A Multicenter Study on Early Diagnosis of NSTE-ACS Patients Based on Machine Learning Model

First Affiliated Hospital of Xinjiang Medical University

Status: Unknown

Conditions: NSTEMI - Non-ST Segment Elevation MI; Unstable Angina

Treatments: Diagnostic Test: The model of machine learning

Study type: Observational

Funder types: Other

Identifiers

NCT04682756

XMa

Details and patient eligibility

About

Early diagnosis of NSTEMI and UA patients is mainly through the construction of machine learning model.

Full description

The patients with NSTEMI and UA were included. After manual labeling, the admiss- ion record characteristics of patients were selected. 75% of the data is used to build the model, and 25% of the data is used to verify the validity of the model. Five classification models of one-dimensional convolution (CNN), naive Bayesian (NB), support vector machine (SVM), random forest (RF) and ensemble learning were constructed to identify and diagnose NSTEMI and UA patients. Multi-fold cross-validation and ROC-AUC curve are used to measure the advantages and disadvantages of the models.

Enrollment: 2,500 patients

Sex: All

Ages: 18 to 75 years old

Volunteers: No Healthy Volunteers

Inclusion criteria

• Patients were included and excluded strictly according to the diagnostic criteria of Chinese guidelines for diagnosis and treatment of Non-STsegment elevation acute coronary syndrome (2016). The patients were admitted to the hospital with chest pain as the main complaint, and were admitted to the first affiliated Hospital of Xinjiang Medical University and the first affiliated Hospital of Medical College of Shihezi Univ- ersity. the patients were diagnosed as NSTEMI and UA by coronary angiography (age range from 30 to 75 years old).

Exclusion criteria

• 1. Patients with STEMI, aortic dissecting aneurysm, pneumothorax and other non-cardiogenic chest pain. 2.Severe hepatorenal failure, primary tumor without surgical treatment, non-severe infection complicated with shock and pregnant women. 3.Previous severe valvular disease, viral myocarditis, pericardial effusion, cardiac pacemaker implantation, cardiogenic shock with serious complications, hypertensive heart disease, various cardiomyopathy, congenital heart disease, etc.

4.Patients with heart disease, AECOPD, lung tumor and hyperthyroidism were diagnosed in the past.

Trial design

2,500 participants in 2 patient groups

CNN model

Description:

Electronic health information of NSTEMI and UA patients in two chest pain centers from 2017 to 2019 was collected，After manual labeling, the characteristics of patient admission records were selected, and through the construction of one-dimensional convolution (CNN) model. Taking the multi-fold cross-validation and ROC-AUC curve as the measurement index, 75% of the data are modeled and 25% of the data are used to verify the effect of the model.

Treatment:

Diagnostic Test: The model of machine learning

XG boost

Description:

Through the construction of XG boost model,taking the multi-fold cross-validation and ROC-AUC curve as the measurement index, 75% of the data are modeled and 25% of the data are used to verify the effect of the model.

Treatment:

Diagnostic Test: The model of machine learning