according to the analysis results;

[0122] Specifically, calculate the loss function value of the predicted uterine contraction conditions and the actually collected uterine contraction conditions within this preset time period as the first loss function value;

[0123] Determine whether the first loss function value is greater than the preset loss function value. If so, collect the parturient's physical sign information before the current time point, construct a data set, retrain the production analysis model, and calculate the loss function value after training as the second loss function value. If the second loss function value is less than or equal to the preset loss function value, the data analysis step uses the retrained production analysis model; if not, use the optimization model to optimize the weights in the production analysis model, and use the loss function as the fitness evaluation function to obtain the optimal weight scheme and update the production analysis model. The optimization model can iteratively optimize the optimal weight scheme to improve the analysis and prediction accuracy of the production analysis model.

[0124] Production advice step: Generate advice information according to the predicted delivery time of the parturient and push it;

[0125] Specifically, calculate the time interval between the current time and the predicted production time. If the time interval is less than the preset time interval, generate and push advice information to go to the hospital immediately or call 120, such as pushing it to the terminals related to the pregnant woman preset, and the terminals of the pregnant woman's doctor.

[0126] If the time interval is greater than the preset time interval, extract the production preparation advice information and the predicted production time as the advice information and push it, such as pushing it to the terminals related to the pregnant woman preset, and the terminals of the pregnant woman's doctor.

[0127] Monitoring and analysis step: Determine whether there are abnormalities in the pregnant woman's physical signs information and the fetus's physical signs information. If so, give an early warning.

[0128] Specifically, if any of the pregnant woman's physical signs information or the fetus's physical signs information does not belong to the corresponding preset range, it is determined that the pregnant woman's physical signs information or the fetus's physical signs information is abnormal, and an early warning is given, such as pushing the early warning information to the terminals related to the pregnant woman, and the terminals of the pregnant woman's doctor.

[0129] This embodiment also provides a device for executing the above-mentioned pending delivery monitoring method.

[0130] Embodiment 2

[0131] This embodiment is basically the same as the above embodiment, except that: the collection of the uterine contraction situation can also be carried out through the set uterine contraction monitoring device.

[0132] The uterine contraction monitoring device includes: a controller and a pressing cylinder body, and a pressure sensor is arranged on the outer peripheral surface of the pressing cylinder body.

[0133] The pressure sensor is used to collect the pressure value of the press and send a pressure signal to the controller.

[0134] The controller is used to receive the pressure signal, obtain the pressure value, and analyze the uterine contraction situation according to the pressure value.

[0135] Specifically, the time when the pressure sensor collects a pressure value greater than the lowest pressure threshold is taken as the start time of uterine contractions. Since the actual transmission delay is low, in this embodiment, the time when the controller receives the pressure signal with a pressure value greater than the lowest pressure threshold is directly taken as the start time of uterine contractions, and the start time of uterine contractions is recorded. The time when the pressure sensor collects a pressure value less than or equal to the lowest pressure threshold after the start time of uterine contractions is taken as the end time of uterine contractions. In this embodiment, the time when the controller receives the pressure signal with a pressure value less than or equal to the lowest pressure threshold is directly taken as the end time of uterine contractions, and the end time of uterine contractions is recorded, thereby completing the time record of one uterine contraction and circularly recording the uterine contraction situation. After the controller obtains the pressure value, according to the preset mapping relationship between the pressure value and the pain level, the pressure value is converted into the pain level and recorded in the record of this uterine contraction situation. After recording multiple uterine contractions, the uterine contraction frequency is calculated according to the time and number of uterine contractions. In addition, for the preset mapping relationship between the pressure value and the pain level, by pre-collecting the continuous pressure values collected during the process of the parturient holding and pressing the columnar body and gradually increasing the applied grip force to the maximum grip force and mapping them to the set pain levels, the mapping relationship between the pressure value and the pain level is formed.

[0136] When the parturient is in pain, the hand will clench tightly to relieve the pain. The more painful, the greater the hand grip force. Therefore, a hand-held uterine contraction monitoring device is set up. When the parturient has uterine contractions and pain, it can provide a hand-holding point for the parturient. And by the parturient pressing the pressing columnar body of the uterine contraction monitoring device, the pressure value of the parturient's pressing can be collected, and the uterine contraction situation can be calculated and analyzed based on this. The collection process is convenient and there is no need to additionally record the uterine contraction situation, which increases the convenience of parturient monitoring.

[0137] In addition, in other embodiments, data is collected through the terminal at the same time. For example, the uterine contraction situation is recorded by an APP. If the uterine contraction situation collected by the uterine contraction monitoring device is different from that collected by the terminal, the uterine contraction situation is corrected. Specifically, a uterine contraction situation source selection signal is obtained. According to the uterine contraction situation source selection signal, it is determined whether the uterine contraction situation source is the terminal or the uterine contraction monitoring device. If it is determined that the uterine contraction situation source is the terminal, when the uterine contraction situation collected by the uterine contraction monitoring device is different from that collected by the terminal, the uterine contraction situation collected by the uterine contraction monitoring device is corrected according to the uterine contraction situation collected by the terminal. The data in the uterine contraction situation collected by the uterine contraction monitoring device that is different from the uterine contraction situation collected by the terminal is corrected to the data of the uterine contraction situation collected by the terminal. And if the modified item is the pain level, according to the modified pain level and the collected pressure value, the preset mapping relationship between the pressure value and the pain level is modified so that the modified pain level and the collected pressure value conform to the mapping relationship.

[0138] If it is determined that the source of the uterine contraction situation is the uterine contraction monitoring device, when the uterine contraction situation collected by the terminal is different from that collected by the uterine contraction monitoring device, the uterine contraction situation collected by the terminal is corrected according to the uterine contraction situation collected by the uterine contraction monitoring device, and the data in the uterine contraction situation collected by the terminal that is different from the uterine contraction situation collected by the uterine contraction monitoring device is corrected to the data of the uterine contraction situation collected by the uterine contraction monitoring device, so as to achieve data correction, ensure data accuracy. Especially after the mapping relationship is corrected, the accuracy of the uterine contraction monitoring device recognition is ensured, and then the accuracy of subsequent analysis is ensured.

[0139] Embodiment III

[0140] This embodiment provides a prenatal monitoring system for implementing the above-mentioned prenatal monitoring method, including:

[0141] An information monitoring module for collecting maternal physical signs information and fetal physical signs information;

[0142] A central control module for preprocessing maternal physical signs information and fetal physical signs information;

[0143] It is also used to construct a production analysis model for analyzing and predicting the uterine contraction situation of the parturient according to the maternal physical signs information;

[0144] It is also used to construct a data set by using historical maternal physical signs information and production time to train the production analysis model;

[0145] It is also used to use the production analysis model to analyze and predict the uterine contraction situation of the parturient according to the maternal physical signs information; according to the predicted uterine contraction situation, analyze and predict the production time;

[0146] It is also used to compare and analyze the predicted uterine contraction situation with the actually collected uterine contraction situation within the preset time period, and optimize the production analysis model according to the analysis results;

[0147] It is also used to generate and push suggestion information according to the predicted production time of the parturient;

[0148] It is also used to judge whether there are abnormalities in the maternal physical signs information and fetal physical signs information. If so, trigger the warning module to give a warning;

[0149] A warning module for giving a warning of data abnormality.

[0150] This system collects and monitors the physical signs information of the pregnant woman in labor and the physical signs information of the fetus. For obtaining the physical signs information of the pregnant woman, a production analysis model is constructed for analysis to predict the uterine contraction situation of the pregnant woman, and then to analyze and predict the production time. According to the predicted production time, corresponding advice information can be generated and pushed to the terminals related to the pregnant woman and the terminals of medical staff. At the same time, abnormal analysis is also carried out on the collected and monitored physical signs information. If there are abnormalities, early warnings will be given in a timely manner, thereby improving the information utilization rate, and being able to better assist the pregnant woman and medical staff in understanding the condition of the pregnant woman, facilitating the timely adoption of corresponding measures, and ensuring the safety of the pregnant woman during childbirth.

[0151] The above are only embodiments of the present invention. Specific structures and characteristics and other common knowledge in the art are not described in detail here. Those of ordinary skill in the art know all the common general technical knowledge in the technical field to which the invention belongs before the application date or the priority date, can know all the existing technologies in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, combine their own abilities to improve and implement this solution. Some typical well-known structures or well-known methods should not become obstacles for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent. The protection scope required by this application should be subject to the content of its claims, and the specific implementation manners and other records in the specification can be used to explain the content of the claims.

Claims

1. A method for monitoring labor, comprising: Information monitoring steps: collecting maternal and fetal vital signs information; Monitoring and analysis steps: determine whether there are any abnormalities in the maternal and fetal physical signs information, and if so, issue an early warning; It is characterized by further comprising: Data processing steps: pre-processing maternal and fetal physical sign information; Model building steps: Build a production analysis model to analyze and predict the uterine contractions of parturients based on their physical signs; Model training steps: Use historical maternal vital signs information and delivery time to build a data set and train the delivery analysis model; Data analysis steps: Use the production analysis model to analyze and predict the uterine contractions of the parturient according to the parturient's physical signs; analyze and predict the delivery time according to the predicted uterine contractions; Model optimization step: perform comparative analysis based on the predicted uterine contraction situation and the actual uterine contraction situation collected within the preset time period, and optimize the production analysis model based on the analysis results; Delivery suggestion steps: Generate and push suggestion information based on the predicted delivery time of the mother.

2. The method for monitoring labor preparation according to claim 1, characterized in that: The maternal physical sign information includes: one or more of body temperature, heart rate, blood pressure, and uterine contraction conditions; wherein the uterine contraction conditions include: one or more of the start time, end time, frequency, and pain level of each uterine contraction; Fetal vital signs information, including: fetal heart rate.

3. The method for monitoring labor preparation according to claim 1, characterized in that: The preprocessing includes: data cleaning and data normalization; Data cleaning includes: processing missing values, removing noise data, and processing outliers; Data normalization scales the data to a specific range, such as [0,1], using Min-Max scaling: Where x is the original data, x min and x max are the minimum and maximum values ​​of the feature in the dataset, respectively, and x new is the normalized data value.

4. The method for monitoring labor preparation according to claim 2, characterized in that: The production analysis model adopts a neural network model to analyze and predict the uterine contractions of parturient women based on heart rate, blood pressure, and uterine contractions. The model building step comprises: Construct input time series: Construct input time series [(t1,p1,h1,b1),(t2,p2,h2,b2),…,(t n ,p n ,h n ,b n )], where represents the time point, t represents the time interval set according to the monitoring requirements, p represents the pain level of uterine contraction at the corresponding time point, if it is zero, it means that there is no uterine contraction at the current time point, h represents the heart rate at the corresponding time point, and b represents the blood pressure at the corresponding time point; Build the LSTM model architecture: Add K LSTM layers to the created sequential model, with R LSTM units in each layer, and return the sequence so that the next layer can continue to process the time series information; add a Dropout layer after each LSTM layer, and connect the last Dropout layer to a fully connected layer to output the prediction results; compile the model using the Adam optimizer and select the mean square error as the loss function.

5. The method for monitoring labor preparation according to claim 4, characterized in that: The model training step comprises: Perform sequence partitioning to split the input time series into input and output pairs, select the pain level, heart rate and blood pressure of uterine contractions corresponding to consecutive i time points as input, and use them to predict the pain level, heart rate and blood pressure of uterine contractions from i+a to i+b time points, where b>a; The construction of the input-output pair uses historical maternal vital signs information, including: maternal vital signs information of women who have given birth, maternal vital signs information of women who are currently expecting to give birth, or one or more thereof; Use input-output pairs for training, and obtain an LSTM model whose loss function meets the preset requirements as the production analysis model.

6. The method for monitoring labor preparation according to claim 2, characterized in that: The data analysis step comprises: S501, initialization judgment times k=1; S502, using a labor analysis model to analyze and predict the uterine contraction of the parturient within a preset time period, that is, the pain level within the preset time period, based on the parturient's physical sign information; S503, analyzing and calculating the duration and interval of uterine contractions according to the output results of the production analysis model; S504, determining whether the duration of uterine contraction is within the preset duration range, and whether the interval duration is within the preset interval duration range and is greater than the second preset interval duration, if so, executing S505; if not, executing S506; S505, extracting a time period in which the duration of uterine contraction is greater than the minimum preset duration in the preset duration range and the interval duration is less than the maximum preset interval duration in the preset interval duration range as the predicted delivery time, and executing S5014; S506, analyzing the relationship between the duration of uterine contractions and the preset duration range, and the relationship between the interval duration and the preset interval duration range; If the duration of the uterine contraction is greater than the preset duration range, and the interval duration is less than the preset interval duration range, execute S507; If the duration of the uterine contraction is less than the preset duration range, and the interval duration is greater than the preset interval duration range, execute S508; If the duration of the uterine contraction is within the preset duration range, and the interval duration is greater than the preset interval duration range, execute S5011; If the duration of the uterine contraction is less than the preset duration range, and the interval duration is within the preset interval duration range, execute S5012; If the duration of uterine contraction is within the preset duration range, and the interval duration is less than the preset interval duration range, execute S507; If the duration of the uterine contraction is greater than the preset duration range, and the interval duration is within the preset interval duration range, execute S507; If the duration of the uterine contraction is greater than the preset duration range, and the interval duration is greater than the preset interval duration range, execute S507; If the duration of the uterine contraction is less than the preset duration range, and the interval duration is less than the preset interval duration range, execute S507; S507, moving the minimum time point of the preset time period forward by a preset time step, that is, closer to the current time point, and executing S509; S508, moving the maximum time point of the preset time period backward by a preset time step, that is, away from the current time point, and executing S509; S509, update k=k+1; S5010, determine whether k is equal to the preset number of times, if yes, execute, if not, execute S502; S5011, extracting the time period in which the duration of uterine contraction is greater than the minimum preset duration in the preset duration range as the predicted delivery time, and executing S5014; S5012, extracting a time period whose intermittent duration is less than the maximum preset intermittent duration within the preset intermittent duration range as the predicted production time, and executing S5014; S5013, outputting the predicted uterine contraction conditions of the parturient in all preset time periods, and providing special reminders of the parturient's conditions; S5014. Output the preset production time.

7. The method for monitoring labor preparation according to claim 5, characterized in that: The model optimization step comprises: Calculating a loss function value of the predicted uterine contraction condition and the uterine contraction condition actually collected within the preset time period as a first loss function value; Determine whether the first loss function value is greater than the preset loss function value. If so, collect the maternal vital signs information before the current time point, build a data set, retrain the production analysis model, and calculate the loss function value after training as the second loss function value. If the second loss function value is less than or equal to the preset loss function value, the data analysis step adopts the retrained production analysis model; if not, adopt the optimization model to optimize the weights in the production analysis model, and use the loss function as the fitness evaluation function to obtain the optimal weight scheme and update the production analysis model.

8. The method for monitoring labor preparation according to claim 6, characterized in that: The production suggestion steps include: Calculate the time interval between the current time and the predicted delivery time. If the time interval is less than the preset time interval, generate and push a suggestion message to go to the hospital immediately or call 120. If the interval time length is greater than the preset interval time length, the production preparation suggestion information and the predicted production time are extracted as the suggestion information and pushed.

9. A device, characterized in that: Used to execute the labor monitoring method as described in any one of claims 1-8.

10. A labor monitoring system, used to execute the labor monitoring method according to any one of claims 1 to 8, comprising: Information monitoring module, used to collect maternal and fetal vital signs information; The central control module is used to determine whether there are any abnormalities in the maternal and fetal vital signs information. If so, the early warning module is triggered to issue an early warning; Early warning module, used for data anomaly early warning; It is characterized in that the central control module is also used to pre-process the maternal and fetal vital signs information; It is also used to build a production analysis model to analyze and predict the uterine contractions of parturients based on their physical signs. It is also used to build data sets using historical maternal vital signs information and delivery time to train delivery analysis models; It is also used to adopt a production analysis model to analyze and predict the uterine contractions of parturients based on their physical signs; and to analyze and predict the time of delivery based on the predicted uterine contractions; It is also used to compare and analyze the predicted uterine contraction conditions and the actual uterine contraction conditions collected within the preset time period, and optimize the production analysis model based on the analysis results; It is also used to generate and push advice information based on the predicted delivery time of the mother.

Images