This is a summary, written by members of the CITF Secretariat, of:

Khan AA, Sbihi H, Irvine MA, Jassem AN, Joffres Y, Klaver B, Janjua N, Bharmal A, Ng CH, Wilmer A, Galbraith J, Romney MG, Henry B, Hoang LMN, Krajden M, Hogan CA. Prediction of SARS-CoV-2 transmission dynamics based on population-level cycle threshold values: A Machine Learning and mechanistic modeling study. medRxiv. 2023 Mar 6. doi:

The results and/or conclusions contained in the research do not necessarily reflect the views of all CITF members.

In a preprint, not yet peer-reviewed, CITF-funded researchers found that machine learning and epidemic transmission modelling could accurately predict cases of SARS-CoV-2 infection during the Omicron wave. The modelling used the PCR test results of asymptomatic individuals. CITF-funded researcher Dr. Marc Romney and former CITF Leadership Group member Dr. Mel Krajden (both at the University of British Columbia) contributed to this study.

Key findings:

  • Several machine learning models, tested on large simulated datasets, accurately predicted simulated SARS-CoV-2 epidemic trends.
  • The epidemic transmission model accurately predicted the actual SARS-CoV-2 incidence observed in British Columbia. The model was also validated using data from an outbreak in a long-term care patient facility.
  • Both approaches had to be adjusted to a new context, in which Omicron (which has a shorter incubation period than other SARS-CoV-2 variants) was predominant and in which vaccination coverage was very heterogeneous in the population. Despite this, the models performed equally well.

Early in the COVID-19 pandemic, SARS-CoV-2 seroprevalence was estimated by analyzing PCR tests performed in the Canadian population. Today, the use of PCR tests is mostly limited to certain groups of symptomatic or vulnerable individuals. Rapid antigen tests (RATs) are the predominant method to test for SARS-CoV-2 infection now in the general population. Because RAT results are not consistently reported to public health officials, new tools are needed to assess the incidence of SARS-CoV-2 in the entire population.

Models can be used to predict SARS-CoV-2 transmission dynamics and provide support that is critical to decision-making and planning for resource allocation, vaccination strategies, and isolation practices. As well, models can be developed for specific settings, such as hospitals and long-term care facilities.

The study included individuals with a PCR-confirmed SARS-CoV-2 infection by nasopharyngeal swab or saline gargle between November 19, 2021 and January 8, 2022. This captured the emergence of the Omicron wave in British Columbia.