#HackfromHome - an anti-viral virtual hackathon.

We’re giving people a chance to respond with action, working together to improve the lives of everyone affect by the Coronavirus. Our goal is to band together to help communities, patients, and their families better using what we know best - technology.

If possible, these ideas should preserve the privacy of individuals through the use of Personal Data Accounts. For more information on Personal Data Accounts read this.

PLEASE DO NOT SUBMIT ANY MORE IDEAS.

IDEA SUBMISSION IS CLOSED.

Capture and Present Better Data on Testing Rates

What's wrong with current data?

The data provided by Authorities on current and increasing infection numbers is missing data on how many tests have been performed to obtain the Active Infection counts we're seeing.


Currently widely tracked and published are:
Regular reporting of:

  • Active Infection Cases

  • Case Fatailites

  • Case "Recoveries" (with different criteria for "recovery" by region)


Sporadic reporting of:

  • Overall number of tests performed (usually country-level), with positive/negative tests counts within that set.

    • different regions are using different tests


This data is likely skewed towards cases presenting at Healthcare providers with more severe symptoms vs. the community spread in the population as a whole.

Why does this matter?

We're getting an incomplete picture of how prevalent the spread of the virus is.

This makes it hard for the general population to judge the risk in their area since a low infection count could just be down to poor intel/a low testing rate.

In turn, this may be an explanation for some regions apparent lack of behaviour change/prevention measures due to underestimated risk from low detection rates.


Poor intel on any community reservoir of infected people may increase the risk of another outbreak through premature relaxing of prevention measures.

What could be different?

Regional data on:

  • Active Infection Cases with:

    • diagnosis criteria

    • initial detection method/presentation

  • Criteria used to judge and or test for COVID19 fatalities among the recently deceased

    • i.e. testing/detection rate among the dead

  • Number of tests with:

    • How the cases were flagged for testing (e.g. high risk area, high risk group, presenting with symptoms at point of care, thermal screening, random sampling etc.)

    • What test(s) were used

    • Which testing facilities


What could we do to help?

  • In lieu of official (and detailed) data from regional authorities, we could create a website or mobile app for self-reporting cases who've been tested, with:

    • test result

    • how they were flagged to be tested

    • symptoms

    • timescales

    • test type if known

    • test location

    • location before presenting for testing

    • home town

Once we have the data, then stronger analysis of/modelling from the existing data can be made, with dashboards, confidence ratings for published infection rates & estimates of likely community spread

Notes & Caveats

  • Data privacy would need to be at the forefront, but the data is probably useful if we can track cases to Town level to help inform local behaviour.

  • Place of test is likely less important than place of likely contacts (catch or spread)

  • Self-reported data means a self-selecting sample - likely biassed towards negative tests, those with less serious symptoms & those who are more comfortable with technology

  • Self-reported may not know the specifics of the test performed

  • Can't verify data is accurate, so likely there will be noise in the data from:

    • data entry mistakes

    • duplicate reporting when reporting on behalf of a patient

    • mis/disinformation/abuse

  • Will
  • Mar 31 2020
  • Attach files
  • Amy Sheon commented
    3 Apr 08:08pm

    Great idea. Make sure to consider the needs of those who are not connected electronically, e.g. no mobile or broadband.

    See my talks Sat on digital divide and vulnerable populations:

    Public Health Perspectives on Covid-19

    Saturday, April 4⋅GMT-4:00 (8:00 am EST)

    Join Zoom Meetinghttps://zoom.us/j/947524053

    and:

    Covid19 and open data sets Saturday, April 4⋅GMT-4:30 (8:30 am EST)

    Join Zoom Meetinghttps://zoom.us/j/947524053

    Amy Sheon

  • Anna Jolliffe commented
    2 Apr 09:26pm

    I like this idea too!

  • Andrew Nesbit commented
    2 Apr 06:14pm

    I love this idea.

  • Will commented
    31 Mar 09:17pm

    Related project starting to track testing date in the USA: https://covidtracking.com/

  • Will commented
    31 Mar 06:06pm

    (I'm having a bad day with this site)

  • Will commented
    31 Mar 06:00pm

    The data provided by Authorities on current and increasing infection numbers is missing data on how many tests have been performed to obtain the Active Infection counts we're seeing.


    Currently widely tracked and published are:
    #### Regular reporting of:

    • Active Infection Cases

    • Case Fatailites

    • Case "Recoveries" (with different criteria for "recovery" by region)


    #### Sporadic reporting of:

    • Overall number of tests performed (usually country-level), with positive/negative tests counts within that set.

      • different regions are using different tests


    This data is likely skewed towards cases presenting at Heathcare providers with more severe symptoms vs. the community spread in the population as a whole.

    ## Why does this matter?

    We're getting an incomplete picture of how prevalent the spread of the virus is.

    This makes it hard for the general population to judge the risk in their area since a low infection count could just be down to poor intel/a low testing rate.

    In turn, this may be an explanation for some regions apparent lack of behaviour change/prevention measures due to underestimated risk from low detection rates.


    Poor intel on any community reservoir of infected people may increase the risk of another outbreak through premature relaxing of prevention measures.


    ## What could be different?
    Regional data on:

    • Active Infection Cases with:

      • diagnosis criteria

      • initial detection method/presentation

    • Criteria used to judge and or test for COVID19 fatalities among the recently deceased

      • i.e. testing/detection rate among the dead

    • Number of tests with:

      • How the cases were flagged for testing (e.g. high risk area, high risk group, presenting with symptoms at point of care, thermal screening, random sampling etc.)

    *What test(s) were used * Which testing facilities
    ## What could we do to help?* In lieu of official (and detailed) data from regional authorities, we could create a website or mobile app for self-reporting cases who've been tested, with: * test result * how they were flagged to be tested * symptoms * timescales * test type if known * test location * location before presenting for testing * home town* Once we have the data, then stronger anlysis of/modelling from the existing data can be made, with dashboards, confidence ratings for published infection rates & estimates of likely community spread

    ## Notes & Caveats* Data privacy whould be at the forefront, but the data is probably useful if we can track cases to Town level to help inform local behaviour.* Place of test is likely less important than place of likely contacts (catch or spread)* Self-reported data means a self-selecting sample - likely biassed towards negatvive tests, those with less serious symptoms & those who are more comfortable with technology * Self-reported may not know the specifics of the test performed* Can't verify data is accurate, so likely there will be noise in the data from: * data entry mistakes * duplicate reporting when reporting on behalf of a patient * mis/disinformation/abuse

  • Will commented
    31 Mar 05:13pm

    Sorry - I did enter a description, but it got lost when submitting from mobile - rewriting now.