CN117747103A - Preeclampsia risk prediction model based on key enzyme fusion index EHI - Google Patents

Abstract

The invention discloses a preeclampsia risk prediction model based on a key enzyme fusion index EHI, and belongs to the technical field of preeclampsia risk assessment. The preeclampsia risk prediction model based on the key enzyme fusion index EHI provided by the invention uses two methods of statistics and random forest, creatively simplifies the preeclampsia screening and early warning mode, has high result prediction accuracy and strong interpretability, has stable effect and is convenient for medical workers to use; the method can effectively improve the accuracy and the popularity of preeclampsia screening, effectively reduce/avoid the phenomena of missed diagnosis and undefined diagnosis, and has higher practical value for preeclampsia screening.

Description

A risk prediction model for preeclampsia based on key enzyme fusion index EHI

Technical field

The present invention relates to the technical field of preeclampsia risk assessment, and in particular to a preeclampsia risk prediction model based on the key enzyme fusion index EHI.

Background technique

As the name suggests, preeclampsia refers to signs of eclampsia, also known as preeclampsia in medicine. It refers to the occurrence of systolic blood pressure ≥140mmHg and/or diastolic blood pressure ≥90mmHg after 20 weeks of pregnancy, accompanied by urine protein ≥0.3g/24h. , or random urine protein (+), or although there is no proteinuria, preeclampsia is diagnosed when any one of the following symptoms is combined: 1) thrombocytopenia; 2) liver function damage; 3) renal function damage; 4) Pulmonary edema; 5) New central nervous system abnormalities or visual impairment and other symptoms. After the occurrence of preeclampsia, if effective intervention and treatment measures cannot be taken in time, it will further develop into severe preeclampsia, with coma and convulsions, and may be complicated by renal failure, heart failure, pulmonary edema, intracranial hemorrhage, placental abruption and other eclampsia. Symptoms threaten the lives of pregnant women and fetuses.

The incidence of preeclampsia (PE) continues to rise. According to the New England Journal of Medicine, 2%-4% of pregnancies worldwide are complicated by preeclampsia, which is the most common maternal death besides postpartum hemorrhage. reason. In addition, preeclampsia will also have an impact on the short-term and long-term health of fetuses and newborns. It is one of the current scientific problems in the field of obstetrics that needs to be solved.

Preeclampsia is a pregnancy disease that accumulates multiple organ dysfunction and has complex clinical manifestations. Once preeclampsia occurs, there is no effective treatment unless the pregnancy is terminated. Therefore, prevention and treatment to prevent the occurrence of preeclampsia or to postpone it as much as possible are extremely important to reduce the harm of preeclampsia to mothers and babies. Currently, the prediction of preeclampsia is based on urine protein prediction methods. The prediction effect is poor, there are missed diagnoses, and the interpretability is not comprehensive. For patients with negative urine protein but other complications of preeclampsia, the diagnosis is not clear. Facilitates identification and treatment of preeclampsia. These disadvantages greatly limit the application of clinical testing to determine the probability of preeclampsia in medical screening.

Contents of the invention

The purpose of the present invention is to provide a preeclampsia risk prediction model based on the key enzyme fusion index EHI to solve the problem that the current preeclampsia prediction method has poor prediction effect, missed diagnosis, incomplete interpretability, and problems with negative urine protein but with Patients with other complications of preeclampsia have problems such as unclear diagnosis, which seriously affects the identification and treatment of preeclampsia.

To achieve the above objectives, the present invention provides a preeclampsia risk prediction model based on the key enzyme fusion index EHI. The prediction model is a machine learning prediction model based on the EHI score of the person to be screened and the preeclampsia EHI risk classification threshold. Determine the risk of preeclampsia in the person to be screened, and output the predicted probability of preeclampsia in the person to be screened based on the random forest model.

Preferably, the EHI score calculation formula of the person to be screened is: Among them, LDH is lactate dehydrogenase, ALP is alkaline phosphatase, AMY is amylase, and U-PRO is urine protein.

Preferably, the EHI risk classification threshold for preeclampsia is: EHI less than 2, defined as "lower risk", EHI greater than or equal to 2 and less than 3, defined as "low risk", EHI greater than or equal to 3 and less than 10, defined as "low risk" "medium risk" and EHI greater than 10 are classified as "high risk".

A method for establishing a preeclampsia risk prediction model based on the key enzyme fusion index EHI as described above. The specific steps are as follows:

S1. Obtain the blood biochemistry and urine routine test results of population samples from the hospital;

S2. Use the LDH, ALP, AMY, and U-PRO data in the test results obtained in step S1 to calculate EHI and establish the EHI risk classification threshold for preeclampsia;

S3. Input the LDH, ALP, AMY, U-PRO and EHI data obtained in steps S1 and S2 into the random forest model for training, obtain the machine learning prediction model and save it;

S4. Enter the sample to be tested into the EHI scoring calculation formula and the machine learning prediction model obtained in S3, and determine the risk and probability of preeclampsia based on the preeclampsia EHI risk classification threshold and the machine learning prediction results.

Preferably, the population samples in step S1 include healthy pregnancy population samples and patient samples diagnosed with preeclampsia.

Preferably, the method for establishing the EHI risk classification threshold for preeclampsia in step S2 is:

S21. Calculate the EHI score of each sample according to the EHI score calculation formula;

S22. Divide the EHI score of each sample into intervals with an interval of 1, and count the EHI score distribution of the sample; divide the number of samples of each type in each interval by the total number of samples of the two types in the interval to obtain the sample of this type. The probability of being within the EHI interval;

S23. Combine multiple intervals with the same trend according to the probability obtained in step S22 to obtain the preeclampsia EHI risk classification threshold.

Preferably, the principle of merging multiple intervals in step S23 is to balance the total number of samples falling in each merged interval.

Preferably, the random forest model establishment method in step S3 is as follows:

S31. Set the label of preeclampsia patients as 1, set the label of healthy pregnant people as 0, and use the LDH, ALP, AMY, U-PRO indicators and EHI scores of the sample as features;

S32. Divide the samples into training sets and test sets equally, with a ratio of 4:1;

S33. Input the training set into the random forest model for training to obtain the random forest model.

The application of a preeclampsia risk prediction model based on the key enzyme fusion index EHI as described above in the assessment of the risk and probability of preeclampsia.

A method for using the preeclampsia risk prediction model based on the key enzyme fusion index EHI as described above, the method is:

S71. After inputting the sample data to be tested into the established preeclampsia risk prediction model based on the key enzyme fusion index EHI, calculate the EHI score according to the EHI score calculation formula;

S72. Output the corresponding preeclampsia risk according to the preeclampsia EHI risk classification threshold;

S73. Load the enzyme/protein index and EHI scoring results in step S71 into the trained prediction model to obtain the predicted probability of suffering from preeclampsia.

The preeclampsia risk prediction model provided by the present invention is a preeclampsia risk prediction model based on the key enzyme fusion index EHI of liquid biopsy technology, using four key enzyme/protein indicators of LDH, ALP, AMY, and U-PRO through mathematical formula calculations. Afterwards, the EHI (Enzyme Hybrid Index) value was obtained, and the probability of suffering from preeclampsia was predicted by inputting the four key enzyme/protein indicators of LDH, ALP, AMY, and U-PRO into the random forest model.

Therefore, the preeclampsia risk prediction model based on the key enzyme fusion index EHI provided by the present invention has the following beneficial effects compared with the existing technology:

(1) The preeclampsia risk prediction model provided by the present invention uses two methods, statistics and random forest, to creatively simplify the preeclampsia screening and early warning model. The result prediction accuracy is high, the interpretability is strong, and the effect is stable. , easy for medical workers to use;

(2) The application of the preeclampsia risk prediction model provided by the present invention can effectively improve the accuracy and popularity of preeclampsia screening, effectively reduce/avoid the phenomenon of missed diagnosis and unclear diagnosis, and has a relatively high effect on the screening of preeclampsia. High practical value.

The technical solution of the present invention will be further described in detail below through the accompanying drawings and examples.

Description of drawings

In order to explain the technical solutions of the embodiments of the present invention more clearly, the drawings needed to be used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. , for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative labor.

Figure 1 is a flow chart for establishing a preeclampsia risk prediction model based on the key enzyme fusion index EHI in Embodiment 1 of the present invention;

Figure 2 is a statistical graph of EHI risk classification thresholds for preeclampsia;

Figure 3 is a flow chart of test sample detection in an example of the effect of the present invention.

Detailed ways

The technical solution of the present invention will be further described below through the drawings and examples.

In order to make the purpose, technical solutions and advantages of the present application more clear, thorough and complete, the technical solutions of the present invention are clearly and completely described below through the drawings and examples. The following detailed descriptions are descriptions of embodiments and are intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meanings as commonly understood by those of ordinary skill in the art to which this application belongs.

The population samples used in the examples were obtained from the Inner Mongolia Autonomous Region People's Hospital. The research in the examples of the present invention was approved by the Ethics Committee of the Inner Mongolia Autonomous Region People's Hospital in 2022 (Approval Number: 202201004L).

Embodiment 1

To establish a preeclampsia risk prediction model based on the key enzyme fusion index EHI, the steps are as follows:

S1: Obtain the blood biochemistry and urine routine test results of 600 samples from healthy pregnant women and 600 samples from patients diagnosed with preeclampsia from the hospital.

S2: Use the lactate dehydrogenase, alkaline phosphatase, amylase, and urine protein data in the test results obtained in step S1 to establish the preeclampsia EHI risk classification threshold using the following method:

S21: Calculate the EHI score of each sample. The calculation formula is:

Among them, LDH is lactate dehydrogenase, the unit is U/L; ALP is alkaline phosphatase, the unit is U/L; AMY is amylase, the unit is U/L; U-PRO is urine protein, the unit is mg/L ;

S22: Divide the EHI score of each sample into intervals of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, +∞ with a spacing of 1, and count the EHI score distribution of the sample;

S23: Merge multiple intervals based on statistical results. The merged statistical results are shown in Figure 2, and the EHI risk classification threshold for preeclampsia is obtained (see Table 1): if EHI is less than 2, it is classified as "lower risk" , if the EHI is greater than or equal to 2 and less than 3, it is classified as "low risk", if the EHI is greater than or equal to 3 and less than 10, it is classified as "medium risk", and if the EHI is greater than 10, it is classified as "high risk".

Table 1 Preeclampsia EHI risk classification thresholds

EHI scoring area and threshold Preeclampsia risk assessment results [0,2) lower risk [2,3) low risk [3,10) medium risk [10,+∞) high risk

S3: Use the test result data obtained in step S1 to establish a random forest model to predict the probability of preeclampsia. The method is as follows:

S31. Set the label of preeclampsia patients as "1", set the label of healthy pregnant people as "0", and score the lactate dehydrogenase, alkaline phosphatase, amylase, urine protein indicators and EHI of the sample. as a characteristic;

S32. Randomly select 480 samples of healthy pregnant people and 480 samples of patients diagnosed with preeclampsia as the training set, and the remaining samples are the test set;

S33. Input the training set into the random forest model for training to obtain the random forest model.

S4. Input the LDH, ALP, AMY, U-PRO and EHI data obtained in steps S1 and S2 into the random forest model to obtain a machine learning prediction model. The obtained machine learning prediction model is the precursor based on the key enzyme fusion index EHI. Epilepsy risk prediction model, save the model. The establishment flow chart is shown in Figure 1.

Effect example

Test the prediction accuracy of the preeclampsia risk prediction model based on the key enzyme fusion index EHI obtained in Example 1. The method is as follows:

Randomly select 3 samples from the population samples collected in the hospital as samples to be tested, and record them as test No. 1, test No. 2, and test No. 3 respectively. LDH, ALP, The test results of AMY and U-PRO are shown in Table 2.

Table 2 LDH, ALP, AMY, U-PRO and EHI results of samples to be tested

No. 1 to be tested No. 2 to be tested No. 3 to be tested LDH 268.00 181.00 278.00 ALP 179.55 136.00 141.90 AMY 50.00 61.00 44.00 U-PRO 1 0 2 EHI 5.97 2.18 17.81

Use formula Calculate the EHI score of each sample to be tested, and the results are shown in Table 2.

Comparing with the preeclampsia EHI risk classification threshold (Table 1) determined in Example 1, the result is that the risk of preeclampsia for sample No. 1 to be tested is medium risk, and the risk of preeclampsia for sample No. 2 to be tested is low. Risk, sample No. 3 to be tested has a high risk of preeclampsia.

Load the LDH, ALP, AMY, U-PRO test results and EHI scoring results of the three samples to be tested in Table 2 into the preeclampsia risk prediction model based on the key enzyme fusion index EHI established in Example 1 to obtain the test results The predicted probability of the sample suffering from preeclampsia is 66.40% for sample No. 1 to be tested, 45.60% for sample No. 2 to be tested, and 45.60% for sample No. 3 to be tested. The probability of getting sick is 100%. The detection flow chart is shown in Figure 3.

After searching the source of the samples to be tested, it was found that the sample No. 1 to be tested was a healthy pregnancy sample, the sample No. 2 to be tested was a healthy pregnancy sample, and the sample No. 3 to be tested was a patient diagnosed with preeclampsia, and the result prediction was accurate.

Therefore, the present invention provides a preeclampsia risk prediction model based on the key enzyme fusion index EHI, which uses two methods, statistics and random forest, to creatively simplify the preeclampsia screening and early warning model, and the result prediction accuracy is High, interpretable, stable effect, easy to use by medical workers; it can effectively improve the accuracy and popularization of preeclampsia screening, effectively reduce/avoid missed diagnosis and unclear diagnosis, and is suitable for preeclampsia screening It has high practical value.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: The technical solution of the present invention may be modified or equivalently substituted, but these modifications or equivalent substitutions cannot cause the modified technical solution to depart from the spirit and scope of the technical solution of the present invention.

Claims (10)

1. The preeclampsia risk prediction model based on the key enzyme fusion index EHI is characterized in that: the prediction model is a machine learning prediction model, the preeclampsia sickness risk of the person to be screened is determined based on the EHI scoring of the person to be screened and the preeclampsia EHI risk classification threshold, and the preeclampsia sickness prediction probability of the person to be screened is output based on the random forest model.

2. The pre-eclampsia risk prediction model based on a key enzyme fusion index EHI according to claim 1, wherein the EHI scoring calculation formula of the person to be screened is:wherein LDH is lactate dehydrogenase, ALP is alkaline phosphatase, AMY is amylase, and U-PRO is urine protein.

3. The pre-eclampsia risk prediction model based on a key enzyme fusion index EHI according to claim 1, wherein the pre-eclampsia EHI risk classification threshold is: EHI is less than 2, and is defined as "lower risk", EHI is greater than or equal to 2 and less than 3, and is defined as "low risk", EHI is greater than or equal to 3 and less than 10, and is defined as "medium risk", EHI is greater than 10, and is defined as "high risk".

4. A method for building a preeclampsia risk prediction model based on a key enzyme fusion indicator EHI as set forth in any one of claims 1 to 3, comprising the following specific steps:

s1, obtaining blood biochemistry and urine routine detection results of crowd samples from a hospital;

s2, calculating EHI by using LDH, ALP, AMY, U-PRO data in the detection result obtained in the step S1 and establishing a preeclampsia EHI risk classification threshold;

s3, inputting LDH, ALP, AMY, U-PRO and EHI data obtained in the steps S1 and S2 into a random forest model to obtain a machine learning prediction model and storing the model;

s4, inputting the sample to be tested into the EHI scoring calculation formula and the machine learning prediction model obtained in the S3, and judging the risk and probability of the preeclampsia according to the preeclampsia EHI risk classification threshold and the machine learning prediction result.

5. The method for establishing a preeclampsia risk prediction model based on a key enzyme fusion indicator EHI according to claim 4, wherein the method comprises the following steps: the population sample in step S1 includes a healthy gestational population sample and a patient sample diagnosed with preeclampsia.

6. The method for establishing a pre-eclampsia risk prediction model based on a key enzyme fusion index EHI according to claim 4, wherein the method for establishing a pre-eclampsia EHI risk classification threshold in step S2 is as follows:

s21, calculating EHI scoring of each sample according to an EHI scoring calculation formula;

s22, dividing the EHI of each sample into intervals with the interval of 1, and counting EHI scoring distribution conditions of the samples; dividing the number of each type of sample in each interval by the total number of the two types of samples in the interval to obtain the probability of the type of sample in the EHI interval;

s23, combining a plurality of intervals with the same trend according to the probability obtained in the step S22 to obtain the preeclampsia EHI risk classification threshold.

7. The method for establishing a pre-eclampsia risk prediction model based on a key enzyme fusion index EHI according to claim 6, wherein the principle of merging the multiple regions in step S23 is as follows: the total number of samples falling within each merge interval is equalized.

8. The method for establishing a pre-eclampsia risk prediction model based on a key enzyme fusion index EHI according to claim 4, wherein the random forest model establishment method in the step S3 is as follows:

s31, marking a label of a preeclampsia patient as 1, marking a label of a healthy pregnant crowd as 0, and marking LDH, ALP, AMY, U-PRO indexes and EHI of a sample as characteristics;

s32, equally dividing the sample into a training set and a testing set, wherein the ratio is 4:1;

s33, inputting the training set into a random forest model for training to obtain the random forest model.

9. Use of a pre-eclampsia risk prediction model based on a key enzyme fusion index EHI as claimed in any one of claims 1-3 in the assessment of pre-eclampsia risk and probability of onset.

10. A method of using a pre-eclampsia risk prediction model based on a key enzyme fusion index EHI as claimed in any one of claims 1 to 3, wherein the method is:

s71, inputting sample data to be tested into an established preeclampsia risk prediction model based on a key enzyme fusion index EHI, and calculating EHI scoring according to an EHI scoring calculation formula;

s72, outputting corresponding preeclampsia diseased risks according to the preeclampsia EHI risk classification threshold;

s73, loading the enzyme/protein index and the EHI scoring result in the step S71 into a trained prediction model to obtain the prediction probability of preeclampsia.