Liberating health data in a digital world: new report details solutions to data access obstacles
Participants at the Schwartz Reisman + Diabetes Action Canada Solutions Workshop came from government, health organizations, hospitals, law firms, research outfits, the private sector, and more to discuss the prospects of improving patient outcomes through innovations in sharing and interpreting healthcare data. A new report details their findings.
Privacy legislation has been instrumental in protecting data rights and data privacy in an increasingly data-driven world. But in Ontario’s healthcare sector, restrictions on access to data are based on outdated privacy legislation that doesn’t account for the realities of a digital world—and the powerful technologies that can analyze data to significantly improve patient outcomes.
A new report from the Schwartz Reisman Institute, in partnership with Diabetes Action Canada, explores how and why existing privacy law is generating significant delays and obstacles, both for conventional researchers and those working in the emerging field of machine learning (ML) for health. Gathering key insights from a recent collaborative, cross-disciplinary workshop, the report makes recommendations about how health data could be liberated from the outdated mechanisms that ostensibly protect privacy, but actually act as obstacles to improved health outcomes for patients.
“The problem is not that we value privacy too much; it’s that we protect it in the wrong way,” says Gillian Hadfield. “There are easily achievable ways to simultaneously protect privacy and get the masses of data we have into the hands of data scientists and public health officials.”
Participants came from a variety of policy sectors, academic disciplines, non-profit organizations, and healthcare institutions to bring their diverse perspectives and backgrounds to a common goal: improving the health and autonomy of those who live with diabetes in Ontario through better access to and analysis of data stored in the health system.
The Solutions Workshop was led by Dan Ryan, professor at the University of Toronto’s Faculty of Information and specialist in design thinking, collaboration in communities of organizations, and innovation in the legal and educational sectors.
U of T iSchool Professor Dan Ryan facilitated the workshop using a variety of techniques, including creative thinking exercises, deep dives into particular problem sets, empathy elicitation sessions, and new solutions for old problems.
Participants discussed five problem sets in which the roots of the challenge lie:
Researchers find it difficult to access data due to complex and expensive data sharing and use negotiations with data custodians.
ML specialists face a lack of understanding about their need for large quantities of data—a quantity erroneously seen as unnecessary.
Research Ethics Boards (REBs) have varying capacities, interpretations of risk, and timetables. When researchers have to consult multiple REBs, lengthy delays result.
Modern, responsive health systems (“learning health systems”) require up-to-date information. But integrating data from different levels (patient, clinical setting, institution, population) is challenging.
Canadian institutions struggle to retain top ML researchers because obstacles to accessing data are bigger than in other jurisdictions. The high-performance computing systems needed for ML are scarce and not well-integrated with conventional systems.
Curious to know what experts at the workshop identified as barriers to data-driven health improvement—and how to dismantle them?
