New research shows that symptoms for COVID-19 early infection vary between ages and between men, women and children. These differences are more evident in younger age groups (16-59 years), compared with older age groups (60-80 years and over), while men experience different symptoms than women during the early stages COVID-19 infection.The paper was published in The Lancet Digital Health today by King's College London researchers. It analyzes data from ZOE COVID's Symptom Study app from April 20th through October 15th 2020. Thanks to a joint initiative between the Department of Health and Social Care, app contributors can be tested as soon as any new symptoms are reported. Researchers used three days of self-reported symptoms to model the signs and symptoms of COVID-19 infection. They were able to detect 80% of cases.Researchers compared the abilities to predict COVID-19 infection early signs using current National Health Service UK criteria with a Hierarchical Gaussian process model, which is a type of machine-learning.The machine learning model could incorporate certain characteristics about the affected person, such as their age, sex, health conditions and other factors. It showed that early COVID-19 symptoms are different between groups.We examined 18 symptoms, each with different significance for early detection. For early detection of COVID-19, the most important symptoms were loss of smell, persistent cough, stomach pain, abdominal pain and blisters on feet. Eye soreness, unusual muscle pain and chest pain are all common signs. The loss of sense of smell was less important for people older than 60 and not as relevant for those over 80. Diarrhoea was a key sign in older age groups (60-79, >80). Fever, although a well-known symptom of disease was present in all age groups, was not an early sign of the disease.Men reported more frequent symptoms such as fatigue, shortness of breath, chills, and shivers than women. Women were more likely, however, to experience loss of smell, chest pain, and persistent cough.These models were created in the COVID Symptom Study app. However, they were also replicated over time which suggests that they could be used by non-app contributors. The models were tested on the Alpha strain of the virus, but the key findings indicate that symptoms from the Delta variant will differ between populations.Claire Steves, Reader, King's College London, was the lead author. She stated: "It is important that people understand that the earliest symptoms can be broad-ranging and may appear differently for each member of a household or family. It is possible to update the testing guidance to allow cases to be detected earlier in light of emerging variants that are highly transmissible. For people suffering from any of these non-core symptoms, lateral flow tests could be used.Dr Liane Dos Santos Cas, the first author, was a King's College London student. She stated: "Currently in the UK, only few symptoms can be used to recommend self-isolation or further testing. We can detect COVID-19 cases more accurately by using AI and a greater number of symptoms, but only after a few days. This method can be used to encourage more people get tested to reduce spread of COVID-19.Dr Marc Modat is a Senior Lecturer at King's College London. He said, "As part our study, we were able to identify the profile of symptoms due COVID-19 differences between one group and another. This means that people should be encouraged to have their test done using information about them, such as their age. A larger number of symptoms may be possible to take into consideration, in order to account for the various manifestations of the disease among different groups.
