Lawson study aims to strengthen algorithm to better track homelessness

The project aims to clear up some of the fog left by data on homelessness that primarily relies on Canada's health care system. Caley Bedore / Global News

Researchers in London, Ont., have launched a multi-year study that aims to better track and understand homelessness in Canada by developing more accurate sources of data.

Homelessness Counts is a project led by the Lawson Health Research Institute and will primarily work by strengthening a pre-existing algorithm through AI, machine learning and consultations with communities across the country.

The algorithm was developed through prior research that partnered with ICES, an independent, not-for-profit corporation primarily funded by Ontario that tracks data derived from health-care interactions in the province.

Story continues below advertisement

Richard Booth, a lead investigator in the new study, says that algorithm fuelled a prediction model to help determine who is experiencing homelessness.

“The major issue with this algorithm … is (the data) only is pulled from health care. You need to go through a health-care agency that is paid for by OHIP to show up in these records,” Booth said.

“We know that there’s housing data that’s not captured in the health-care records, we know that there’s shelter data, especially, that’s not captured in health care, we know that there’s things around money that we can’t get, but are probably predictive of whether someone will remain homeless or not.”

Booth added that the latter part of the study will build on the algorithm through AI and machine learning, but not before consulting with communities across the country, including those who are experiencing homelessness.

Cheryl Forchuk, another lead investigator, said consultations have already taken place in Brantford, Kingston, London, North Bay and Thunder Bay, but the study plans on expanding beyond Ontario to include the territories, eastern and western provinces, and Montreal.

Get the latest National news. Sent to your email, every day.

During initial consultations, researchers asked service providers for those experiencing homelessness what data sets they use to track people in their system.

Story continues below advertisement

Forchuk said the most commonly reported method revolved around internal spreadsheets created by individual agencies, which would contain valuable information not found in the health-care system.

These data sets created issues in capturing people who could be defined as experiencing homelessness and led to a lack of an easily accessible and centralized source.

“Some of the biggest problems (service providers) had … the fact that they don’t have the data, the data is not integrated, it doesn’t take into account cultural sensitivities,” Forchuk said.

“Essentially a lot of trust issues around data sharing are not always accommodated.”

Craig Cooper, London’s director of housing stability services, joined Wednesday’s launch of the study to discuss the city’s experience with data and tracking homelessness.

Cooper said London previously collected data through an internal spreadsheet that gathered information from local service providers. This process led to challenges with data quality and duplication.

“You really could never get that total number of individuals seeking service in the community because people were moving from shelter to shelter, service provider to service provider, on a daily or even weekly basis,” Cooper said.

Story continues below advertisement

In late 2017, the city then signed on to the HIFIS database, which helped provide more clear and accurate information.

“What we’ve been able to do is have a consistent platform where all of our funded agencies are now inputting data. It allows to understand any duplicates, remove those duplicates and get a real better picture of who is accessing our system,” Cooper said.

HIFIS also allows for reporting on specific demographics and has led to past success in tracking London’s homeless veteran population, as well as the city’s homeless Indigenous population. Cooper noted that this work is done in co-ordination with outreach teams in the city.

“That co-ordinated access system has really allowed us to match individual to resources, it’s allowed us to prioritize individuals as a community, based on what we’re seeing as the significant needs in community,” Cooper said.

“We’re really using data a lot differently than we had even a couple of years ago and we’re seeing great outcomes.”

The project has four years of funding from the Public Health Agency of Canada.

Story continues below advertisement

Researchers aim to consult with at least 12 communities in Canada, but the hope is even more will open discussions during the study.

Booth added that the research aims to take an ethical approach to privacy, confidentiality and data gathering.

“We have to be really careful because we can do a lot of things, but should we do a lot of things?” Booth said.

“We want to make better services for individuals, but we can’t make better services for individuals until we know who they are, where they are and what they’re up to.”

Click to play video: 'Coronavirus: Ontario study finds people on homeless spectrum hit hard by COVID-19'
Coronavirus: Ontario study finds people on homeless spectrum hit hard by COVID-19

Sponsored content