a Retrospective Study of Neural Network Model to Dynamically Quantificate the Severity in COVID-19 Disease
Overview
- Phase
- Not Applicable
- Intervention
- Not specified
- Conditions
- COVID-19 Disease
- Sponsor
- Xinqiao Hospital of Chongqing
- Enrollment
- 1000
- Locations
- 1
- Primary Endpoint
- Calibration
- Status
- Completed
- Last Updated
- 2 years ago
Overview
Brief Summary
The research aim to collect large samples of COVID-19 disease patients with clinical symptoms, laboratory and imaging examination data. Screening the biological indicators which are related to the occurrence of severe diseases. Then, investigators using artificial intelligence (AI) technology deep learning method to find a prediction model that can dynamically quantify COVID-19 disease severity.
Investigators
Jianguo Sun
Deputy Director,Head of Oncology department, Principal Investigator, Clinical Professor
Xinqiao Hospital of Chongqing
Eligibility Criteria
Inclusion Criteria
- •Patients of COVID-19 disease confirmed by virus nucleic acid RT-PCR and CT
Exclusion Criteria
- •unconfirmed suspected cases
- •Patients during pregnancy and lactation
- •incomplete clinical data
- •inestigators considered patients ineligible for the trial
Outcomes
Primary Outcomes
Calibration
Time Frame: up to 3 months
The calibration curves analysis is used to show error between the predicted clinical phenotype with prediction model and actual clinical phenotype.
Net benefit
Time Frame: up to 3 months
Decision curve analysis was used to determine whether the models could be considered useful tools for clinical decisionmaking by comparing the net benefits at any threshold.
discrimination
Time Frame: up to 3 months
The performance of our prediction model is evaluated with the receiver operating characteristic (ROC) curves, areas under the curves (AUCs) and concordance index (c-index).