Chest pain, persistent cough, stomach pain – these early corona symptoms usually differ depending on the age group. This is shown by a new study. (Symbol image)

© Imago

According to a new study, early corona symptoms differ significantly depending on the age group. An AI model makes this knowledge possible.

London / Kassel – The symptoms of a corona infection are diverse, in addition to typical cold signs such as cough, sore throat and a runny nose, abdominal pain or shortness of breath can also indicate the coronavirus, among other things. However, especially in the upcoming colder seasons, it would be practical to quickly see whether it is the flu or actually Corona *.

Researchers at King’s College London have now addressed this question. In a new study, the team tried to train artificial intelligence (AI) to determine the likelihood of a Covid-19 infection based on early symptoms.

The study published in the journal “The Lancet Digital Health” also revealed differences in early symptoms in the various age groups: For example, the symptoms in the first three days in 16 to 39 year olds differ significantly from those in over 60 and over 80 -Year-olds.

AI model examines early symptoms to detect corona infection

In previous research approaches, many AI models only took into account the symptoms at the peak of an illness, the scientists explained. At the beginning of a corona infection, however, certain symptoms could be better indicators than in the later course of the disease. A press release said that the new results could be used to adapt recommendations for home isolation and tests at an early stage.

University: King’s College London
Founding: 1829
Number of students: 27.629 (Stand: 2016)
Admission rate: 13 percent (as of 2014)

Data from the Covid Symptom Study app were used for the study. This was developed by Zoe, a health science company. Zoe collaborated with King’s College London and Massachusetts General Hospital in Boston for the development. Users of the app can provide information about their Covid 19 symptoms and share their PCR test results with the research team.

For their study, the team examined the statements of 182,991 sick people about 18 symptoms in the first three days after the onset of the symptoms. The data was made available between the end of April and mid-October 2020. The AI ​​was trained with this data, which then analyzed the data of 15,049 other test subjects between mid-October and the end of November 2020.

Study: Loss of smell and chest pain are the most common corona symptoms

Relevant symptoms for the early detection of a Covid-19 infection were therefore loss of smell, chest pain, persistent cough, stomach pain, blisters on the feet, strained eyes and unusual muscle pain.

Across the sexes, the most common symptoms were loss of smell and chest pain. Although fever is one of the known corona symptoms, it was often not meaningful in the first three days. A study by Harvard University recently showed which people can be corona superspreaders *.

Early corona symptoms differ depending on the age group

However, the symptoms that indicated an infection with the coronavirus differed in the different age groups. While persistent coughing was a clear sign of an illness in people between 40 and 59 years of age, in younger people the loss of smell instead indicated a Sars-Cov-2 infection and the infectious disease Covid-19.

According to Artificial Intelligence, diarrhea, sore throats or muscle pain were the first signs in people over the age of 80. However, the corona symptoms * also differ between vaccinated and unvaccinated people.

New AI model detects corona infections based on symptoms more often than other models

The AI, which was co-developed by King’s College London, does not always determine a Covid 19 disease, but it is still more accurate than other algorithms that diagnose or predict a corona infection. The sensitivity, i.e. the value with which corona positives were also correctly predicted, was 73 percent. Lying down was the specificity, i.e. the probability that corona negatives are also predicted as correctly negative, at 72 percent.

The authors of the study now hope that the model will be well received. Liane dos Santos Canas, first author of the research, also hopes that knowing about the different symptoms can help get people to have tests as early as possible so that the risk of spread is minimized. (Nail Akkoyun) * is an offer from IPPEN.MEDIA.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.