Difference between revisions of "Project:Data anthropology and ethics"

From Wikibase Personal data
Jump to navigation Jump to search
(No difference)

Revision as of 12:30, 9 April 2020

Thema:data anthropology and ethics
Project leader: Pidoux

Next meetings

April 09, 2020 at 6pm for reviewing the questionnaire.

Challenge

The full initial challenge that can be found here: https://jessicapidoux.info/?p=201

Roadmap

Goals
A non-profit project based on a sociological study, from questionnaires and tested with computational models
A methodology for adapting to local models
Social responses for communities
Open Source: models and methodology
Different from a digital contact tracing proposal
Digital contact tracing can be discussed with Paul-Olivier Dehaye also part of our team
Obtaining public funding from public institutions in different countries (Switzerland, Canada, Netherlands
Hiring full time people
The members of this team are "consultant experts" to implement the project. We provide the questionnaire, scientific article, consulting and follow up


Possible official sponsors

personaldata.io
MyData
dhcenter UNIL-EPFL

Documentation



Minutes 2020/04/08

  • Use the methodology The Role of Age Distribution and Family Structure on COVID-19 Dynamics: A Preliminary Modeling Assessment for Hubei and Lombardy .

Paper + data --> The Role of Age Distribution and Family Structure on COVID-19 Dynamics: A Preliminary Modeling Assessment for Hubei and Lombardy (Q4393) data are avaialble --> https://github.com/bwilder0/COVID19-Demography file:World_Age_2019.csv
Korean data will be used to make extraprolation and valide the model.
Social contours: make assumption , then mapping on how the population looks like. .
https://forum.personaldata.io/t/questionnaire-social-contours/398
Homophily: tendency of individuals to associate and bond with similar other

https://github.com/RPetitpierre/COVID19-Demography

Versus Virus challenge No. 46

date: 3 april -5 April 2020

In versus virus it is number 46 and was merged with other challenges with other purposes, that is why is important that you read our initial challenge in the linke above.

website: www.CO19andME.com


Ideas to work on:
1. NLP tool according to Paul-Olivier post on slack
participants:
2. Data analysis of questionnaires results
participants:
3. Modelling lifestyles from personas and questionnaires
participants: Jessica
4. Identifying factors of risk according to lifestyles and virus characteristics
participants:

HackVid 28.03

1. Started by public health surveillance, and their ethics and laws. Rules of privacy already allow the state to limit the freedom. Enhanced with digital systems, like asking for digital surveillance. What are limits, etc.

2. Starts from the bottom up, difficult to conceptualize. Not part of standard public health surveillance. If imagine new ecosystem of contributors, not to enhance surveillance. Should they need a framework for themselves beyond the laws. What's is the driver?
what's are the limits?
Should we need a framework?

  • Start with anthropology

whats' the model
resources
how many models?
identify how people are living? How exposed they're?
social points of contagion
what's the implication for ethics?

700 answers from the French version of Questionnaire COVID-19 lifestyles and contagion

Very interesting input specially the supermarket visit.

Personas: https://docs.google.com/document/d/1uQS9AOr8bMr3fJwqmfPVLobvW9JioV_qAGAzPT2FUvE/edit?usp=sharing
11 types of at risk profiles were detected

  • kid + single parents
  • teenager + student
  • single person without children with office job (homeworking)
  • doctor in hospital (borderer worker)
  • nurse in mobility
  • delivery person
  • cashier or seller in a supermarket + risky partner
  • independent worker
  • unemployed person healed from COVID-19
  • isolated elder person
  • family with home-school


  • From the data perspective

summary of the discussion:
1) personas can be helpful as elements in a methodology to analyze the spread of a disease and the risk factors
2) data collaboratives can be used to collect the data needed to build and refine such models
3) at the same time, these data and models generate risks of stigmatization
4) personas can also be helpful as elements in a methodology to analyze the spread of stigmatization risks
5) disease risks and stigmatizations risks are structurally related, as groups at high risk of contagion face higher stigmatization risk, irrespective of other (positive/negative) characteristics
6) additionally, other characteristics (e.g. negatively perceived social traits, e.g. poverty, social deviance) combine with at risk behaviour to generate higher stigmatization risks

Outputs:
1. Categorization of the personas for modeling the drivers of risks of contagion and being stigmatized.
2. Three levels of analysis to the modeling:

  • risks of contagion according to the virus and COVID-19 characteristics
  • lifestyles factors that would increase the risks of contagion
  • ethical issues : stigmatization

3. Use data for public health purpose only


Which data are relevant from the personas for design)
Possible sources of data: potentially portable COVID testing.
An ultrasensitive, rapid, and portable coronavirus SARS-CoV-2 sequence detection method based on CRISPR-Cas12
https://www.biorxiv.org/content/10.1101/2020.02.29.971127v1

[feature 1: degree of risk for themselves, likelihood to get infected] [how to know?]
-> result of a model with a lot of variables

[feature 2: Based on the configuration of the shelter, what are the degree of risk of contaminating others, likelihood to infect others] [how to know?]
variables: shelter configuration, occupation,


[feature 3: typical contagion channels] [how to know?]
variables: transportation, shared facilities, occupation, leisure, food supply, hospital, pharmacy,


[feature 4: easy to detect/contact/trace] [how to know?]
variables: level of education, gender, age, access to tech, awareness, location (urban/suburban/remote), level of trust in government & health system,

[feature 5: risk of harming by detection/contact/tracing ] [how to know?]
variables two families of stigmatization risks: A) socially undesired behaviour. e.g., illegal activities, -> have incentive to avoid surveillance -> risk is not intrinsically related to the generation of useful knowledge from an epidemiological point of view -> surveillance can try to mitigate the problem by getting as much information as possible about the disease-related aspects of behaviour, ignoring or masking the non-disease-related aspects of the behavior

B) all groups combining a high degree of risk of contagion with a high risk of contaminating others, based on demographic, lifestyle, occupational characteristics -> stigmatization risk is intrinsic to (and highly correlated with) the generation of useful knowledge from an epidemiological point of view -> surveillance can try to mitigate the problem by a) intervening on the granularity of the description of risk-related phenomena (may be in trade-off with data utility), b) adopting a language that does not revolve around concrete sociological categories keywords: stigmatization, discrimination, group exclusion.

As example, 2 related controversies that have already happened in France
 - stigmatization of people living in poor suburb of Paris associating hospital saturation with assuming not respecting the quarantine (press/social media comments)
- stigmatization of health workers (neighbour leaving note on nurses' flat door asking to go live elsewhere)

twitter exchange https://twitter.com/MariannesNoires/status/1243476777147731968
about this article
https://www.lemonde.fr/planete/article/2020/03/26/il-n-y-a-plus-une-seule-place-de-reanimation-dans-le-93_6034502_3244.html

[feature 6: communication risks, e.g. miscommunication, fear, stigmatization] [how to assess[h] this?]
variables: age, level of education,


Challenge for other hackathons

1. identify the relationships that explain and/or predict specific personas becoming drivers and spreaders of a disease
2. identify the relationships that explain and/or predict specific personas becoming stigmatized as drivers and spreaders of a disease, and the risk factors that would lead such stigmatization having heavy consequences on their well-being
3. build an early warning system for new social categories at risk of being identified as super spreaders and suffer the highest social consequences of this association
4-How to model the factors in lifestyle that would increase the risk of spreading/contagion, and identify the virus risks associated to this propagation

Retrieved from "https://wiki.personaldata.io/w/index.php?title=Project:Data_anthropology_and_ethics&oldid=85488"