Clinical and Radiomic Model of COVID-19

Completed

• Conditions: Coronavirus; Machine Learning
• Interventions: Diagnostic Test: Machine learning model

• Registration Number: NCT04337502
• Lead Sponsor: Maastricht University

Brief Summary

To develop and validate a machine-learning model based on clinical, laboratory, and radiological characteristics alone or combination of COVID-19 patients to facilitate risk Assessment before and after symptoms and triage (home, hospitalization inward or ICU).

Detailed Description

In December 2019, a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; earlier named as 2019-nCoV), emerged in Wuhan, China. The diseases caused by SARS-CoV-2 is COVID-19. As of March 8, 2020, more than 100 000 COVID-19 patients have been reported globally (more than 80 000 cases in China, more than 20 000 in other countries), and 3 600 patients (3 100 in China, 500 outside of China) have died. The outbreak of COVID-19 constitutes a Public Health Emergency of International Concern.

Among COVID-19 patients, around 80% are mild (non-severe) illness patients, who usually heal within two weeks. However, another 20% of patients may aggravate into a severe or critical illness which results in a longer hospital stay, and the mortality rate for such patients is 13.4%. Therefore, inchoate identification of the high-risk severe patients is extremely important for patient management and medical resource allocation. General quarantine and symptomatic treatment can be used for most non-severe patients, while a higher level of care and green channel to the intensive care unit (ICU) are helpful for severe patients. Previous studies have summarized the clinical and radiological characteristics of severe COVID-19 patients, while which factors are important predictors is still unclear.

Machine learning is a branch of artificial intelligence that enables us to learn knowledge and potential laws from the given data and to build a model for solving problems as human needs. In recent years, machine learning has been developed as a novel tool to analyze large amounts of data from medical records or images. Previous modeling studies focused on forecasting the potential international spread of COVID-19.

Therefore, our purpose is to develop and validate a machine-learning model based on clinical, laboratory, and radiological characteristics alone or combination of COVID-19 patients in the early stage without severe illness from multiple centers for the prediction of severe (or critical) illness in the following hospitalization to facilitate risk Assessment before and after symptoms and triage (home, hospitalization inward or ICU).

Study Timeline

• Study Start: 2019-12-23
• Primary Completion: 2020-01-20
• Status Verified: 2020-03
• Study Completion: 2020-03-03
• Study First Submitted: 2020-04-06
• Study First Submitted QC: 2020-04-06
• Last Update Submitted: 2020-04-06
• Study First Posted: 2020-04-07
• Last Update Posted: 2020-04-07

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 300

Inclusion Criteria

• confirmed COVID-19 patients by high-throughput sequencing or real-time reverse-transcriptase polymerase-chain-reaction (RT-PCR) assay for nasal and pharyngeal swab specimens.

Exclusion Criteria

1. patients with severe illness when admitted;
2. time interval > 2 days between the admission and examinations;
3. absent data or delayed results

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
severe group	Machine learning model	The severe group was designated when the patients had one of the following criteria during hospitalization issued by the Chinese National Health Committee (Version 3-5). 1) Respiratory distress with respiratory frequency ≥ 30/min; 2) Pulse Oximeter Oxygen Saturation ≤ 93% at rest; 3) Oxygenation index (artery partial pressure of oxygen/inspired oxygen fraction, PaO2/FiO2) ≤ 300 mmHg; 4) One of the conditions as following: a) respiratory failure occurs and requires mechanical ventilation; b) Shock occurs; c) ICU admission is required for combined organ failure.
non-severe group	Machine learning model	The non-severe group was designated when the patients did not occur in the mentioned severe criteria until discharged from the hospital.

Primary Outcome Measures

Name	Time	Method
Predictive performance	Janunary 1, 2020, to February 13, 2020	AUC, accuracy, sensitivity, and specificity

Trial Locations

Locations (1)

The central hospital of Wuhan; 🇨🇳 Wuhan, Hubei, China