AUTHOR=Bowie Cam , Friston Karl TITLE=A follow up report validating long term predictions of the COVID-19 epidemic in the UK using a dynamic causal model JOURNAL=Frontiers in Public Health VOLUME=Volume 12 - 2024 YEAR=2024 URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2024.1398297 DOI=10.3389/fpubh.2024.1398297 ISSN=2296-2565 ABSTRACT=Background -this paper asks whether Dynamic Causal modelling (DCM) can predict the long-term clinical impact of the COVID-19 epidemic. DCMs are designed to continually assimilate data and modify model parameters, such as transmissibility of the virus, changes in social distancing and vaccine coverage-to accommodate changes in population dynamics and virus behaviour. But as a novel way to model epidemics do they produce valid predictions? We presented DCM predictions 12 months ago, which suggested an increase in viral transmission was accompanied by a reduction in pathogenicity. These changes provided plausible reasons why the model underestimated deaths, hospital admissions and acute-post COVID-19 syndrome by 20%. A further 12-month validation exercise could help to assess how useful such predictions are. Methods -we compared DCM predictions-made in October 2022-with actual outcomes over the twelvemonths to October 2023. The model was then used to identify changes in COVID-19 transmissibility and the sociobehavioural responses that may explain discrepancies between predictions and outcomes over this period. The model was then used to predict future trends in infections, long-Covid, hospital admissions and deaths over 12-months to October 2024, as a prelude to future tests of predictive validity. Findings -Unlike the previous predictions-which were an underestimate-the predictions made in October 2022 overestimated incidence, death and admission rates. This overestimation appears to have been caused by reduced infectivity of new variants, less movement of people and a higher persistence of immunity following natural infection and vaccination. Interpretation -despite an expressive (generative) model, with time-dependent epidemiological and sociobehavioural parameters, the model overestimated morbidity and mortality. Effectively, the model failed to accommodate the "law of declining virulence" over a timescale of years. This speaks to a fundamental issue in long-term forecasting: how to model decreases in virulence over a timescale of years? A potential answer may be available in a year when the predictions for 2024-under a model with slowly accumulating T-cell like immunity-can be assessed against actual outcomes.