Transcript Slide 1
Social technologies for community response to epidemics Chris Watkins Department of Computer Science Royal Holloway, University of London The effect of public health measures on the 1918 influenza pandemic in US cities M Bootsma and N Ferguson, PNAS 2007 Public health interventions and epidemic intensity during the 1918 influenza pandemic R Hatchett, C Mecher, and M Lipsitch, PNAS 2007 US cities that implemented NPIs in 1918 had lower mortality rates than those cities that did not. NPI: non-pharmaceutical intervention, such as • closing schools • banning large public gatherings • isolation of the sick • ... Could we implement more effective public health measures than in 1918? How much could we reduce the intensity of a severe pandemic by enabling people not to catch it? What social technologies are available? On-line social networking Localisation and tracking Voted discussion systems Distributed community support On-line social networking Facebook, but changing rapidly Localisation and tracking Smartphones, using wifi signal strength Our digital footprints Voted discussion systems Reddit, Yahoo answers, MOOCs Distributed community support Protocols for local coordination and discussion ??? Pandemics: mild and severe Case A: mild Probable in next 20 years Health service emergency: daily life as usual Case B: severe Possible Societal emergency: supplychain disruption? People unwilling to change behaviour much People willing to change behaviour given tools and a plan Extraordinary measures? Ordinary public health measures Pandemics: mild and severe Case A: mild Realistic to assume exponential growth and uncontrolled spread Case B: severe Realistic policy aim is mitigation Policy aim could be suppression / sub-exponential growth Travel restrictions futile Fatalistic to assume exponential growth and uncontrolled spread If aim is to contain local outbreaks, travel restrictions justified. Is there a plan for case B? Changing community behaviour Goals worth achieving Communicate a plan Open discussion Provide information and tools The People Changing community behaviour Goals worth achieving Communicate a plan Open discussion Provide information and tools The People A computer scientist’s reaction: What?! 1.4 < R0 < 2.5 !? That’s incredibly valuable information. So all we have to do to contain an epidemic is to ensure that each person who gets sick infects one person fewer! On-line social networking • Less than 10 years old • Social graph: database of posts, pictures, links connecting people with unique ids. • Who do we really meet? Facebook knows. – Co-tagging in photographs; auto face recognition – Locations of home and work known. – Rich local network => contact network? On-line networking: research questions Does the spread of infectious disease correlate with the contact network inferred from Facebook? If so: - emerging disease surveillance - information on personal infection risk - social distancing: when should I stay in? - can social conventions be altered so that people post updates of their health? Localisation and trail recording Smartphones: - have their position recorded by network approx 100 metres - could run an app that repeatedly records pattern of wifi signal strengths. Localisation within buildings, to a single room or within a few metres. - Trails of locations recorded in encrypted form and uploaded for encounter analysis. Digital footprints: - Payment cards, travel cards. - Correlation of multiple evidence of movement and activity Research questions Could localisation and trail recording be viably used for - real-time epidemiology? - automatic contact tracing? How close to a complete real-time picture of an epidemic could we get with current technology? Voted Discussion Systems Mostly less than 10 years old Reddit: as many visitors as New York Times. Slashdot, Yahoo answers, Quora, (Digg), (Stumbleupon), and many more. MOOCs are newest and most sophisticated: rapid development ! • • • • Searchable discussion threads on many topics Individual comments get voted up or down System estimates which posts will be up-voted: avoids ‘first post’ problem Users accumulate individual ‘karma’ score Effects are: • Posts that are angry, stupid, badly written, crazy, ignorant, or impolite get voted down out of sight. • Everyone wants to be upvoted: huge incentive to – to write well and thoughtfully, – to obey community standards – for relevant and courteous discussion Questions To what extent can communications technology enable people to collectively avoid infection? What information can be generated ? How can this information best be used?