Can Google searches help to predict Covid peaks?
Researchers say they can help us tackle the pandemic and future outbreaks
Before the pandemic, health experts typically advised the public not to research their ailments online. Ask Dr Google about your symptoms and you could end up severely misdiagnosing your condition, making it sound much better or worse than it really is. Hypochondriacs especially ran the risk of disappearing down a rabbit hole of misplaced anxiety.
But Covid-19 has changed all that. From the early weeks of the disease, the Government and NHS listed all the symptoms online, as they urged potential sufferers to self-isolate. New symptoms have been added when they’ve been confirmed.
Experts continue to warn about the nonsense you can read online – from 5G spreading the virus to eating garlic to stop yourself catching it - but the internet quickly became a potent tool for informing the public about this new disease.
Unsurprisingly, Covidrelated searches dominated Google in 2020 (and probably will again in 2021). ‘Coronavirus symptoms’ was the fourth most-searched term in the UK last year, and eighth worldwide. ‘How to make a face mask?’ was the UK’S most-asked ‘How to’ question, followed by ‘How to make hand sanitiser?’ (see more at www.snipca.com/36953).
Now scientists think that by analysing what people search for, they can predict a peak in cases 17 days in advance. The claim comes from researchers at University College London (UCL), who found an association between searches about skin rash and a loss of smell – both recognised Covid symptoms – and a rise in cases. To get more accurate results, they adjusted their analysis to reduce “public interest bias”, which is the effect media coverage has on online searches.
The research, published in Nature Digital Medicine, builds on existing studies that show a link between internet searches and the spread of seasonal flu (see UCL’S Flu Detector: www.snipca.com/37322). Dr Vasileios Lampos, who led the work, says it “provides a new set of tools that can be used to track COVID-19”. His team has been sharing the findings weekly with Public Health England to support its response to the disease.
Dr Lampos said that the methodology works in different countries “irrespective of cultural, socioeconomic and climate differences”, including the US, Greece, Canada, Italy, Australia and South Africa, as well as the UK. The same pattern appeared no matter where they ran their model (see screenshot left). He now hopes that analysing internet searches will become an accepted part of tackling all health emergencies, helping the world “be better prepared for the next pandemic”.
The tech giants are playing their part. In September, Google released a collection of over 400 search trends designed to help researchers study the link between queries about symptoms and the spread of the virus, though for now this data is available in the US only.
In a blog post ( www.snipca.com/37320), Google Health researcher Evgeniy Gabrilovich said the data can provide “an earlier and more accurate indication of the reemergence of the virus in different parts of the country”.
He added that because the searches cover a broad range of conditions, including diabetes and stress, they could also be useful in studying “the secondary health effects of the pandemic”.
It seems clear that while searching for symptoms carries inevitable risks, and misinformation is rife online, there’s a net benefit to society. When Covid is finally under control, asking Dr Google may actually be encouraged.
‘Coronavirus symptoms’ was the fourth most-searched term in the UK last year, and eighth worldwide