Researchers from the University of Copenhagen in Denmark have determined that artificial intelligence (AI) applied to health data can predict, with a 90% rate of accuracy, if a person will succumb to COVID-19.

To develop the AI system, the research team fed Computerome, a secure supercomputer for personal data, health data from more than 3,900 patients in Denmark, training the computer to recognize patterns and correlations among risk factors.

According to the team’s findings, the health factors with the greatest influence on whether an uninfected person will, once infected, be put on a respirator or succumb to COVID-19 include (in order of importance) body mass index, age, high blood pressure, gender (with men being more likely to succumb to the virus than women), neurological diseases, chronic obstructive pulmonary disease (COPD), asthma, diabetes and heart disease.

People with one or more of these health factors could potentially be prioritized when decisions concerning vaccine distributions are being made.

Additionally, the AI could be used to make predictions about the number of people likely to be hospitalized or who require respirators, enabling health care facilities to better prepare for surges.

The team is currently working to develop the technology enough so that the system can eventually predict the number of respirators needed for specific healthcare facilities according to a region’s COVID-19 positive rates data five days in advance.

The research appears in the journal Scientific Reports.

To contact the author of this article, email mdonlon@globalspec.com