Rapid exclusion of COVID-19 infection using AI, ECG technology – sciencedaily

The AI-enhanced ECG was able to detect COVID-19 infection in the test with a positive predictive value – infected people – of 37% and negative predictive value – uninfected people – of 91%. When additional normal controls were added to reflect a 5% prevalence of COVID-19 – similar to a real world population – the negative predictive value increased to 99.2%. The results are published in Proceedings of the Mayo Clinic.

COVID-19 has an incubation period of 10 to 14 days, which is long compared to other common viruses. Many people do not have symptoms of infection and could unknowingly put others at risk. Additionally, the turnaround time and clinical resources required for current testing methods are considerable, and access can be an issue.

“If validated prospectively using smartphone electrodes, it will make the diagnosis of COVID infection even simpler, highlighting what could be done with international collaborations,” said Paul Friedman, MD, chairman of the cardiovascular medicine department at the Mayo Clinic in Rochester. Dr. Friedman is the lead author of the study.

Awareness of a global health crisis brought together stakeholders from around the world to develop a tool that could address the need to quickly, non-invasively and cost-effectively eliminate the presence of acute COVID infection. 19. The study, which included data from racially diverse populations, was conducted by a global consortium of volunteers spanning four continents and 14 countries.

“The lessons of this global working group have shown what is feasible and the need has prompted members of industry and academia to come together to resolve the complex issues of data collection and transfer from multiple centers. with their own ECG systems, electronic and variable health records. access to their own data, ”says Suraj Kapa, ​​MD, cardiac electrophysiologist at the Mayo Clinic. “The relationships and data processing frameworks refined through this collaboration may support the development and validation of new algorithms in the future.”

Researchers selected patients with ECG data around the time their diagnosis of COVID-19 was confirmed by genetic testing for the SARS-Co-V-2 virus. This data was compared to the control with similar ECG data from patients who were not infected with COVID-19.

The researchers used more than 26,000 ECGs to train the AI ​​and nearly 4,000 more to validate its readings. Finally, the AI ​​was tested on 7,870 previously unused ECGs. In each of these sets, the prevalence of COVID-19 was approximately 33%.

To accurately reflect a real-world population, over 50,000 additional normal ECGs were then added to achieve a 5% prevalence rate of COVID-19. This increased the negative predictive value of AI from 91% to 99.2%.

Zachi Attia, Ph.D., an engineer from the Mayo Clinic in the Department of Cardiovascular Medicine, explains that prevalence is a variable in calculating positive and negative predictive values. Specifically, as the prevalence decreases, the negative predictive value increases. Dr Attia is co-first author of the study with Dr Kapa.

“Accuracy is one of the biggest hurdles in determining the value of any test for COVID-19,” says Dr. Attia. “Not only do we need to know the sensitivity and specificity of the test, but also the prevalence of the disease. The addition of the additional control ECG data was essential to demonstrate how a variable prevalence of the disease – as we have it. we met with regions with very different disease rates at different stages of the pandemic – would have an impact on how the test would work. “

“This study demonstrates the presence of a biological signal in the ECG compatible with COVID-19 infection, but it included many sick patients. Although this is a promising signal, we need to test it well. prospectively in asymptomatic people using electrodes on smartphones to confirm that it can be practically used in the fight against the pandemic, “notes Dr. Friedman.” Studies are underway to answer this question. “

About this study

This study was conceived and designed by researchers at the Mayo Clinic, and the work was made possible in part through a philanthropic donation from the Lerer Family Charitable Foundation Inc., and through the voluntary support of participating physicians and hospitals. from around the world who have contributed to an effort to combat the COVID-19 pandemic. Technical support was offered by GE Healthcare, Philips and Epiphany Healthcare for the transfer of the ECG data.

Source of the story:

Materials provided by Mayo Clinic. Original written by Terri Malloy. Note: Content can be changed for style and length.