A method for optimal selection of stroke-related biomarkers

Abstract

The invention relates to a method for optimally selecting stroke-related biomarkers, which is mainly divided into two links. The first link is to diagnose whether a patient is suffering from stroke, and the second link is to distinguish patients with stroke. What kind of stroke is it. In each link, through mutual information analysis, the mutual information value of stroke and disease, and which type of stroke is analyzed separately, and used as a basis for screening. Then use principal component analysis to reduce the redundancy of information, and finally use the method of support vector machine to adjust and correct the screened biomarkers to complete the entire optimization selection process. The invention reduces the redundant waste of information, realizes the correction of the selected results; reduces the types of biochemical detection items, reduces the diagnosis cost, improves the correct rate of diagnosis, and can realize the prediction of stroke risk at the same time; provides a It is a general means of processing biological information, and the same method can be used for optimal selection of biomarker information of other diseases.

Description

technical field [0001] The invention provides a method for optimally selecting stroke-related biomarkers, which is a method for optimally selecting stroke-related biomarkers based on mutual information analysis-principal component analysis-two-level support vector machine, and belongs to the field of biomedical engineering . Background technique [0002] Stroke has a very serious impact on people's health and life. With its high morbidity, high mortality and high disability rate, it poses a huge threat to human health and has become the second most common cause of death in the world. It is difficult to detect stroke. The current routine detection items for patients with cerebrovascular disease include: [0003] a) Mainly blood pressure, electrocardiogram, fundus examination, blood routine, urine routine, blood lipids, blood sugar, hemorheology, blood [0004] After routine testing of homocysteine ​​and antiphospholipid antibodies; [0005] b) Patients with cerebrovascular...

AI Technical Summary

Problems solved by technology

[0010] Traditional macromolecular markers (macromolecular proteins, antigens, hormones, etc.) are detected by immunohistochemistry or Western blot based on antibodies, which have low sensitivity, complicated operation, and difficult antibody preparation; gene markers mainly use reverse transcription-polymerase Chain reaction detection, although the sensitivity is high, but it is easy to have cross-contamination during operation, resulting in false positives

For stroke, there is no single, effective biomarker. Concentrated biomarkers must be combined to understand and understand stroke more comprehensively and play its clinical role

At present, when doing biochemical testing, in order to fully understand the disease, many markers are often tested. However, the detection of some biomarkers is unnecessary, and at the same time, there will be information overlap between some biomarkers, resulting in information overload. Redundancy

Images