COVID PRECISE is a large international consortium of experts performing research on COVID-19 related prediction models.
The current COVID-19 pandemic has learnt us that our – always too scarce – healthcare resources need to be as precisely as possible tailored to those who need them most. Personalized COVID-19 care. This applies to preventive resources (e.g. mouth masks and vaccinations), to diagnostic testing, to treatments, and to other hospital, intensive care and post-discharge resources.
Multivariable prediction models are increasingly being used in medicine to tailor or personalize healthcare resource use. The number of COVID-19 related prediction models in the medical literature is literally increasing by the day. We even see them appearing already in medical apps and other technological applications and medical guidelines.
However, only the accurate, high quality and generalisable prediction models should be implemented in our daily healthcare systems. COVID PRECISE aims to gather and provide the existing evidence and data on all COVID-19 related prediction models, and provide recommendations on those prediction models that are most likely suitable in daily healthcare to serve tailored, high quality COVID-19 care.
We address three types of prediction models:
- Models to predict COVID-19 susceptibility in the general population (high-risk group identification) to guide tailored use of preventive healthcare resources;
- Models to predict presence of COVID-19 in patients presenting with suggestive but not exclusive symptoms or signs (diagnostic classifications) to tailor the diagnostic capacity to symptomatic patients with a high probability of indeed having COVID-19 infection.
- Models to predict a compromised course in patients diagnosed with COVID-19 (prognostication) to tailor hospital resources including treatments and intensive care resources to those with an estimated poor prognosis.
Our research involves:
- Meta-research or aggregate reviews of all existing (peer-reviewed and pre-print) COVID-19 prediction models, for the above three categories.
- Individual participant data (meta-)analysis, by combining as much as possible datasets from multiple hospitals, practices and countries that allow for developing, validating and tailoring accurate and generalizable COVID-19 prediction models.