Canada is entering the important yet dangerous phase of the COVID-19 pandemic: the reopening of industry. As such, there is an urgent need for a quick and accurate screening tool to help ensure people re-entering the workplace are COVID-19 negative. This proposal offers an innovative, simple-to-implement and quick screening tool for this purpose. This study hypothesize that breathing sounds of a COVID-19 positive person would have different characteristics even if the person is asymptomatic. This study aim the development of an integrated diagnostic pattern recognition tool in the form of a smartphone app, using audio and temperature as inputs to identify COVID-19 positive individuals. The proposed digital technology will screen individuals as healthy or possibly COVID-19 positive. The latter group will then be recommended for further testing. The goal of the proposed app is to provide much more accurate early screening (currently only temperature is taken), and to reduce the burden of COVID-19 tests. This digital technology will be used and tested in Manitoba initially and later nationally in Canada, with the potential of being publicly available in the future. To use the proposed screening tool, a smartphone is held within 1 cm of an individual's mouth and the individual instructed to take five deep breaths through the mouth. The individuals' breathing sounds will be recorded by the smartphone, while the participant's temperature will also be recorded by the heat camera. The app will first use its acoustic analysis to identify sounds as healthy or abnormal. If the outcome is abnormal, then a questionnaire will be provided, along with a further acoustic analysis to rule out other common comorbid conditions (e.g. chronic lung disease). Finally, based on the inputs, the diagnostic algorithm will decide if the individual should be referred for further testing or not. Since the proposed end product is a smartphone app, the two software partner companies will play a crucial role in the final integration and development.
The goal of this project is to develop a digital technology including an interactive app on a
dedicated smartphones to record, analyze and classify a person as likely COVID-19 positive or
negative. The specific objectives are:
1. Develop an initial simple app to record breathing sounds, check the quality of sounds in
real time, message and instruct if the sounds fail the quality check.
2. Recruit and collect breathing sounds of at least up to 200 individuals at one of the
COVID-19 testing stations in Winnipeg.
3. Using data of 50 COVID-19 positive cases and 50 negative ones of the above collection as
the training data, develop a breathing sounds analysis algorithm with high sensitivity
and specificity for smartphones to classify the recorded breathings sounds as healthy or
abnormal. Our current database allows for immediate detection of some underlying
unrelated pathologies, which in the initial stages will be supported by the included
questionnaire. This can be further enhanced at the end-product stage as each worker's
history is built up and daily compared.
4. Develop a smart diagnostic algorithm for smartphones as an app using the above sound
analysis outcome, the subject's temperature and answers to a few questions and screen
for possibility of having Covid-19 in less than 30 seconds.
5. Evaluate the above diagnostic algorithm for the remaining collected data (100 subjects),
and plan for the next phase of study as a triage in one of the city's hospitals.
6. Prepare the final market prototype of the technology and disseminate the results.
Diagnostic Test: suspected of COVID-19 infection
suspected of COVID-19 infection
Inclusion Criteria:
- Adults being tested for COVID-19 by swab fluid test
- Must have a smart phone and Internet access
Exclusion Criteria:
- age<18 years
- cognitively impaired
- having no access to a smart phone or Internet
University of Manitoba
Winnipeg, Manitoba, Canada
Zahra Moussavi, Ph.D., Principal Investigator
University of Manitoba