Posted 1 hour ago,
Reading 2 mins.
in collaboration with
Dr Gérald Kierzek (Medical Director of Doctissimo)
The long Covid is a pathology that is still poorly understood. American scientists have tried to better define the characteristics of this disease, whose symptoms last for many months, using data from nearly 100,000 cases.
Long Covid is marked by many symptoms including shortness of breath, fatigue, fever, headaches, a kind of “brain fog” and other neurological issues.
Objective: to better define long covid
These symptoms can last several months or more after an initial diagnosis of Covid-19, but they are not clear enough and “no examination or test exists to concretely determine this pathology and make the diagnosis in a reliable and safe manner” explains Dr. Kierzek, emergency doctor and medical director of Doctissimo. The objective of this study was therefore to better define long Covid and to determine new therapeutic approaches.
Methodology followed on nearly 100,000 people
To carry out this study, scientists at the University of Colorado examined the patient demographics, health care utilization (such as doctor visits), diagnoses and medications of 97,995 adult patients with Covid-19. .
They used this information, along with data from almost 600 long Covid patients, to create three machine learning models to identify long Covid patients. The latter were classified as such from the moment they presented symptoms of Covid at least 90 days after their initial contamination.
“The symptoms of long Covid are subjective, and can evoke other pathologies, there are also psychosomatic causes. insists Dr. Kierzek. “The interest of this type of work is to determine this pathology as well as possible, to be able to make a clear diagnosis to the patients who suffer from it and to avoid them medical wandering..
Three basic groups
These three models created therefore grouped together: all patients affected by Covid-19, patients hospitalized with Covid-19 and patients with Covid-19 but not hospitalized. Machine learning systems were launched and classified the approximately 100,000 patients from the database. For this, the profiles were crossed with those that closely matched those with a long Covid, looking for their common characteristics.
The modeling proved to be relevant, as the people identified as being at risk for long Covid were similar to the patients seen in the clinics receiving these patients. “In this type of study, with a large cohort, the interest is to target symptoms on a large scale, which we, doctors, would not notice by seeing patients individually” adds Dr. Kierzek. This then makes it possible to dwell more on these symptoms to understand what could have led them to a long Covid (risk factor? specific treatment during the acute phase of Covid which could have favored a long Covid?…) and possibly to diagnose it.
Consult a GP online
New tools for medicine
Although these computer tools “do not replace traditional medicine and consultation with a doctor” as Dr. Kierzek again points out, they can bring certain advances. This is what Professor Emily Pfaff, Ph.D., clinical computer scientist at the University of North Carolina at Chapel Hill, in the United States, and co-author of this work, recalls: “It made sense to leverage modern data analytics tools and a single big data resource, where many features of the long Covid can be represented.“. The ultimate goal is to integrate this type of model to improve their performance and become a tool to help diagnose long Covid.