🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

If I could retweet this thread a 1000 times, I would. What happened with monkeypox is simply astonishing. "Preparedness" looks like an empty word if plans do not translate into actions, or if actions are delayed until the issue has become far too apparent and hard to control.

I was "lucky" to contract ME/CFS close to experts in Oxford, so it "only" took me 8 months to get diagnosed and stabilized back in 1998. I have not had access to see another ME/CFS specialist since then, 2.5 decades and two more post-viral syndromes later

Minimizers: "Our immune systems evolved to function best with regular exposure to airborne viruses." Evolutionary biologists: "The adaptive immune system found in jawed vertebrates (including humans) evolved 500 million years ago in a fish."

The risk of COVID transmission from 7 million contacts, as assessed by the @NHS digital app "the cumulative duration of exposure to infected individuals is a key predictor of transmission" nature.com/articles/s4158… Luca Ferretti Christophe Fraser Group nature

New in nature: we used digital measurements for 7 million people exposed to confirmed COVID-19 cases to determine risks for virus transmission, captured well by NHS COVID-19 app. We found that number of hours of exposure is a major predictor, along with proximity.

The 7 million people - ‘contacts’ - were contact traced by the NHS COVID-19 app. We analysed measurements of their risky exposure to a case linked to the later outcome of whether they reported a positive test in the app (unreported infections are missing). What did we find?

The fraction of contacts testing positive rose steeply with three simple metrics of risk: the max app-estimated risk score at any time during the exposure, the cumulative (summed) risk score over the exposure, and the duration of exposure. See first pic. Shading = 95% uncertainty

We estimated and removed the background risk of getting infected, and grouped the contacts by both exposure duration and app-estimated risk score per unit time (derived from proximity). The fraction of contacts testing positive increased with both factors separately:

The data tell us who tested positive and who didn’t. Each person’s risk was an unobserved combination of risks from the background, the exposure duration, and the exposure proximity. We disentangled and estimated the separate contributions using statistical modelling...

Shown here is the estimated probability of getting infected (and later reporting positive) by one specific case during a 30-minute exposure to them, as a function of the app-measured risk score. When the app thinks there’s twice as much risk of transmission, there really is!

Once we know risk factors, how helpful that is depends on their distribution in the population. We show that here, a-c, for all 7 million contacts (red) and for those testing positive with the background subtracted (‘transmissions’, blue). Transmissions are shifted to higher risk

We used the data to classify contacts approximately reflecting their exposure contexts. e.g. contacts exposed for at least 8 hours in a day are likely household contacts; these made up 6% of contacts but 40% of transmissions (previous pic) showing the increased risk in households

We tested whether the app could have done better by using machine learning methods to estimate risk. We found that simple measures, such as the exposure duration, performed almost as well.

Key takeaways: 1) as the duration of exposure increases, so does transmission risk, in a continuous manner not reflected in binary risk decisions like more/less than 15 mins. This likely holds for other respiratory infections. Duration is easily estimated without apps.

2) Duration and proximity both determine risk. ‘Physical distancing’ strategies to reduce risk should therefore consider the relevance of time as well as space.

3) For COVID, transmission risk continues increasing even after several full days of exposure: if someone in your house tests positive, it’s not a foregone conclusion that you’ve already got it too, you can limit your risk.

4) We have shown that digital contact tracing, like manual contact tracing, is a tool not just for reducing transmission but for understanding it, which helps us learn how to reduce transmission in other ways. We should get better at doing both things before the next pandemic!

5) This is encouraging validation of the risk model used in the NHS COVID-19 app, and for the prospects of digital contact tracing and precision epidemiology more generally. 👇

Authors: Luca Ferretti Chris Wymant @chriswymant.bsky.social James Petrie Adam Fowler 🇬🇧🇺🇦 Daphne Michalsen Tsallis mishkendall AliceLedda, PhD 💙 Francesco Di Lauro Mark Briers Christophe + others. Teams at Pandemic Sciences Institute Big Data Institute Nuffield Department of Medicine UK Health Security Agency. Thanks NHS COVID-19 app & users! Paper & links: 045.medsci.ox.ac.uk/risk_measureme…

A recorded seminar on this: youtube.com/watch?v=wfBiF7… My last four papers: 2020 Science, Ferretti & Wymant... Fraser 2021 Nature, Wymant & Ferretti... Fraser 2022 Science, Wymant... Fraser 2023 Nature, Ferretti & Wymant... Fraser Slow & steady. Wonderful collaborators.