CT Biomarkers Identification by Artificial Intelligence for COVID-19 Prognosis
- Conditions
- Covid-19
- Interventions
- Diagnostic Test: Imaging by thoracic scanner
- Registration Number
- NCT04418245
- Lead Sponsor
- Centre Hospitalier Universitaire de Nīmes
- Brief Summary
The study hypothesis is that low-dose computed tomography (LDCT) coupled with artificial intelligence by deep learning would generate imaging biomarkers linked to the patient's short- and medium-term prognosis.
The purpose of this study is to rapidly make available an early decision-making tool (from the first hospital consultation of the patient with symptoms related to SARS-CoV-2) based on the integration of several biomarkers (clinical, biological, imaging by thoracic scanner) allowing both personalized medicine and better anticipation of the patient's evolution in terms of care organization.
- Detailed Description
Not available
Recruitment & Eligibility
- Status
- WITHDRAWN
- Sex
- All
- Target Recruitment
- 1000
- Patients positive for SARS-CoV-2 according to RT-PCR test between 1st March and 31st May 2020
- Patients undergoing low dose CT scan to establish Covid-19 lung damage
- Available for at least 8 days follow-up
• Patients opposing the retrospective use of their data
Study & Design
- Study Type
- OBSERVATIONAL
- Study Design
- Not specified
- Arm && Interventions
Group Intervention Description Patients positive for SARS-CoV-2 Imaging by thoracic scanner -
- Primary Outcome Measures
Name Time Method Patient requiring more than 3 liters of oxygen to maintain a saturation >95% (intensive care unit or resuscitation department) Day 8 Yes/no
Percentage of lung affected by condensation on scan Day 0 % calculated by deep learning
Percentage of lung affected by ground glass opacity on scan Day 0 % calculated by deep learning
Vital status Day 8 Dead/alive
Percentage of lung affected on CT Day 0 % ground glass and condensation calculated by deep learning
- Secondary Outcome Measures
Name Time Method C-reactive protein levels Admission Day 0 mg/L
Time until onset of symptoms Admission Day 0 Days
Vital status Day 30 Dead/alive
Length of hospitalization Maximum 30 days Days
Percentage of lung affected by ground glass opacity on scan Day 16 % calculated by deep learning
Percentage of lung affected by condensation on scan Day 16 % calculated by deep learning
rehospitalization Day 30 Yes/no
Duration of intubation Day 30 Days
D Dimers level Admission Day 0 µg/L
Time between RT-PCR positive results and first scan Admission Day 0 Hours
Medical history of cardiovascular disease Admission Day 0 Yes/no: hypertension, coronary artery disease, congestive heart failure, cardiac arrhythmia
Percentage of lung affected on CT Day 16 % ground glass and condensation calculated by deep learning
BMI> 30 Admission Day 0 Yes/no:
Software operating time End of study (August 2020) Speed of image loading and image processing depending of brand of scanner
lymphocytemia Admission Day 0 g/L
Diabetes Admission Day 0 Yes/no
Medical history of immunosuppressed condition Admission Day 0 Yes/no: steroid use, pre-existing immunological condition, current chemotherapy for cancer
Calculate a prognostic score from clinical, biological and CT parameters Day 8 Deep learning algorithm
lactate dehydrogenase Admission Day 0 U/L
Calculate a prognostic score from clinical and biological parameters only Day 8 Deep learning algorithm
Medical history of respiratory disease Admission Day 0 Yes/no: Chronic obstructive pulmonary disease, chronic respiratory failure
Current or previous history of smoking Admission Day 0 Yes/no:
Age Admission Day 0 Years
Compare receiver operating curves of prognostic scores with and without CT parameters Day 8
Trial Locations
- Locations (6)
CHU la Timone
🇫🇷Marseille, France
CHU Montpellier
🇫🇷Montpellier, France
CHU Strasbourg
🇫🇷Strasbourg, France
CHU Poitiers
🇫🇷Poitiers, France
CHU de Nimes
🇫🇷Nîmes, France
CHU Martinique
🇲🇶Fort-de-France, Martinique