Application of lipid marker in preparation of product for diagnosing or predicting gestational diabetes mellitus

Abstract

The invention provides an application of a lipid marker in preparation of a product for diagnosing or predicting gestational diabetes mellitus. According to the application, the lipid marker comprises any one or more of the following lipid molecules: C59H102O6N1, C55H104O6N1, C65H112O6N1, C59H110O6N1, C36H63O4, C46H84O8N1P1Na1, C52H85O8N1P1, C50H85O8N1P1, C45H87O13N1P1, C41H84O6N2P1 or C44H85O8N1P1 and the like, and the lipid marker can be used for preparing the lipid marker. Or a corresponding lipid metabolism protein of the lipid molecule. The problem that products for diagnosing or predicting gestational diabetes in the prior art are poor in effect can be solved, and the method is suitable for the field of gestational diabetes diagnosis product application.

Description

Technical Field

[0001] The present invention relates to the application field of gestational diabetes diagnosis products. Specifically, it relates to the application of a lipid biomarker in the preparation of a product for diagnosing or predicting gestational diabetes. Background Art

[0002] Abnormal glucose metabolism during pregnancy, also known as gestational diabetes mellitus (GDM), refers to abnormal blood glucose levels during pregnancy. The presence of GDM significantly increases the risk of adverse pregnancy outcomes. These adverse pregnancy outcomes include cesarean section, shoulder dystocia, birth injury, pregnancy-induced hypertension (including preeclampsia), etc., and may have an impact on the long-term cardiovascular metabolism of the parturient. In addition, the probability that women with GDM will develop type 2 diabetes after childbirth is about 50%. Therefore, the diagnosis and management of GDM are crucial for ensuring the health of mothers and infants. Early detection and active management of GDM can help reduce the risk of adverse pregnancy outcomes and the likelihood of patients developing type 2 diabetes in the future [1]. The results of randomized clinical trials show that intervening in pregnant women at high risk of abnormal glucose metabolism (GDM) in the early stage of pregnancy can significantly reduce the incidence of GDM and the incidence of maternal and fetal complications, while improving pregnancy outcomes. This indicates that it is very necessary to conduct risk assessment and intervention on GDM in the early and mid-pregnancy. Risk assessment of GDM can be carried out by identifying pregnant women with relevant risk factors, such as obesity, family history of diabetes, previous history of GDM, etc. By conducting risk assessment on these high-risk pregnant women in the early and mid-pregnancy, early intervention measures can be taken to provide personalized management and monitoring to prevent or mitigate the development and adverse consequences of GDM [2].

[0003] Although there have been some studies on the diagnosis and prediction of GDM, the accuracy of the existing diagnostic or predictive methods for GDM is relatively low, making it difficult to meet the clinical needs. Moreover, abnormal glucose metabolism during pregnancy is a complex disease affected by multiple factors, and single clinical information is difficult to accurately reflect the current pregnancy situation. Summary of the Invention

[0004] The main object of the present invention is to provide the application of a lipid biomarker in the preparation of a product for diagnosing or predicting gestational diabetes, so as to solve the problem of poor effectiveness of the existing products for diagnosing or predicting gestational diabetes.

[0005] To achieve the above object, according to the first aspect of the present invention, there is provided the application of a lipid biomarker in the preparation of a product for diagnosing or predicting gestational diabetes, and the lipid biomarker includes any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolism proteins of the lipid molecules.

[0006] Furthermore, the lipid markers include the following combinations of lipid molecules: A) C59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92O8N2P1; or B)C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1.

[0007] Furthermore, the product is used to detect the content of lipid markers in a biological sample of an individual; preferably, the detection method for the content of lipid markers includes mass spectrometry detection, nuclear magnetic resonance detection or probe detection; preferably, the content includes the relative abundance of mass spectrometry; preferably, the product includes a standard, and the standard is a composition formed by any one or more lipid markers, or a set of pure standard products formed by separately setting any one or more lipid markers; preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

[0008] To achieve the above object, according to the second aspect of the present invention, there is provided a kit for diagnosing or predicting pregnancy-induced hypertension in a subject, the kit including a detection reagent for detecting the content of lipid markers, and the lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolism proteins of the lipid molecules.

[0009] Furthermore, the lipid markers include the following combinations of lipid molecules: A) C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B) C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1; Preferably, the detection reagent includes a standard, and the standard is a composition formed by any one or more lipid molecules, or a collection of pure standard products formed by separately setting any one or more lipid molecules; Preferably, the detection reagent includes a probe capable of specifically binding to a lipid marker, and the probe contains a group for detection; Preferably, the group for detection includes a fluorescent group, an isotope group, an enzyme group, or a biotin group.

[0010] To achieve the above object, according to the third aspect of the present invention, a screening method for lipid markers for diagnosing or predicting pregnancy-induced hypertension is provided. The screening method includes: a) obtaining the clinical status and biological samples of a sample population, detecting the biological samples to obtain various lipid molecules and the contents of lipid molecules in the biological samples. The sample population includes healthy pregnant women and pregnant women with pregnancy-induced hypertension. Correspondingly, the clinical status includes being healthy and having pregnancy-induced hypertension; b) screening significantly different lipid molecules that are different between pregnant women with pregnancy-induced hypertension and healthy pregnant women according to the lipid molecules and the contents of lipid molecules through statistical methods; preferably, the statistical methods include t-test, generalized linear model, rank test, logistic regression, fold change, multiple hypothesis testing correction or Kruskal-Wallis test; preferably, b) includes: using a logistic regression model, with the lipid molecules and the contents of lipid molecules as independent variables and the clinical status as the dependent variable, to screen and obtain significantly different lipid molecules.

[0011] To achieve the above object, according to the fourth aspect of the present invention, a method for constructing a model for diagnosing or predicting pregnancy-induced hypertension is provided. The construction method uses the lipid markers obtained by the above screening method for lipid markers for diagnosing or predicting pregnancy-induced hypertension and applies machine learning methods to construct a model for diagnosing or predicting pregnancy-induced hypertension.

[0012] Furthermore, the construction method uses lipid markers and clinical parameters to construct a model for diagnosing or predicting pregnancy-induced hypertension; preferably, the clinical parameters include one or more of the age, BMI, smoking status or drinking status of the sample population; preferably, machine learning includes logistic regression screening method and / or neural network classifier model.

[0013] To achieve the above object, according to the fifth aspect of the present invention, an electronic device for diagnosing or predicting pregnancy-induced hypertension is provided. The electronic device includes: an acquisition and detection module, which is configured to acquire a biological sample, detect the biological sample, and obtain the content of lipid markers in the biological sample; a diagnosis or prediction module, which has a model for diagnosing or predicting pregnancy-induced hypertension built in. The diagnosis or prediction module is configured to input the content of lipid markers into the model for diagnosing or predicting pregnancy-induced hypertension and output a prediction result according to the model for diagnosing or predicting pregnancy-induced hypertension.

[0014] Furthermore, the lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolic proteins of the lipid molecules.

[0015] Furthermore, the lipid markers include the following combinations of lipid molecules: A) C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B) C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1; Preferably, the diagnostic or predictive model for gestational hypertension is the diagnostic or predictive model for gestational hypertension obtained by using the above construction method. Preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

[0016] To achieve the above object, according to the sixth aspect of the present invention, there is provided a computer-readable storage medium, which includes a stored program. When the program runs, it controls the device where the storage medium is located to execute the above screening method for lipid markers for diagnosing or predicting gestational hypertension, the above construction method for a diagnostic or predictive model for gestational hypertension, or a diagnostic or predictive model for gestational hypertension obtained by using the above construction method for a diagnostic or predictive model for gestational hypertension.

[0017] To achieve the above object, according to the seventh aspect of the present invention, there is provided an electronic device, which is used to run a program. When the program runs, it executes the above screening method for lipid markers for diagnosing or predicting gestational hypertension, the above construction method for a diagnostic or predictive model for gestational hypertension, or a diagnostic or predictive model for gestational hypertension obtained by using the above construction method for a diagnostic or predictive model for gestational hypertension.

[0018] By applying the technical solution of the present invention, a batch of lipid markers capable of diagnosing or predicting gestational hypertension have been screened. By using one or more of the above lipid markers, a product for diagnosing or predicting gestational diabetes can be prepared, which can assist in diagnosing or predicting gestational hypertension and provide a more definite and effective diagnostic result for the subject. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The specification drawings forming a part of this application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0020] Figure 1 Shows the prediction flow chart of the lipidome for early pregnancy GDM according to Embodiment 1 of the present invention.

[0021] Figure 2 Shows the schematic diagram of the neural network model of the lipidome for early pregnancy GDM according to Embodiment 1 of the present invention.

[0022] Figure 3 Shows the prediction AUC effect diagram of the lipidome and clinical data for early pregnancy GDM according to Embodiment 2 of the present invention.

[0023] Figure 4A hardware structure block diagram for diagnosing or predicting pregnancy-induced hypertension according to an embodiment of the present invention is shown. Detailed implementation manners

[0024] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present invention will be described in detail below in conjunction with the embodiments.

[0025] As mentioned in the background art, due to the simplicity and easy availability of clinical test data, there are many representative works on the prediction of GDM based on sociodemographic characteristics, clinical variables in the early pregnancy, and laboratory indicators, and preliminary prediction effects have been achieved. Some studies on the prediction of gestational glucose metabolism disorders (GDM) in the early and middle pregnancy, some of which focus on using clinical information for prediction, while others are based on studies in aspects such as metagenomics, proteomics, and genomics. The prediction model based on clinical information can utilize the clinical characteristics and test results of pregnant women, such as age, body mass index (BMI), blood glucose level, etc., to establish a prediction model. These models can help doctors evaluate the risk of GDM in the early and middle pregnancy, so as to take appropriate intervention measures. In addition, studies based on metagenomics, proteomics, and genomics are also exploring the potential in predicting GDM in the early and middle pregnancy. These studies can analyze the gene expression, protein composition, and metagenomic characteristics of pregnant women to search for biomarkers or genetic variations related to the risk of GDM. These studies are expected to provide more comprehensive and accurate prediction models, further improving the early identification and intervention of GDM [3]. However, there are relatively few studies on the lipidome in the prediction of GDM.

[0026] Yan-Ting Wu et.al. analyzed the medical records of 16,819 women diagnosed with gestational diabetes mellitus, and screened 7 features from sociodemographic characteristics, clinical variables in the early pregnancy, and laboratory indicators based on the K-NearestNeighbor algorithm, and its prediction AUC was 0.77 [4].

[0027] Hou, G. et.al from the BGI Research Institute collected samples from 100 GDM patients and 100 normal controls in three trimesters respectively. They found a total of 7 triglycerides and 5 diglycerides with significant quantitative differences in the first and second trimesters, and the prediction AUC for early pregnancy GDM could reach 0.88 [5]. The main problem with this study is that the single linear alignment based on differential expression may miss important non-linear differential characteristic molecules. Therefore, there is still a huge room for improvement.

[0028] Currently, there are also studies on predicting GDM from the microbial genome. After collecting saliva from 44 GDM samples for 16S rRNA sequencing, Jinfeng Wang et al. predicted based on two oral microbes, Lautropia Neisseria and Veillonella, with an AUC of 0.83[6].

[0029] Although there have been some studies on the diagnosis and prediction methods for gestational diabetes mellitus in the prior art, the diagnosis and prediction methods obtained in the prior art still have difficulty meeting the requirements.

[0030] In the prediction study of gestational glucose metabolism disorders (GDM), there are still some challenges, including the high cost and immature technology of proteomic detection methods, the poor prediction results of clinical information, and the simplicity of lipidomic prediction and screening methods in the prior art, which may miss important features.

[0031] 1) High cost and immature technology of proteomic detection methods: Proteomic analysis requires high costs and complex experimental techniques. Currently, high-throughput methods for protein detection are still under development, with technical problems and challenges. This may limit the application of proteomics in GDM prediction and make it relatively less used in clinical practice.

[0032] 2) Poor prediction results of clinical information: Although clinical information plays a certain role in GDM prediction, the current prediction results are relatively poor. This may be because gestational glucose metabolism disorders are a complex disease affected by multiple factors. As shown in the study by Yan-Ting Wu et al., single clinical information is difficult to accurately reflect the current pregnancy situation. Therefore, more research efforts are needed, such as adding new omics data, to improve and optimize the prediction model of clinical information.

[0033] 3) Simple lipidomic prediction and screening methods in the prior art: In past studies, the prediction and screening methods for lipidomics were relatively simple and may have overlooked some important features. Lipid metabolism plays an important role in gestational glucose metabolism disorders, but there is still room for further development in the detailed analysis and screening of lipidomic features in current research. More complex and comprehensive lipidomic analysis methods may help improve the accuracy and reliability of the prediction model.

[0034] In this application, the inventors attempted to screen the lipid molecules of the subjects and found that biomarkers for gestational diabetes could be identified from the perspective of lipid molecules. By obtaining the content of lipid markers and calculating the prediction probability through a model, it is possible to assist in the diagnosis of gestational diabetes or predict the risk of developing gestational diabetes, providing more definite and effective diagnostic results for the subjects. Based on the above lipid markers related to gestational diabetes, by using one or more of these lipid markers, products for diagnosing or predicting gestational diabetes can be prepared to predict the condition of gestational hypertension or the risk of developing gestational hypertension in the subjects. Therefore, a series of protection schemes for this application are proposed.

[0035] In the first typical embodiment of this application, there is provided an application of a lipid marker in the preparation of a product for diagnosing or predicting gestational diabetes, and the lipid marker includes any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolism proteins of lipid molecules.

[0036] The lipidome refers to the composition and metabolic characteristics of lipids in the body, which is closely related to abnormal glucose metabolism. Therefore, it is of great significance to further study and explore the potential of the lipidome in the prediction of GDM. Generally speaking, not limited to clinical information, research at multiple levels such as metagenomics, proteomics, genomics, and lipidomics is expected to provide a more accurate and comprehensive prediction model for predicting whether pregnant women in the early and mid - pregnancy (0 - 12 weeks) will develop into GDM after a period of pregnancy, and provide more information and tools for early intervention and management.

[0037] The above lipid biomarker or combination of lipid biomarkers has the ability to prepare products for diagnosing or predicting gestational diabetes. Using the above products, it is possible to determine whether a subject has gestational diabetes or predict the risk of developing gestational diabetes. Using any one of the above lipid biomarkers, or a combination of any several lipid biomarkers, can achieve the diagnosis or prediction of gestational diabetes and has the ability to prepare such products.

[0038] Correspondingly, different lipid metabolism proteins are closely related to the contents of their corresponding lipid markers. The higher the content of a lipid metabolism protein, the lower the content of the lipid molecules it metabolizes in a biological sample. Therefore, the corresponding lipid metabolism proteins of the above lipid markers can also be used for the diagnosis or prediction of gestational diabetes, and thus can be applied to the preparation of products for diagnosing or predicting gestational diabetes. In the prior art, methods including but not limited to Olink technology, immunoblotting experiments, or enzyme-linked immunosorbent assays can all detect the above lipid metabolism proteins, and related products are all within the protection scope of the above claims. Using the above products, it is possible to predict whether a pregnant woman in the early and middle trimesters will develop into GDM in the late trimester after a certain period of pregnancy. In this application, the early trimester of pregnancy refers to the first 12 weeks of pregnancy, starting from the implantation of the fertilized egg. The middle trimester of pregnancy refers to the 13th to 26th weeks of pregnancy. The late trimester of pregnancy refers to the 27th week of pregnancy until delivery.

[0039] In a preferred embodiment, the lipid markers include a combination of the following lipid molecules: A) C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B) C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1.

[0040] The combination of the above 27 lipid molecules or the combination of 5 lipid molecules applied simultaneously in the above product can achieve the best prediction effect. Using the combination of the above 27 or 5 lipid molecules can also achieve the diagnosis or prediction of gestational diabetes.

[0041] In a preferred embodiment, the product is used to detect the content of lipid markers in a biological sample of an individual; preferably, the detection method for the content of lipid markers includes liquid chromatography detection, mass spectrometry detection, nuclear magnetic resonance detection or probe detection; preferably, the content includes the relative abundance of mass spectrometry; preferably, the product includes a standard, and the standard is a composition formed by any one or more lipid molecules, or a set of pure standard products formed by each of any one or more lipid molecules separately set; preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

[0042] All methods for detecting lipid content in the prior art are applicable to this application. Accordingly, detection reagents are selected according to different detection methods. For example, detection reagents corresponding to chromatographic detection, mass spectrometric detection or nuclear magnetic resonance detection may include standards. Such standards are compositions formed by any one or more of the above lipid molecules, or a set of pure product standards formed by separately setting any one or more of the above lipid molecules, and are used to calibrate the signal characteristics, content characteristics, etc. of lipid markers in chromatographic, mass spectrometric or nuclear magnetic resonance detection. The above pure product standards are products formed by a certain lipid molecule with a purity meeting the requirements of the standard in an independent package. The above chromatographic detection includes, but is not limited to, qualitative and / or quantitative detection of lipid markers using existing technologies such as HPLC, UPLC, GC, etc. The above nuclear magnetic resonance detection includes, but is not limited to, qualitative and / or quantitative detection of lipid markers using nuclear magnetic resonance hydrogen spectrum, nuclear magnetic resonance carbon spectrum or other types. When detecting the content of the above lipid markers by mass spectrometric detection, the content of the lipid markers includes the relative abundance of each lipid molecule in the mass spectrometric detection data.

[0043] For the above probe detection and qualitative or quantitative detection of lipid markers using probes, in an optional embodiment, it includes detecting using a probe capable of specifically binding to a lipid marker, and the probe is further connected to a group for labeling, including but not limited to a fluorescent group, an isotope group, an enzyme group, a biotin group, etc., all of which can achieve the detection of lipid markers.

[0044] In the second typical embodiment of this application, a mass spectrometry test method for diagnosing or predicting pregnancy-induced hypertension in a subject is provided. The method includes a detection reagent for detecting the content of a lipid marker, and the lipid marker includes any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolism proteins of the lipid molecules.

[0045] In a preferred embodiment, the lipid markers include a combination of the following lipid molecules: A) C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B) C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92O8N2P1; Preferably, the detection reagent includes a standard product, which is a composition formed by any one or more lipid molecules, or a collection of pure product standards formed by separately setting any one or more lipid molecules; Preferably, the detection reagent includes a probe that can specifically bind to a lipid biomarker, and the probe contains a group for detection; Preferably, the group for detection includes a fluorescent group, an isotope group, an enzyme group, or a biotin group.

[0046] In the third typical embodiment of the present application, a screening method for lipid biomarkers for diagnosing or predicting pregnancy-induced hypertension is provided. The screening method includes: a) obtaining the clinical status and biological samples of a sample population, detecting the biological samples to obtain various lipid molecules and the contents of lipid molecules in the biological samples. The sample population includes healthy pregnant women and pregnant women with pregnancy-induced hypertension. Correspondingly, the clinical status includes health and pregnancy-induced hypertension; b) according to the lipid molecules and the contents of lipid molecules, screening for significantly different lipid molecules that are different between pregnant women with pregnancy-induced hypertension and healthy pregnant women by statistical methods.

[0047] In the above screening method for lipid biomarkers, first, the biological samples and the condition of pregnancy-induced hypertension of a sample population including healthy people and patients with pregnancy-induced hypertension are obtained. Lipid detection is performed on the biological samples to obtain the contents of different lipid compounds in each biological sample. Statistical methods are used to test the lipids that are different between patients with pregnancy-induced hypertension and healthy people, which are significantly different lipid molecules. The amount of these lipids present in the sample population is related to the disease condition. There are significant differences in the amounts of significantly different lipids in the bodies of healthy people and patients with pregnancy-induced hypertension, but it is impossible to confirm the specific effects of significantly different lipids on diagnosis or prediction, nor is it clear how to perform diagnosis or prediction. In order to ensure and improve the effect of diagnosing or predicting pregnancy-induced hypertension, methods such as using a logistic regression parameter screening model are used to screen according to the contents of significantly different lipids and the disease condition of the sample population, and lipids with large significant differences are screened out.

[0048] In a preferred embodiment, b) includes: using a logistic regression model, with lipid molecules and the contents of lipid molecules as independent variables and clinical status as the dependent variable, screening for significantly different lipid molecules.

[0049] The above statistical methods for determining significantly different lipid molecules include, but are not limited to, one or more of T-test, Wilcoxon test, logistic regression, KW test (Kruskal-Wallis test), FoldChange, Bonferroni correction, or other existing statistical methods. In the embodiments of the present application, a logistic regression model is used to screen for significantly different lipid molecules, and the above 27 most significant lipid molecules are obtained.

[0050] In the fourth typical embodiment of the present application, a method for constructing a diagnostic or predictive model for pregnancy-induced hypertension is provided. The above construction method uses the lipid markers obtained by the above screening method for diagnostic or predictive lipid markers for pregnancy-induced hypertension, and applies machine learning methods to construct a diagnostic or predictive model for pregnancy-induced hypertension.

[0051] In a preferred embodiment, the construction method uses lipid markers and clinical parameters to construct a diagnostic or predictive model for pregnancy-induced hypertension; preferably, the clinical parameters include one or more of the age, BMI, smoking status, or drinking status of the sample population; preferably, machine learning includes logistic regression screening and / or neural network classification models.

[0052] The smoking status in the above clinical parameters includes smoking or not smoking, and may further include the degree of smoking, including but not limited to the number of years of smoking, the number of cigarettes smoked per day, the smoking index, etc.; similarly, the drinking status includes drinking or not drinking, and may further include the degree of drinking, including but not limited to the number of years of drinking, the daily alcohol consumption, etc.

[0053] In the fifth typical embodiment of the present application, an electronic device for diagnosing or predicting pregnancy-induced hypertension is provided. The electronic device includes: an acquisition and detection module configured to acquire a biological sample, detect the biological sample, and obtain the content of lipid markers in the biological sample; a diagnosis or prediction module with a built-in diagnostic or predictive model for pregnancy-induced hypertension, and the diagnosis or prediction module is configured to input the content of the lipid markers into the diagnostic or predictive model for pregnancy-induced hypertension and output a prediction result according to the diagnostic or predictive model for pregnancy-induced hypertension.

[0054] In a preferred embodiment, the lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or the corresponding lipid metabolism protein of the lipid molecule; preferably, the lipid marker includes a combination of the following lipid molecules: A) C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B) C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C47 H 80 O 12 P1 and C 45 H 92 O8N2P1; Preferably, the diagnostic or predictive model for pregnancy-induced hypertension is the diagnostic or predictive model for pregnancy-induced hypertension obtained by using the above-mentioned method for constructing a diagnostic or predictive model for pregnancy-induced hypertension.

[0055] Preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

[0056] In the sixth typical embodiment of the present application, a computer-readable storage medium is provided. The storage medium includes a stored program. When the program runs, it controls the device where the storage medium is located to execute the above-mentioned method for screening lipid markers for diagnosing or predicting pregnancy-induced hypertension, the above-mentioned method for constructing a diagnostic or predictive model for pregnancy-induced hypertension, or the diagnostic or predictive model for pregnancy-induced hypertension obtained by using the above-mentioned method for constructing a diagnostic or predictive model for pregnancy-induced hypertension.

[0057] In the seventh typical embodiment of the present application, an electronic device is provided. The electronic device is used to run a program. When the program runs, it executes the above-mentioned method for screening lipid markers for diagnosing or predicting pregnancy-induced hypertension, the above-mentioned method for constructing a diagnostic or predictive model for pregnancy-induced hypertension, or the diagnostic or predictive model for pregnancy-induced hypertension obtained by using the above-mentioned method for constructing a diagnostic or predictive model for pregnancy-induced hypertension.

[0058] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the present invention.

[0059] From the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus hardware devices such as detection devices. Based on such an understanding, the data processing part of the technical solution of the present application can be embodied in the form of a software product. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of the various embodiments or some parts of the embodiments of the present application.

[0060] This application can be used in numerous general-purpose or special-purpose computing system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on.

[0061] The method provided by this application can be executed in a terminal, a computer terminal, or a similar computing device. Taking the operation on a terminal as an example, Figure 4 It is a hardware structure block diagram of a method for constructing a diagnostic or predictive model for pregnancy-induced hypertension or a terminal for diagnosing or predicting pregnancy-induced hypertension according to an embodiment of the present invention. As Figure 4 shown, the terminal may include one or more ( Figure 4 only one is shown in the figure) processor A1 (processor A1 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory B1 for storing data. Optionally, the above terminal may further include a transmission device C1 for communication functions and an input / output device D1. Those of ordinary skill in the art can understand that Figure 4 the structure shown is only schematic and does not limit the structure of the above terminal. For example, the terminal may further include more or fewer components than Figure 4 shown in the figure, or have a different configuration from Figure 4 shown in the figure.

[0062] The memory B1 can be used to store computer programs. For example, software programs and modules of application software, such as computer programs corresponding to methods such as read segment splicing, clustering, and consensus processing in the embodiments of the present invention. The processor A1 executes various functional applications and data processing by running the computer programs stored in the memory B1, that is, implements the above methods. The memory B1 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory B1 may further include a memory remotely set relative to the processor A1, and these remote memories can be connected to the terminal through a network. Examples of the above network include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0063] The transmission device C1 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by a communication provider of a terminal. In one example, the transmission device C1 includes a network adapter (Network Interface Controller, abbreviated as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device C1 may be a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0064] Obviously, those skilled in the art should understand that some modules or steps of the present application above can be implemented on a general computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Optionally, they can be implemented with program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device, or they can be separately made into individual integrated circuit modules, or multiple modules or steps among them can be made into a single integrated circuit module for implementation. In this way, the present application is not limited to any specific combination of hardware and software.

[0065] The beneficial effects of the present application will be further explained in detail below in conjunction with specific embodiments.

[0066] Embodiment 1

[0067] In the present application, the following method is used to screen significant difference lipid molecules and establish a model. The technical solution is as Figure 1 shown.

[0068] 1. Sample collection: Collect maternal peripheral blood from GDM pregnant women and normal pregnant women, obtain it around 16 weeks of gestation, and immediately store it at 4°C, and perform plasma separation within 8 hours. Immediately store it at -80°C after plasma separation and wait for the next step of processing.

[0069] 2. Lipid mass spectrometry analysis: Perform lipid mass spectrometry analysis on the collected plasma samples, and measure and identify the composition of lipid molecules through mass spectrometry technology.

[0070] 3. Preprocessing of sample data and collection of clinical data: Preprocess the mass spectrometry data, such as denoising, feature selection, etc. At the same time, collect relevant clinical data, such as the age, weight of pregnant women, etc., as additional information for model input.

[0071] 4. Logistic regression parameter screening model: Using the logistic regression model, taking the lipid molecular mass spectrometry analysis data as the independent variable and the GDM status as the dependent variable, and using parameter screening methods such as the stepwise method to select the most significant differential characteristic lipid molecules in predicting GDM. The specific method is as follows. First, we establish an empty regression model according to the following formula 1 to model the fetal genotype. Next, we sequentially add the 1442 lipid parameters detected by mass spectrometry into this model and perform a significance analysis of the final prediction results with the previous model. We select the parameter with the largest significant difference to construct the model of the first parameter as shown in formula 2.

[0072]

[0073] where P(e y ) is the logistic function, representing the probability of an event;

[0074] β0 is the intercept term of the logistic regression model.

[0075]

[0076] X1 is an independent variable, representing a characteristic of the lipid molecular mass spectrometry analysis data,

[0077] β1 is the regression coefficient corresponding to X1.

[0078] Following the above steps, we sequentially add the remaining unselected parameters until no parameter can cause a significant change in the parameter model. The finally selected parameters are X1, X2,…,Xn shown in formula 3.

[0079]

[0080] X2,...,X n are other selected characteristics,

[0081] β2,...,β n are the regression coefficients corresponding to these characteristics.

[0082] 5. Neural network modeling: As Figure 2 shown, we use the selected differential characteristic lipid molecules and clinical parameters as input variables to establish a highly accurate GDM prediction model. Specifically as follows: First, we use a neural network autoencoder to perform dimensionality reduction encoding on a total of 31 parameters, including 27 lipid molecules (shown in Table 3) + 4 clinical parameters (the age, BMI, smoking status, and drinking status of the sample population). The dimensionality reduction process can be expressed as:

[0083] L = g n (…(g2(g1(LIP·W1 + b1))·W2 + b2)+…)·Wn +b n ) Formula 4

[0084] LIP represents a data matrix of 31 parameters (27 lipid molecules + 4 clinical parameters).

[0085] W1, W2,..., W n and b1, b2,..., b n are the weight matrix and bias terms of the neural network.

[0086] g1, g2,..., g n are activation functions, usually used to increase the non - linearity of the model.

[0087] The above LIP is our 31 parameters. We use W1, W2,..., down to W n to achieve the effect of dimensionality reduction. Note that during training, we will build a reverse dimensionality - increasing network to let the input data restore itself to minimize the data loss after dimensionality reduction.

[0088] The final result will output a probability of GDM as shown in Formula 5. Wc will be an (n×1) - dimensional output matrix to output the n dimensionality - reduced features encoded by the previous layer to 1 probability neuron.

[0089] Result = [L]·W c +b c Formula 5

[0090] W c and b c are the weight matrix and bias terms of the last layer of the neural network.

[0091] [L] is the low - dimensional feature representation encoded by the neural network for dimensionality reduction.

[0092] Result is the probability of GDM (Gestational Diabetes Mellitus) output by the neural network.

[0093] In the construction of the above - mentioned neural network model, an auto - encoder is used for model construction to learn the compact representation of data. It includes an encoder and a decoder. The encoder maps the high - dimensional input data to a lower - dimensional representation, and then the decoder maps this representation back to the original input space. During training, the auto - encoder tries to minimize the reconstruction error to retain the key features of the input data. Finally, the encoding part of the auto - encoder can be used as the dimensionality - reduced representation for the next - step classification analysis.

[0094] The encoder is the first part of the autoencoder and accepts the original high-dimensional input data. The encoder maps the input data to a lower-dimensional representation. This lower-dimensional representation is usually considered a compact representation of the input data, capturing the most important features in the input data. The goal of the encoder is to learn how to best compress the input data and finally obtain Equation 5 in the above steps, realizing the calculation of the GDM probability using the neural grid.

[0095] The decoder is the second part of the autoencoder. It accepts the lower-dimensional representation from the encoder and attempts to map it back to the original high-dimensional input space. The task of the decoder is to reconstruct the original data from the low-dimensional representation. The goal of this step is to minimize the data loss after dimensionality reduction so that the reconstructed data is as similar as possible to the original data. That is, the above reverse dimensionality increase network is realized through the decoder, thereby reducing the data loss after dimensionality reduction.

[0096] 6. Result verification: The established prediction model is verified for its results. To evaluate the performance and prediction ability of the model, 10-fold cross-validation is carried out. Each time, 80% of the data is used as the training set and 20% of the data is used as the test set. The performance and prediction ability of the model are evaluated by verifying the prediction accuracy and reliability of the validation samples.

[0097] When using the above constructed neural network model for result verification and calculation, 10-fold cross-validation is used, and different combinations of sample data are adopted to realize the training and testing of the model. During the training process, the parameters and disease conditions of the samples are input into the model for training at the same time; while during the testing process, only the parameters of the samples are input, and the performance of the prediction model is verified by comparing the results output by the model with the actual disease conditions of the samples. Using the above method, the performance and prediction ability of the constructed model can be accurately reflected, and the model obtained through this verification can also realize the calculation and prediction of other external sample data.

[0098] When calculating the probability using multiple parameters in the sample and the above model after dimensionality reduction, multiple parameters (i.e., the above 27 lipid molecules + 4 clinical parameters) need to be combined into a multi-dimensional vector, and then the encoding part of the autoencoder is used to map it to the dimensionality-reduced representation space. This will obtain a low-dimensional feature vector consistent with the autoencoder, and thus the low-dimensional feature vector is used for calculation and prediction in the neural network model.

[0099] Through this method, plasma mass spectrometry, clinical data, and machine learning models are combined to establish a model capable of predicting the risk of abnormal glucose metabolism during pregnancy. Through steps such as preprocessing and parameter screening, the most relevant features and parameters can be extracted, and then a prediction model with high accuracy can be established. Finally, the effectiveness and practicality of the model are verified through result validation. This comprehensive method is expected to provide strong support for the early identification and intervention of abnormal glucose metabolism during pregnancy.

[0100] Example 2

[0101] A total of 85 GDM samples and 89 healthy controls were recruited in the hospital for analysis as shown in Table 1. All samples were divided into a training set and a validation set at a ratio of 80%:20% and subjected to 10-fold cross-validation analysis according to the description in step 6 of Example 1.

[0102] First, the results obtained by fitting the model based on all lipid parameters of mass spectrometry are shown in Table 2, and the prediction accuracy is only 61.76% with AUC = 0.75.

[0103] Therefore, we used the 27 characteristic lipids obtained by the screening method (step 4 of Example 1) from the samples in the first trimester of pregnancy to build a model to predict the incidence of GDM, and we obtained a model with an accuracy of 77.08% and AUC = 0.84.

[0104] After we added four clinical characteristics, age, BMI, whether smoking, and whether drinking alcohol, the AUC could reach 0.91 and the accuracy was increased to 88.24%.

[0105] We also separately built a model for clinical data, as Figure 3 shown, using only the four clinical parameters could achieve an effect with AUC 0.70 and an accuracy of 66.63%; while combining clinical data and 27 lipids could reach AUC 0.91 and an accuracy of 88.24%.

[0106] Figure 3 In , TPR (True Positive Rate), also known as Recall, refers to the ratio of the number of samples correctly identified as positive by the classifier to the total number of positive samples among the positive samples. It measures the ability of the model to identify positive samples, that is, among all actual positive cases, how many are correctly classified as positive. TPR = TP / (TP + FN), where TP represents the true positive cases (the number of samples correctly classified as positive), and FN represents the false negative cases (the number of samples that are actually positive but are misclassified as negative).

[0107] In Figure 3The ROC curve (Receiver Operating Characteristic curve) plotted in the (AUC graph) is a curve that describes the variation of model performance at different classification thresholds. AUC represents the area under the ROC curve and is used to quantify model performance. The value of AUC ranges from 0 to 1, and the closer the AUC value is to 1, the better the model performance.

[0108] Therefore, it can be seen that both using only the lipidome (Lipo) and using only clinical information (Clinic) can predict GDM well, but integrating these two pieces of information can further improve the prediction effect of the model.

[0109] Table 1. Mean and 95% confidence interval of sample clinical information

[0110] GDM samples (n = 85) Healthy controls (n = 89) Maternal age 30.65(26.75-34.55) 30.61(27.98-33.24) Gestational age at blood sampling 12.04(11.35-12.73) 12.03(11.22-12.84) Pre-pregnancy BMI 25.41(22.14-28.68) 22.32(20.12-24.52)

[0111] Table 2: Prediction of early pregnancy GDM based on lipidome data

[0112]

[0113]

[0114] Table 3: 27 selected significant lipid molecules

[0115]

[0116] Example 3

[0117] Furthermore, based on the original 27, another 5 key lipid molecules were selected through the parameter importance ranking of the model to construct a prediction model for gestational diabetes mellitus (GDM). The information of the 5 lipid molecules is shown in Table 4. These lipid molecules include triglycerides, phosphatidylethanolamines, sphingomyelins, and phosphatidylinositol lipids, which are biologically significantly representative.

[0118] Table 4

[0119]

[0120]

[0121] We found that the new model using these five lipid molecules still performs excellently in predicting GDM. Specifically, the area under the curve (AUC) of the new model reached 0.83, indicating its high diagnostic performance. The accuracy rate was 0.85, and this result shows that the model has high reliability in identifying GDM patients and non-GDM patients.

[0122] It should be noted that although the performance of the new model is slightly lower than that of the original model (the AUC of the original model is 0.91 and the accuracy is 0.88), this gap is acceptable considering the significant reduction in the number of lipid molecules used. In addition, the optimization of the new model has reduced both the experimental cost and the time cost while maintaining good prediction results.

[0123] In summary, our research shows that by carefully selecting lipid molecules and combining clinical parameters, early prediction of GDM can be effectively carried out from the perspective of lipidomics. This finding is of great significance for improving the clinical management of pregnant women and reducing the risk of pregnancy complications.

[0124] From the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects: In this application, the above lipid markers are used, and clinical parameters are further used to jointly diagnose or predict the GDM situation of the subject, and a better judgment of GDM can be made from the perspective of lipidomics. The above method can predict GDM at 11 - 13 weeks of early pregnancy. The predicted AUC is 0.91 and the accuracy is 0.88. This method has a high prediction ability and is superior to the existing technical level. In addition, by screening out 27 lipids and combining clinical information, not only the cost is reduced, but also the prediction effect is ensured. The ability to accurately predict GDM in early pregnancy can help medical professionals and pregnant women take early intervention measures to reduce the risk of pregnancy complications. This technology not only performs well in terms of accuracy, but also pays attention to cost - effectiveness, so it has the potential for practical application.

[0125] The references involved in this application are as follows:

[0126] 1. Ben - Haroush, A., Yogev, Y., & Hod, M. (2004). Epidemiology of gestational diabetes mellitus and its association with Type 2 diabetes. Diabetic Medicine, 21(2), 103 - 113.

[0127] 2. Savvidou, M., Nelson, S. M., Makgoba, M., Messow, C. M., Sattar, N., & Nicolaides, K. (2010). First-trimester prediction of gestational diabetes mellitus: examining the potential of combining maternal characteristics and laboratory measures. Diabetes, 59(12), 3017-3022.

[0128] 3. Lamain-de Ruiter, M., Kwee, A., Naaktgeboren, C. A., Franx, A., Moons, K. G., & Koster, M. P. (2017). Prediction models for the risk of gestational diabetes: a systematic review. Diagnostic and Prognostic Research, 1(1), 1-9.

[0129] 4. Wu, Y. T., Zhang, C. J., Mol, B. W., Kawai, A., Li, C., Chen, L.,... & Huang, H. F. (2021). Early prediction of gestational diabetes mellitus in the Chinese population via advanced machine learning. The Journal of Clinical Endocrinology & Metabolism, 106(3), e1191-e1205.

[0130] 5. Hou, G., Gao, Y., Poon, L. C., Ren, Y., Zeng, C., Wen, B.,... & Nicolaides, K. H. (2023). Maternal plasma diacylglycerols and triacylglycerols in the prediction of gestational diabetes mellitus. BJOG: An International Journal of Obstetrics & Gynaecology, 130(3), 247 - 256.

[0131] 6. Wang, J., Zheng, J., Shi, W., Du, N., Xu, X., Zhang, Y.,... & Zhao, F. (2018). Dysbiosis of maternal and neonatal microbiota associated with gestational diabetes mellitus. Gut, 67(9), 1614 - 1625.

[0132] 7. Rasanen, J. P., Snyder, C. K., Rao, P. V., Mihalache, R., Heinonen, S., Gravett, M. G.,... & Nagalla, S. R. (2013). Glycosylated fibronectin as a first - trimester biomarker for prediction of gestational diabetes. Obstetrics & Gynecology, 122(3), 586 - 594.

[0133] 8. Koos, B. J., & Gornbein, J. A. (2021). Early pregnancy metabolites predict gestational diabetes mellitus: implications for fetal programming. American journal of obstetrics and gynecology, 224(2), 215 - e1.

[0134] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. Use of a lipid biomarker in the preparation of a product for diagnosing or predicting gestational diabetes, characterized in that, The lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or The corresponding lipid metabolism proteins of the lipid molecules.

2. The application according to claim 1, wherein The lipid markers include a combination of the following lipid molecules: A)C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B)C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1 3. The application according to claim 1 or 2, characterized in that, The product is used to detect the content of the lipid markers in a biological sample of an individual; Preferably, the detection method for the content of the lipid markers includes mass spectrometry detection, nuclear magnetic resonance detection or probe detection; Preferably, the content includes the relative abundance of mass spectrometry; Preferably, the product includes a standard, and the standard is a composition formed by any one or more of the lipid markers, or a set of pure standard products formed by separately setting any one or more of the lipid markers; Preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

4. A kit for diagnosing or predicting pregnancy-induced hypertension in a subject, characterized in that, The kit includes a detection reagent for detecting the content of lipid markers, and the lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or The corresponding lipid metabolism proteins of the lipid molecules.

5. The kit according to claim 4, wherein The lipid markers include a combination of the following lipid molecules: A)C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B)C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1; Preferably, the detection reagent includes a standard, and the standard is a composition formed by any one or more of the lipid molecules, or a set of pure standard products formed by separately setting any one or more of the lipid molecules; Preferably, the detection reagent includes a probe that can specifically bind to the lipid marker, and the probe contains a group for detection; Preferably, the group for detection includes a fluorescent group, an isotope group, an enzyme group or a biotin group.

6. A screening method for lipid markers for diagnosing or predicting pregnancy-induced hypertension, characterized in that, The screening method includes: a) Obtain the clinical status and biological samples of the sample population, detect the biological samples, and obtain various lipid molecules and the content of the lipid molecules in the biological samples. The sample population includes healthy pregnant women and pregnant women with pregnancy-induced hypertension. Correspondingly, the clinical status includes health and pregnancy-induced hypertension; b) According to the lipid molecules and the content of the lipid molecules, screen for significantly different lipid molecules that are different between the pregnant women with pregnancy-induced hypertension and the healthy pregnant women by statistical methods; Preferably, the statistical methods include t-test, generalized linear model, rank test, logistic regression, fold change, multiple hypothesis testing correction or Kruskal-Wallis test; Preferably, the step b) includes: Using a logistic regression model, with the lipid molecules and the content of the lipid molecules as independent variables and the clinical status as the dependent variable, screen to obtain the significantly different lipid molecules.

7. A method for constructing a diagnostic or predictive model for pregnancy-induced hypertension, characterized in that, The construction method uses the lipid markers obtained by the screening method of the lipid markers for diagnosing or predicting pregnancy-induced hypertension according to claim 6, and applies machine learning methods to construct the diagnosis or prediction model for pregnancy-induced hypertension; Preferably, the construction method uses the lipid markers and clinical parameters to construct the diagnosis or prediction model for pregnancy-induced hypertension; Preferably, the clinical parameters include one or more of the age, BMI, smoking status or drinking status of the sample population; Preferably, the machine learning includes a logistic regression screening method and / or a neural network classifier model.

8. An electronic device for diagnosing or predicting pregnancy-induced hypertension, characterized in that, The electronic device includes: An acquisition and detection module, which is configured to acquire a biological sample, detect the biological sample, and obtain the content of lipid markers in the biological sample; A diagnosis or prediction module, which has a pregnancy-induced hypertension diagnosis or prediction model built in. The diagnosis or prediction module is configured to input the content of the lipid markers into the pregnancy-induced hypertension diagnosis or prediction model and output a prediction result according to the pregnancy-induced hypertension diagnosis or prediction model; Preferably, the lipid markers include any one or more of the following lipid molecules: C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 or C 45 H 92 O8N2P1; or The corresponding lipid metabolism proteins of the lipid molecules; Preferably, the lipid markers include a combination of the following lipid molecules: A)C 59 H 102 O6N1, C 55 H 104 O6N1, C 65 H 112 O6N1, C 59 H 110 O6N1, C 36 H 63 O4, C 46 H 84 O8N1P1Na1, C 52 H 85 O8N1P1, C 50 H 85 O8N1P1, C 45 H 87 O 13 N1P1, C 41 H 84 O6N2P1, C 44 H 85 O8N1P1, C 50 H 86 O8N1P1Na1, C 41 H 80 O7N1P1Na1, C 46 H 87 O7N1P1, C 43 H 74 O5N1, C 49 H 94 O6N2P1, C 52 H 95 O7N1P1, C 54 H 102 O 13 N1, C 47 H 80 O 12 P1, C 49 H 83 O9N1P1, C 49 H 82 O 13 P1, C 49 H 85 O 10 N1P1, C 45 H 77 O7N1P1, C 43 H 77 O7N1P1, C 48 H 87 O 10 N1P1, C 43 H 79 O7N1P1 and C 45 H 92 O8N2P1; or B)C 59 H 102 O6N1, C 46 H 84 O8N1P1Na1, C 41 H 84 O6N2P1, C 47 H 80 O 12 P1 and C 45 H 92 O8N2P1; Preferably, the pregnancy-induced hypertension diagnosis or prediction model is the pregnancy-induced hypertension diagnosis or prediction model obtained by using the construction method described in claim 7; Preferably, the biological sample includes plasma, amniotic fluid, serum or urine.

9. A computer-readable storage medium, characterized in that, The storage medium includes a stored program. When the program runs, the device where the storage medium is located is controlled to execute the screening method of lipid markers for diagnosing or predicting pregnancy-induced hypertension described in claim 6, the construction method of the pregnancy-induced hypertension diagnosis or prediction model described in claim 7, or the pregnancy-induced hypertension diagnosis or prediction model obtained by using the construction method of the pregnancy-induced hypertension diagnosis or prediction model described in claim 7.

10. An electronic device, characterized in that, The electronic device is used to run a program. When the program runs, it executes the screening method of lipid markers for diagnosing or predicting pregnancy-induced hypertension described in claim 6, the construction method of the pregnancy-induced hypertension diagnosis or prediction model described in claim 7, or the pregnancy-induced hypertension diagnosis or prediction model obtained by using the construction method of the pregnancy-induced hypertension diagnosis or prediction model described in claim 7.

Images