CfRNA marker for predicting premature birth risks
A risk, RAB27B technology, applied in the field of biomedicine, can solve the problem of non-invasive prenatal diagnosis technology blank
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0039] Example 1 Screening for Gene Markers Related to Preterm Birth
[0040] 1. Sample
[0041] 208 samples with complete clinical information were screened from the samples kept in the hospital, including 156 normal pregnant women and 52 pregnant women with premature delivery. They were randomly divided into a training group (156 cases) and a verification group (52 cases). The clinical information of the included patients Including maternal age, BMI value, pregnancy history, interval between two pregnancies, drinking history, smoking history, drug abuse history, whether there is a history of late miscarriage and / or premature delivery, history of cervical surgery, results of vaginal ultrasonography, whether the fetus and amniotic fluid volume are abnormal , Whether it is multiple pregnancy, whether there are pregnancy complications or complications, whether it is assisted by assisted reproductive technology, mode of production, whether it is full-term birth, whether it is dys...
Embodiment 2
[0049] Embodiment 2 feature selection process
[0050] 1. Use the LASSO algorithm to select important feature variables
[0051] The loss function of LASSO is minimized as follows:
[0052]
[0053] Based on the glmnet package in R language, the loss function is solved. Simply speaking, different w can be obtained by selecting different λ values. By choosing the optimal parameters, the error rate is minimized. Using LASSO to select feature variables, 16 cfRNAs were screened out.
[0054] 2. Use the Boruta algorithm to select important feature variables
[0055] Boruta Algorithm It is a wrapper algorithm around Random Forest. When fitting a random forest model to a dataset, you can recursively address the underperforming features at each iteration. This method minimizes the error of the random forest model, which will eventually form a minimized optimal feature subset.
[0056] The steps of the boruta algorithm are as follows:
[0057] 1) First, it adds randomness to ...
Embodiment 3
[0064] Example 3 Random forest model predicts the risk of premature birth in pregnant women
[0065] Estimation procedure for random forest
[0066] 1. Specify the value of m, that is, randomly generate m variables for the binary tree on the node. The selection of binary tree variables still meets the principle of minimum node impurity. In this model, m=2;
[0067] 2. Use the Bootstrap self-help method to randomly select k sample sets in the original data set with replacement to form k decision trees, and for the unextracted samples to be used for the prediction of a single decision tree, when k=300 in this model, The model is basically stable;
[0068] 3. According to the random forest model RandomForest_model composed of k decision trees, the unknown_sample to be classified is predicted. The function used is predict(RandomForest_model, data=unknown_sample_data), and the principle of prediction is simple average;
[0069] 4. The importance function calculates the importance...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More