COVID-19: An opportunity to help by modelling testing and tracing to inform the UK government

Tim Colbourn, a lead­ing UK epi­demiol­o­gist, is look­ing for help with mod­el­ling the effec­tive­ness of test­ing and trac­ing for COVID-19, along with com­par­ing its effect and cost to other in­ter­ven­tions. He has com­posed a doc­u­ment in or­der to sug­gest that a form of test­ing and trac­ing be im­ple­mented by the UK gov­ern­ment: “PTTI: A Pop­u­la­tion-wide Test­ing, con­tact Trac­ing and Iso­la­tion strat­egy for the UK”, [Tim Colbour­net al., 2020] https://​​docs.google.com/​​doc­u­ment/​​d/​​1vTr4IhVdfr5e-CcDfqgv3zjP-Da1Qh5bas09UHsGODU/​​edit

The eco­nomic case for PTTI is clear: at a cost of £33.6bn over 2yrs, it averts an es­ti­mated 271,000 deaths (£124,000 per death averted), 541,000 in­ten­sive care unit (ICU) cases, 812,000 hos­pi­tal­iza­tions, and 34,600,000 Covid-19 cases. Th­ese rep­re­sent £18.0 billion in saved costs for the NHS. There are an ad­di­tional £8.1 billion in pre­vented pro­duc­tivity losses due to ill­ness and lock­down. Then there are all the other costs of lock­down not in­cluded (Table 4). For the next ver­sion we hope to have more de­tailed mod­els, and in­clude more al­ter­na­tive sce­nar­ios e.g. PTTI with­out the test­ing i.e. pop­u­la­tion-wide clini­cal case iden­ti­fi­ca­tion, con­tact trac­ing and iso­la­tion (PTTI), face masks, and even a man­aged herd im­mu­nity strat­egy for com­par­i­son. We wel­come in­put from those in­ter­ested!

Cur­rently, at least in the UK, gov­ern­ment ap­pointed mod­el­lers are not mod­el­ling any al­ter­na­tives to in­ter­mit­tent lock­down and herd im­mu­nity. No other groups seem to be pub­lish­ing such mod­els ei­ther (with the par­tial ex­cep­tion of the pa­per https://​​arxiv.org/​​pdf/​​2003.09861.pdf, which is go­ing to be pub­lished in up­dated form soon). There are also some other mod­el­ling di­rec­tions that can be ex­plored, which haven’t yet been ex­plored in the mod­els gov­ern­ments are us­ing:

  • In­clud­ing the spa­tial di­men­sion to ac­count for the clus­tered na­ture of trans­mis­sion and lo­cal hotspots

  • Do­ing eco­nomic cost benefit like de­ci­sion anal­y­sis so we can know how benefi­cial var­i­ous poli­cies are in com­par­i­son to each other

  • Model­ling other things like the pub­lic use of face­masks, or highly effec­tive drugs be­com­ing available

This is an op­por­tu­nity to im­prove how the UK gov­ern­ment con­trols the spread of COVID-19 af­ter lock­down has ended by helping model the im­ple­men­ta­tion and effects of test­ing and trac­ing. This may also help other gov­ern­ments if they fol­low the UK’s lead or adapt the strat­egy sug­gested.

While we don’t want to for­get our com­par­a­tive ad­van­tages or op­ti­miz­ing for the long run I agree in part with the sen­ti­ment of this post. Less­wrong has already done good work in pre­dict­ing, in­vent­ing, shar­ing op­por­tu­ni­ties, and or­ga­niz­ing to help with COVID-19. Here is an­other op­por­tu­nity that we col­lec­tively have the ex­pe­rience and ex­per­tise re­quired to help with.

If you want to help with mod­el­ling, com­ment in the doc­u­ment above. One com­ment should in­clude your con­tact de­tails so that Tim can con­tact you and so that you can find each other. I’ve started a com­ment thread in the doc­u­ment for this pur­pose.


Some pos­si­bly use­ful re­sources to prompt im­proved mod­els:

My guess at the best type of soft­ware for it: