Development of Machine Learning Algorithms for Inferring Biomarkers Underlying Multi-Modal Physiological Signals of Patients with COVID-19

COVID-19 is a global pandemic disease and the best way to stop it is controlling its spread and treating the infected individuals. Detailed measures of clinical characteristics and outcomes in patients with COVID-19 like Reverse transcription-polymerase chain reaction (RT-PCR) are not accessible to a large population and require patients to spend hours waiting at the hospitals. As well, it is not yet known that the lung is the only host of this virus; an inflammation of the heart has been recently reported in patients with COVID-19. Most recent studies imply that COVID-19 might directly impact on the heart. Therefore, relying on one method for detecting COVID-19 is not sufficient, multi-modal sensors indicating different physiological activities are required. There is no unique solution to capture different but relevant physiological signals underlying COVID-19. Given medical imaging techniques (chest CT scans), pulmonary function tests (PFTs), and electrophysiological recordings (e.g., ECG and blood pressure) of patients with COVID-19, we, in collaboration with Dena Corporation and University of Toronto, aim to develop machine learning and signal processing algorithms for detecting biomarkers underlying COVID19 and inferring their correlational patterns.

Faculty Supervisor:

Milad Lankarany

Student:

Behnaz Poursartip;Idir Mellal;Vaibhav Bachuwar

Partner:

Dena Corporation

Discipline:

Sector:

Information and cultural industries

University:

University of Toronto

Program: