Scientists at MarSU determine coronavirus pneumonia using artificial intelligence
Timely diagnosis of pulmonary diseases remains important for many people on our planet. Today, there is an urgent need for regular monitoring of lung conditions to prevent or detect pneumonia. Correct and quick diagnosis is an important step in the treatment.

A digital medicine project has been launched at MarSU, aimed at searching for anomalies in x-rays using deep machine learning. Radiography is a technically simple and most accessible method for the preliminary diagnosis of the disease for the population, including tracking the disease at an early stage. This type of research is widespread throughout the world and remains a priority in medical practice, despite the presence of other, more complex and effective diagnostic methods.
Thanks to the scientific developments of university scientists, pneumonia can be determined with an accuracy of 98%. Neural networks analyze the digitized X-ray images of the lungs and make a conclusion, which indicates the presence or absence of signs characteristic of coronavirus pneumonia, which allows us to calculate the probability of a diagnosis of “COVID-19”.
There are many datasets containing x-rays of the lungs, both healthy and pathological. The research by MarUU research practitioners was carried out both on open dataset bases (Kaggle and others), and on a full-scale database of real images of medical institutions.

The project will be further developed. A study is underway to train neural networks to determine viral and bacteriological pneumonia. It is planned to open a research laboratory in the field of digital medicine at the medical faculty.