A decision support tool to improve testing

One of the main recommendations of the World Health Organization (WHO) for containing the spread of COVID-19 is rapid testing. This led to an extremely high demand for diagnostic tests during the first outbreak of cases in the spring of 2020. Roche, which produces molecular and serological tests for the virus in Canada, was unable to distribute the right quantities of tests to laboratories across the country. To avoid disrupting coronavirus testing efforts, the biotech company had to rethink its procurement strategy, and quickly.

This technology literally "learned" how to predict future demand for COVID-19 diagnostic tests.

To that end, Roche called on the expertise of Andrea Lodi, Canada Excellence Research Chair in Data Science for Real-Time Decision-Making at Polytechnique Montréal. The researcher, who is also a member of the Group for Research on Decision Analysis (GERAD), began working intensively on an allocation optimization tool based on artificial intelligence at the end of March. Drawing on huge amounts of data, this technology literally "learned" how to predict future demand for COVID-19 diagnostic tests and then distribute the tests produced according to anticipated propagation scenarios.

Barely six weeks later, Andrea Lodi delivered a 100% operational decision support tool, thanks in part to his collaborators Guillaume Rabusseau and Guy Desaulniers, artificial intelligence supercluster Scale AI and the digital intelligence company Ivado Labs. Roche has since been using this tool to ensure that the right number of diagnostic tests are sent to the right places in real time. Even better, the multinational company can now certify to the various Canadian laboratories—its customers—that it is supplying them fairly and equitably, without bias or favouritism. This innovative technology could soon be deployed elsewhere—in the United States, for example—thanks to its open science approach, which requires rapid open sharing of data and research results.

Andrea Lodi is also working on another project related to COVID-19 with the Centre hospitalier de l'Université de Montréal. The CHUM is looking to optimize the operating room schedule based on the number of beds available in intensive care. This work is still in progress.