London is no longer a coronavirus infection hotspot, according to the latest data from a symptom tracking app.

Just two per cent of people are likely to have Covid-19, according to the modelling – down from more than six per cent just over two weeks ago.

Nationally, the predicted number of coronavirus cases fell from 2 million to just over 500,000 in two weeks – down more than 70 per cent.

The app – developed by researchers at Kings College London and health tech startup Zoe – asks users to record any signs of illness, or an absence of symptoms.

Millions of people are logging their health on the app, including more than 164,000 in London.

From the data, scientists model the predicted rate of infection in each area – although they can not account for people who have the virus but are not showing symptoms.

Researchers believe the rate of infection has been cut in three in the capital in just two weeks.

Tower Hamlets (1.68 per cent), Greenwich (1.71 per cent) and Croydon (1.73 per cent) now appear to have the lowest rates of coronavirus.

The City of London (2.78 per cent), Barking and Dagenham (2.56 per cent) and Newham (2.45 per cent) remain the worst affected areas.

But every borough has seen significant decreases since April 1, when infection rates appeared to peak in London.

According to the modelling, the City of London (8.68 per cent), Lewisham (6.59) and Hammersmith and Fulham (6.48) had the most cases at the start of the month.

Lead researcher Professor Tim Spector, of Kings College London, said it was “very reassuring” to see the predicted number of cases falling – but warned this was “definitely not the time for complacency”.

He said: “We believe our population’s symptoms are changing around two weeks before most people are admitted to hospital.

“The data from the app is giving us insight into just how common the virus is and how differently it affects people.”

He added: “What the data tell us is that there is still a large number of infectious people in the UK with mild symptoms, so to quickly lift the lockdown would not be appropriate.”

But Professor Spector said his team was working with the NHS to see if the data can be used “to speed up and guide the lockdown lift”.