Newborn acute respiratory distress syndrome antenatal prediction model and system

Abstract

The invention provides a neonatal acute respiratory distress syndrome antenatal prediction model, which is characterized in that 15 key variables for predicting neonatal acute respiratory distress syndrome are screened out by using an LASSO regression algorithm according to perinatal high-risk factors; and taking the key variables as input, taking the neonatal acute respiratory distress syndrome as output, and training by using a BP neural network to obtain the neonatal acute respiratory distress syndrome antenatal prediction model. According to the method, personalized early prediction of the NARDS can be realized, and the possibility of occurrence of the neonatal acute respiratory distress syndrome can be predicted before a fetus is born; in the prediction process, the examination result can be obtained only by inputting various existing examination data into the prediction model, the method is simple and easy to operate, visual, reliable and rapid, evaluation can be conducted at any time, and the requirements for professional knowledge of medical staff and hospital instruments and equipment are low.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence medical prediction technology, and in particular to a prenatal prediction model and system for neonatal acute respiratory distress syndrome. Background Art

[0002] In August 2017, the concept of neonatal acute respiratory distress syndrome (NARDS), also known as the Montreux criteria, was first proposed internationally. As one of the most serious neonatal lung diseases, severe hypoxemia and respiratory distress can lead to changes in systemic hemodynamics and tissue ischemia and hypoxia, causing multiple organ damage, a series of complications, and even death. These children are often seriously ill, progress rapidly, and have high mortality and disability rates. The specific cause and mechanism are still unclear. The current treatment status of NARDS is that even in the neonatal treatment centers of a few large tertiary hospitals or specialized hospitals in first-tier cities in China, the mortality rate of NARDS is as high as 30-40%. Most NARDS children in hospitals need to be transferred to hospitals with critical neonatal treatment capabilities for treatment. Due to the critical condition and rapid progression, the waiting time or transfer time is too long, which often leads to the worsening of the condition of the children, leaving serious complications or even death.

[0003] In view of the above situation, preparations for treatment should be made in advance for fetuses that may develop neonatal acute respiratory distress syndrome after birth. Therefore, there is an urgent need for a method to predict the possibility of neonatal acute respiratory distress syndrome before the fetus is born. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention proposes a prenatal prediction model and system for neonatal acute respiratory distress syndrome to address the deficiencies in the prior art.

[0005] The technical solution adopted by the present invention is as follows:

[0006] In a first aspect, a method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome is provided, comprising the following steps:

[0007] According to perinatal high-risk factors, the LASSO regression algorithm was used to screen out key variables for predicting neonatal acute respiratory distress syndrome;

[0008] Taking key variables as input and neonatal acute respiratory distress syndrome as output, an artificial neural network was used for training to obtain a prenatal prediction model for neonatal acute respiratory distress syndrome.

[0009] Furthermore, the perinatal high-risk factors include: maternal age, maternal education level, maternal occupation, maternal pre-pregnancy BMI, maternal weight gain, maternal drug abuse / smoking / drinking, oral folic acid / regular prenatal examinations, prenatal examination hospital level, delivery hospital level, whether embryo transfer, twins / multiple births, primipara / multipara, history of cesarean section, cesarean section / natural delivery, intrauterine growth retardation, threatened premature labor, premature rupture of membranes, chorioamnionitis, infectious diseases one week before delivery, white blood cell count, platelet count, hemoglobin, fetal distress, amniotic fluid turbidity, too much / too little amniotic fluid, umbilical cord abnormalities, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension and preeclampsia, thyroid dysfunction during pregnancy, anemia during pregnancy, pregnancy complications, antenatal glucocorticoid use, gestational age, arterial blood, standard residual base and lactate.

[0010] Furthermore, when using the LASSO regression algorithm for screening, the minimum lambda criterion was adopted.

[0011] Furthermore, the key variables include: the level of the delivery hospital, whether it is twins or multiple births, primipara / multipara, premature rupture of membranes, chorioamnionitis, platelets, polyhydramnios, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension / preeclampsia, anemia during pregnancy, complications during pregnancy, antenatal corticosteroid use and gestational age.

[0012] Furthermore, the artificial neural network is a BP neural network.

[0013] In a second aspect, a prenatal prediction model for acute respiratory distress syndrome in newborns is provided, which is constructed using the method for constructing a prenatal prediction model for acute respiratory distress syndrome in newborns described in the first aspect.

[0014] In a third aspect, a system for prenatal prediction of neonatal acute respiratory distress syndrome is provided, comprising:

[0015] A key variable indicator acquisition device is used to acquire a variety of key variables of pregnant women, including the delivery hospital level, whether it is twins or multiple births, primipara / multipara, premature rupture of membranes, chorioamnionitis, platelet count, polyhydramnios / oligohydramnios, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension / preeclampsia, anemia during pregnancy, complications during pregnancy, antenatal glucocorticoid use, and gestational age;

[0016] The device for predicting neonatal acute respiratory distress syndrome before delivery is communicatively connected to the key variable indicator acquisition device, and is used to predict the possibility of neonatal acute respiratory distress syndrome after the fetus is born based on multiple key variable data acquired by the key variable indicator acquisition device and using the prenatal prediction model for neonatal acute respiratory distress syndrome as described in claim 6.

[0017] In a fourth aspect, a computer program product is provided, comprising a computer program / instruction, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0018] In a fifth aspect, a computer-readable storage medium storing a computer program is provided, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0019] It can be seen from the above technical solution that the beneficial technical effects of the present invention are as follows:

[0020] 1. It can realize personalized early prediction of NARDS and predict the possibility of neonatal acute respiratory distress syndrome before the fetus is born.

[0021] 2. The prediction process only requires inputting the existing examination data into the prediction model to obtain the examination results. It is simple, easy to operate, intuitive, reliable, fast, and can be evaluated at any time. It also has low requirements on the professional knowledge of medical staff and hospital instruments and equipment.

[0022] 3. At the same time, prenatal intervention can be carried out on high-risk pregnant women based on early prediction results, and early prevention and treatment strategies can be established to reduce the incidence and mortality of NARDS. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the prior art description. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0024] Figure 1 This is a schematic flow chart of a method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome in Example 1 of the present invention;

[0025] Figure 2 It is the LASSO regression variable shrinkage diagram in Example 1 of the present invention;

[0026] Figure 3 This is the CV diagram of the LASSO regression screening variables in Example 1 of the present invention;

[0027] Figure 4This is a schematic diagram of the BP neural network architecture in Example 1 of the present invention. DETAILED DESCRIPTION

[0028] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0029] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0030] Example 1

[0031] After research, the inventors found that the mother's condition before and during pregnancy has a huge impact on the fetus / newborn. For example, underlying diseases before pregnancy, infectious diseases one week before delivery, fetal distress, etc. are all independent risk factors for the development of ARDS. If combined with adverse conditions, it may lead to respiratory distress in the newborn after birth.

[0032] In view of the above research results, this embodiment provides a prenatal prediction model for neonatal acute respiratory distress syndrome, and the method for constructing the model includes the following steps:

[0033] S1. Based on perinatal high-risk factors, use LASSO regression algorithm to screen out key variables for predicting neonatal acute respiratory distress syndrome

[0034] The following table lists the perinatal high risk factors:

[0035] Table 1 Perinatal high-risk factors

[0036]

[0037]

[0038]

[0039] For the perinatal high-risk factors in the above table, the following explanations are given:

[0040] Umbilical cord abnormalities (coded as X30 in Table 1) include being too long or too short, or having edema, torsion or kinking.

[0041] Placental abnormalities (coded as X31 in Table 1) include placental abruption, placenta previa, or abnormal placental shape, size, and weight.

[0042] Pregnancy complications (coded as X39 in Table 1) refer to underlying diseases that exist before pregnancy, such as pregnancy complicated by congenital heart disease, pregnancy complicated by hyperthyroidism, etc.

[0043] Other diseases complicated by pregnancy (coded as X37 in Table 1) refer to diseases caused by pregnancy other than the pregnancy complications listed in the included risk factors.

[0044] The diagnosis of each physiological index in the above table is based on the standard definition of the disease by obstetricians, including: white blood cell count, platelet count, hemoglobin, gestational age, arterial blood pH, standard residual base, lactic acid and other indicators as quantitative indicators, and the actual test values ​​are the standard.

[0045] When using the LASSO (Least absolute shrinkage and selection operator) regression algorithm to screen key variables, the clinical data were randomly split into a training set (n = 1437) and a test set (n = 617) in a ratio of 7:3. The screening process is as follows: Figure 2 , Figure 3 As shown in the figure, 15 key variables were screened out according to the minimum lambda standard, and the relationship between these 15 key variables and neonatal ARDS was analyzed in combination with clinical details:

[0046] (1) Delivery hospital level (coded as X12 in Table 1)

[0047] If the delivery hospital is a tertiary hospital, the prevention and treatment conditions for neonatal ARDS during pregnancy are significantly better than those in secondary and lower hospitals. Therefore, the probability of neonatal ARDS is higher if you choose "2 secondary hospitals and below".

[0048] (2) Twins / multiple births (coded as X14 in Table 1)

[0049] Compared with single pregnancy, twin / multiple pregnancy is more prone to premature birth and fetal complications, so the probability of neonatal ARDS is higher if you choose "1 yes".

[0050] (3) Primipara / multipara (coded as X15 in Table 1)

[0051] Compared with primiparas, multiparas have more experience in self-management, health care and awareness of danger signals during pregnancy. Therefore, the probability of neonatal ARDS in primiparas is higher.

[0052] (4) Premature rupture of membranes (coded as X21 in Table 1)

[0053] Pregnant women with premature rupture of membranes are more likely to develop secondary intrauterine infection, or premature rupture of membranes itself may be the result and early warning of intrauterine infection. Therefore, choosing "1 yes" has a higher probability of neonatal ARDS.

[0054] (5) Chorioamnionitis (coded as X22 in Table 1)

[0055] Chorioamnionitis may lead to intrauterine infection of the fetus, which in turn may cause neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is higher.

[0056] (6) Platelet count (coded as X25 in Table 1)

[0057] Changes in platelet count may be the result of infection or blood system disease in pregnant women, which in turn leads to the occurrence of ARDS in neonates.

[0058] (7) Polyhydramnios / oligohydramnios (coded as X29 in Table 1)

[0059] Excessive amniotic fluid may be the result of fetal malformation, while insufficient amniotic fluid may lead to fetal intrauterine hypoxia, which in turn leads to the occurrence of neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is greater.

[0060] (8) Placental abnormalities (coded as X31 in Table 1)

[0061] Placental abnormalities may cause fetal intrauterine hypoxia, which in turn leads to neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is greater.

[0062] (9) Intrahepatic cholestasis of pregnancy (coded as X32 in Table 1)

[0063] The high bile acid level in pregnant women with ICP is one of the mechanisms that cause fetal lung damage, so choosing "1 is yes" increases the probability of neonatal ARDS.

[0064] (10) Gestational diabetes (coded as X33 in Table 1)

[0065] GDM in pregnant women may cause fetal lung immaturity and respiratory distress, so choosing "1 yes" increases the probability of neonatal ARDS.

[0066] (11) Hypertension during pregnancy, preeclampsia (coded as X34 in Table 1)

[0067] Gestational hypertension / preeclampsia may lead to intrauterine growth retardation and intrauterine hypoxia, which in turn may lead to neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is higher.

[0068] (12) Anemia during pregnancy (coded as X36 in Table 1)

[0069] Anemia during pregnancy may lead to fetal growth restriction and fetal intrauterine hypoxia, which in turn may lead to neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is greater.

[0070] (13) Pregnancy complications (coded as X39 in Table 1)

[0071] Complications during pregnancy may lead to intrauterine growth retardation and intrauterine hypoxia, which in turn may lead to neonatal ARDS. Therefore, if you choose "1 yes", the probability of neonatal ARDS is higher.

[0072] (14) Antenatal glucocorticoid use (coded as X40 in Table 1)

[0073] Antenatal administration of glucocorticoids can promote fetal lung maturation, anti-inflammatory, and promote lung fluid clearance, so the probability of neonatal ARDS is higher if you choose (2 No).

[0074] (15) Gestational age (coded as X41 in Table 1)

[0075] Gestational age affects the maturity of fetal lung development and thus plays a role in the occurrence of neonatal ARDS.

[0076] S2. Take the key variables as input and neonatal acute respiratory distress syndrome as output, use artificial neural network for training, and obtain the prenatal prediction model of neonatal acute respiratory distress syndrome.

[0077] In a specific implementation, the artificial neural network uses a BP (back propagation) neural network. The BP neural network is a multi-layer feedforward neural network trained by error back propagation, using the BP algorithm (back propagation algorithm) and gradient search technology to minimize the mean square error between the actual output value and the expected output value of the network.

[0078] The structure of the BP neural network in this embodiment is as follows Figure 4 As shown, it includes an input layer, a hidden layer and an output layer, and constructs 15 key variables: "delivery hospital level, whether twins or multiple births, primipara / multipara, premature rupture of membranes, chorioamnionitis, platelet count, polyhydramnios / oligohydramnios, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension / preeclampsia, anemia during pregnancy, complications during pregnancy, antenatal glucocorticoid application, gestational age" data into a data set and writes it into the input layer. After calculation by the hidden layer, the output layer outputs the possibility of NARDS occurring in pregnant women in the late pregnancy before delivery after the birth of their fetus as the calculation result of the BP neural network. The calculated possibility of NARDS is divided into three types: low, general, and high. In a specific embodiment, the hidden layer is 3 layers, and the activation function can be selected from Sigmoid or RELU. The constructed antenatal prediction model for neonatal acute respiratory distress syndrome has a prediction accuracy as shown in Table 2 below, and the prediction accuracy is 93.68%.

[0079] Table 2

[0080]

[0081] The neonatal acute respiratory distress syndrome prenatal prediction model constructed in this embodiment is mainly aimed at near-term / full-term fetuses, with a gestational age of 34 weeks and above being the focus, because premature infants generally attract attention, while these near-term and full-term newborns are easily neglected, resulting in serious consequences.

[0082] By using the prenatal prediction model for neonatal acute respiratory distress syndrome provided by the embodiment, personalized early prediction of NARDS can be achieved, and the possibility of neonatal acute respiratory distress syndrome can be predicted before the fetus is born. The prediction process only needs to input the existing various examination data into the prediction model to obtain the examination results. It is simple and easy to operate, intuitive, reliable, fast, and can be evaluated at any time, and has low requirements for the professional knowledge of medical staff and hospital equipment; at the same time, prenatal intervention can be carried out on high-risk pregnant women according to the early prediction results, and early prevention and treatment strategies can be constructed to reduce the incidence and mortality of NARDS.

[0083] Example 2

[0084] Based on the prenatal prediction model of neonatal acute respiratory distress syndrome constructed in Example 1, this embodiment provides a prenatal prediction system for neonatal acute respiratory distress syndrome, including: a key variable indicator acquisition device and a prenatal prediction device for neonatal acute respiratory distress syndrome.

[0085] The key variable indicator acquisition device is used to acquire 15 key variables. In a specific implementation, the key variable indicator acquisition device is implemented in any form, and can be a tablet computer, a mobile phone, etc.; for example:

[0086] The level of the delivery hospital, whether it is twin or multiple pregnancy, primipara / multipara, and gestational age can be input into the computer and stored through a questionnaire survey;

[0087] The data of the pregnancy test results, such as whether the fetal membranes are ruptured prematurely, whether there is chorioamnionitis, platelet count, too much / too little amniotic fluid, whether there is placental abnormality, whether there is intrahepatic cholestasis of pregnancy, whether there is gestational diabetes, whether there is gestational hypertension / preeclampsia, whether there is anemia during pregnancy, whether there are pregnancy complications, and the use of antenatal corticosteroids, can be entered into the computer and stored.

[0088] The antenatal prediction device for neonatal acute respiratory distress syndrome is communicatively connected to the key variable indicator acquisition device, and is used to predict the possibility of neonatal acute respiratory distress syndrome after the fetus is born using the antenatal prediction model for neonatal acute respiratory distress syndrome based on the 15 key variable data acquired by the key variable indicator acquisition device.

[0089] In some embodiments, a computer program product is also provided, including a computer program / instruction, which, when executed by a processor, implements the steps of the method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome described in Example 1.

[0090] In some embodiments, a computer-readable storage medium storing a computer program is also provided. When the computer program is executed by a processor, the steps of the method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome described in Example 1 are implemented.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome, characterized in that: The following steps are involved: According to perinatal high-risk factors, the LASSO regression algorithm was used to screen out key variables for predicting neonatal acute respiratory distress syndrome; Taking key variables as input and neonatal acute respiratory distress syndrome as output, an artificial neural network was used for training to obtain a prenatal prediction model for neonatal acute respiratory distress syndrome.

2. The method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome according to claim 1, characterized in that: The perinatal high-risk factors include: Maternal age, maternal education level, maternal occupation, maternal BMI before pregnancy, maternal weight gain, maternal drug abuse / smoking / drinking, oral folic acid / regular prenatal check-ups, prenatal check-up hospital level, delivery hospital level, whether embryo transfer is performed, twins / multiple births, primipara / multipara, history of cesarean section, cesarean section / natural birth, intrauterine growth retardation, threatened premature labor, premature rupture of membranes, chorioamnionitis, infectious diseases one week before delivery, white blood cells, platelets, hemoglobin, fetal distress, amniotic fluid turbidity, too much / too little amniotic fluid, umbilical cord abnormalities, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension and preeclampsia, thyroid dysfunction during pregnancy, anemia during pregnancy, pregnancy complications, antenatal glucocorticoid use, gestational age, arterial blood, standard residual base and lactate.

3. The method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome according to claim 1, characterized in that: When using the LASSO regression algorithm for screening, the minimum lambda criterion is adopted.

4. The method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome according to claim 1, characterized in that: The key variables include: The level of the delivery hospital, whether it is twins or multiple births, primipara / multipara, premature rupture of membranes, chorioamnionitis, platelets, polyhydramnios, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension / preeclampsia, anemia during pregnancy, complications during pregnancy, use of antenatal corticosteroids and gestational age.

5. The method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome according to claim 1, characterized in that: The artificial neural network is a BP neural network.

6. A prenatal prediction model for neonatal acute respiratory distress syndrome, characterized in that: The method for constructing a prenatal prediction model for neonatal acute respiratory distress syndrome according to any one of claims 1 to 5 is used to construct the model.

7. A prenatal prediction system for neonatal acute respiratory distress syndrome, characterized in that: include: A key variable indicator acquisition device is used to acquire a variety of key variables of pregnant women, including the level of the delivery hospital, whether it is twins or multiple births, primipara / multipara, premature rupture of membranes, chorioamnionitis, platelets, polyhydramnios, placental abnormalities, intrahepatic cholestasis of pregnancy, gestational diabetes, gestational hypertension / preeclampsia, anemia during pregnancy, complications during pregnancy, antenatal glucocorticoid use, and gestational age; The device for predicting neonatal acute respiratory distress syndrome before delivery is communicatively connected to the key variable indicator acquisition device, and is used to predict the possibility of neonatal acute respiratory distress syndrome after the fetus is born based on multiple key variable data acquired by the key variable indicator acquisition device and using the prenatal prediction model for neonatal acute respiratory distress syndrome as described in claim 6.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images