Artificial Intelligence Algorithms for Discriminating Between COVID-19 and Influenza Pneumonitis Using Chest X-Rays

• Conditions: Pneumonia, Viral; Influenza With Pneumonia; Pneumonia, Interstitial; Pneumonia, Ventilator-Associated; Pneumonia Atypical; COVID-19; Flu Like Illness; Flu Symptom
• Interventions: Diagnostic Test: Scanning Chest X-rays and performing AI algorithms on images

• Registration Number: NCT04313946
• Lead Sponsor: Professor Adrian Covic

Brief Summary

This project aims to use artificial intelligence (image discrimination) algorithms, specifically convolutional neural networks (CNNs) for scanning chest radiographs in the emergency department (triage) in patients with suspected respiratory symptoms (fever, cough, myalgia) of coronavirus infection COVID 19. The objective is to create and validate a software solution that discriminates on the basis of the chest x-ray between Covid-19 pneumonitis and influenza

Detailed Description

This project aims to use artificial intelligence (image discrimination) algorithms;

* specifically convolutional neural networks (CNNs) for scanning chest radiographs in the emergency department (triage) in patients with suspected respiratory symptoms (fever, cough, myalgia) of coronavirus infection COVID 19;

* the objective is to create and validate a software solution that discriminates on the basis of the chest x-ray between Covid-19 pneumonitis and influenza;

* this software will be trained by introducing X-Rays from patients with/without COVID-19 pneumonitis and/or flu pneumonitis;

* the same AI algorithm will run on future X-Ray scans for predicting possible COVID-19 pneumonitis

Study Timeline

• Study First Submitted: 2020-03-17
• Study First Submitted QC: 2020-03-17
• Study Start: 2020-03-18
• Study First Posted: 2020-03-18
• Status Verified: 2020-04
• Last Update Submitted: 2020-04-23
• Last Update Posted: 2020-04-27
• Primary Completion: 2020-08-16
• Study Completion: 2020-08-18

Recruitment & Eligibility

• Status: UNKNOWN
• Sex: All
• Target Recruitment: 200

Inclusion Criteria

• flu-like symptoms: myalgia, cough, fever, sputum
• Chest X-Rays
• COVID-19 biological tests

Exclusion Criteria

• patient refusal
• uncertain radiographs
• uncertain tests results

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
Symptomatic Patients	Scanning Chest X-rays and performing AI algorithms on images	Our goal is to identify an artificial intelligence algorithm that can be run on lung radiographs in patients with influenza / respiratory viral symptoms who come to the emergency department / triage. This algorithm aims to identify the radiographs of patients with COVID-19 and those with influenza pneumonitis, with accuracy verified by COVID-19 tests.

Primary Outcome Measures

Name	Time	Method
COVID-19 negative X-Rays	6 months	Number of participants with pneumonitis on Chest X-Ray and COVID 19 negative
COVID-19 positive X-Rays	6 months	Number of participants with pneumonitis on Chest X-Ray and COVID 19 positive

Trial Locations

Locations (3)

U.O. Multidisciplinare di Patologia Mammaria e Ricerca Traslazionale; Dipartimento Universitario Clinico di Scienze Mediche, Chirurgiche e della Salute Università degli Studi di Trieste; 🇮🇹 Cremona, Italy

Department of Cardiology at Chelsea and Westminster NHS hospital; 🇬🇧 London, United Kingdom

University of Medicine and Pharmacy Gr T Popa; 🇷🇴 Iaşi, Romania