English
French
Spanish
The COVID pandemic has accelerated social challenges across Canada. As an example, opioid overdose deaths in Ontario were up by 35-40% during the pandemic (source: https://www.ctvnews.ca/health/pandemic-aggravates-opioid-crisis-as-overdoses-rise-and-services-fall-out-of-reach-1.5189677) and that’s not unique to the province, nor is this unique to one issue – similar reports are emerging regarding domestic violence, mental health, addictions, homelessness, income and food insecurity. As a result, Canada’s social safety has been strained significantly under these pressures.
Despite $280B+ annual investments in over 250,000 different services aimed at tackling these challenges, the lack of proactive data-driven decision making has meant that responses during COVID remained fragmented and uncoordinated thanks to a patchwork of siloed funding streams, jurisdictional accountability pinball, and limited systems-level line of sight to impact. The long-term impacts of the pandemic are yet to be reaped – we have a choice to make on whether we will leverage this crisis as an opportunity to reset the social safety net, or whether we will continue on this track.
One of the silver linings to the pandemic has been the acceptance that digitization is not just here to stay, but that it must be a priority in our approaches to social issues. We have seen how technology innovation can support transformation across sectors, and the social impact space is no exception. This doesn’t discount the critical need for applying and developing these tools with a Reconciliation and equity lens. Quite the opposite: the challenge to the tech sector is to innovate while advancing the Truth and Reconciliation Commission (TRC) Calls to Action – an equity lens to data architecture, data analysis and visualization, machine learning and predictive analytics are practical means of advancing commitments to human rights.
While there has been an unprecedented shift in trying to confront COVID through R&D and technology innovation focused on treatment or vaccination, investment in COVID social impacts from this angle has been less common. Some noteworthy exceptions have come through government support with a focus on building new digital means for connecting people to mental health supports or developing enhanced data monitoring tools on social challenges. Of course, the challenge here is that these efforts remain in their infancy, and have not been scaled in time for maximum impact during the pandemic. We’ve also missed opportunities to get ahead of incoming challenges because of the lack of available data and inconsistency of its reporting and transparency. And of course, we’ve missed the important connection across social issues, assuming that ministerial responsibilities or donor preferences equate to actual human experiences which are always complex, dynamic and layered.
For our part, as a social technology innovation B-Corp, we saw an opportunity to bring Artificial Intelligence (AI) / Machine Learning (ML), to help decision-makers think ahead from a systems planning and systems integration lens. With investment from Canada’s Digital Technology Supercluster, we were able to develop AI/ML algorithms that could predict three targets of homelessness, suicide and domestic violence on a community by community basis. With these algorithms, we can model different societal and economic effects that are correlated to a municipality’s social needs to understand how changes in our society affect the delivery of the social safety net. The algorithms developed for this project are now part of HelpSeeker’s Community Success Hub – click here to find out more.
A core issue with our current analysis of social needs is an overreliance on population-based formulas. These methods examine simple causality and do not take into account the complexity of our social needs. This risks entrenching inequity depending on a community’s resources – we need to bring a more sophisticated and localized lens to decision making. To advance a more nuanced and community-based approach, we focused on individual municipalities and were able to develop a core set of features that were predictive of future homelessness, domestic violence and suicide (Homelessness was measured through Point in Time (PiT counts), domestic violence was measured through compilation of police reports created by Discourse magazine and suicide numbers were collected from Statistics Canada). For example, this work showed that for homelessness, the features with the greatest effect on predicting future rates were:
- Youth population
- Number of professional, scientific and technical workers
- Municipality
- Migrants to BC
- Education-related searches
Yet, even this is an oversimplification of how we developed a more accurate prediction. For all the features used by the algorithm, they can have a positive or negative effect on the predicted homeless rate. For example, high amounts of certain professions reduce the predicted homelessness rate, while others increase it. Meanwhile, the amount of an individual feature (e.g. population) has different effects. For example, an increase in population of 10% would not have 10x the influence of a 1% increase in population. The complex relationship we describe here is only possible through advanced technology such as AI/ML.
This work also highlighted the need to focus on local needs rather than broadly generalizable techniques. There is a disparity between different communities, and they have different social needs depending on the local environment. For example, there was a link between the rate of certain types of crime and the number of suicides in a community. Yet, similar to suicide the crime rate and driving factors is not necessarily a simple correlation and instead requires a thoughtful approach to local needs to decrease the incidence of suicide.
What’s the gist then?
Firstly, we can’t view social needs in siloes as they are complex and multifactorial. We need to think across departments, government levels, and immediate pain points and create an integrated social policy and funding response framework; as per the homelessness example above, addressing future challenges means homelessness policy should be developed in an integrated manner with education and youth support. While we know this is much easier said than done, this is the essential heavy-lifting our decision-makers should champion across departments, ministries, government, and sectors. That also means we need to agree on KPI, reporting and data standards, and how we will use technology to reset the social safety net to maximize the ROI to Canadians.
The social sector recognizes that tech can enhance cost efficiency and service delivery which will support Canadians’ overall wellbeing and help resolve long-standing social challenges, but this means the funders (government and philanthropy) have to think holistically and proactively.
