Seite - 576 - in VULNERABLE - The Law, Policy and Ethics of COVID-19

VULNERABLE576 detailed other intervention measures, primarily related to reduction of vulnerability to acute infection and shielding of the high-risk popu- lations without complete lockdowns. For instance, according to the Food and Agriculture Organisation of the UN (FAO),17 more than one third of the population of Afghanistan is considered severely food insecure, a condition that, according to available literature, would make them more vulnerable to acute infec- tion than their better-nourished counterparts. Consequently, given the feedback loop between food insecurity, COVID-19, and intervention scenarios, an ethical and accountable modelling methodology would take a very different approach than the original Imperial College model. Rather than focusing on an age parameterization, which would show increased vulnerability to severe infection as a person ages, a model for younger, but more food insecure, countries would include a breakdown of the level of malnutrition and food insecurity. This example reveals that ethical modelling in the pandemic is not only a technical problem, but one that must take into account a con- textual understanding of the realities in each country, as well as the impact of computing and disseminating model results. Furthermore, the accuracy and diversity of the data are also important issues. The most vulnerable communities could be under- or over-represented in a dataset, a common error that could lead to a wrong interpretation and an inappropriate decision. While age is not a vulnerability that can be reversed, food inse- curity most certainly is, and so are certain treatable co-infections. Models that include these and other kinds of contextual parameters will provide realistic and attainable options for governments trying to mitigate the crisis. Responsible Governance of Data The responsible use of data is an important component of accountable AI in the COVID-19 pandemic. This is of particular rel- evance in situations of vulnerability, where individuals do not feel they are able to consent to breaches in their privacy and personal data use. As highlighted by the Office of the Privacy Commissioner of Canada, during a public health crisis, emergency legislation may 17. “Integrated Food Security Phase Classification: Afghanistan” (May 2020), online: The  Integrated  Food  Security  Phase  Classification <http://www.ipcinfo.org/ ipc-country-analysis/details-map/en/c/1152636/>.