Advances in AI and machine learning could lead to better health care: lawyers

AI has potential to make health care delivery more efficient, but is not without risk
Advances in AI and machine learning could lead to better health care: lawyers

As the commercialization of artificial intelligence and machine learning grows, the next frontier for its use will be in the world of health care. This use will potentially change health outcomes for the better – and brings questions surrounding privacy and data use to the forefront.

“We’ve been seeing the crossover between information technology companies and health-care-related companies applying AI and machine learning to solve health-care-related problems,” says Chad Bayne, partner with Osler Hoskin & Harcourt LLP in its emerging and high-growth companies practice. These crossover matters can range from medical devices to therapeutics to patient management and triage.

Bayne notes that Canada “punches above its weight” in AI and machine learning science in centres of excellence, including Toronto, Montreal, Kitchener-Waterloo, and the University of Alberta. These, in turn, have helped create companies focused on the intersection of information technology and AI, whether it is to improve drug discovery, develop a screening system that can predict which patients will need treatment, or find out what to do to avoid the need for treatment at all.

Artificial intelligence combines computer science with data to enable machines to mimic human intelligence when solving problems. Machine learning is a sub-field of artificial intelligence in which algorithms help processes or devices achieve learning through experience and new data. These algorithms can be “locked,” so their function does not change, or “adaptive,” meaning their behaviour can change over time.

AI often conjures images of independent, “thinking” robots. However, its actual uses are far more practical, whether it is handling routine and repetitive administrative tasks, such as triaging resources, or used for more complex purposes like detecting a potential medical condition in a patient’s electronic medical record.

“Properly harnessing AI technology will have huge ramifications for improving our health care system, making it more efficient,” says Danielle Miller Olofsson, a senior associate with the corporate law group at Stikeman Elliott LLP’s Montreal office.

“We as a society are starting to realize that as we’re aging, the health care system may become overburdened,” she says. “So we need solutions that are as much about preventing disease as managing it when it happens.”

There are many examples of Canadian hospitals and health care organizations using AI to improve efficiencies and outcomes for better patient care in a clinical setting. These include:

  • looking at administrative applications, such as using AI to manage hospital resources better (one example would be aligning schedules to match patient traffic)
  • clinical applications for improving patient care and outcomes
  • detection and diagnosis, to improve their speed and accuracy 
  • research and development in areas like drug discovery and medical devices

While the potential for using AI in health care is vast, Miller Olofsson says it is a “delicate balance” of private- and public-sector players coming together for the greater good. She points out that “these are extremely challenging issues that are also very political,” depending on the regulatory landscape.

For example, on one end, there is the European Union, which Miller Olafsson describes as having a very “human-centred” approach to AI and its commercial use. On the other end, jurisdictions like China embrace the potential of this technology at the expense of the impact on privacy and human rights.

North America, including Canada, has taken more of a wait-and-see approach to using AI and its challenges in potentially sensitive sectors like health care.

One consideration that Canadians might want to think about, she adds, is that, given we have a public, universal health care system, is there an argument that we are more permissive in sharing data for AI purposes? 

“Is integration into a larger network breaking down our traditional perception of what is private and what is public?” 

Of course, transparency and privacy concerns are significant, she notes, but if the information from our public health care system benefits everyone, is it inefficient to ask for consent for every use?

On the other hand, cybersecurity is another essential consideration, as “we’ve come to learn that there are a lot of malevolent actors out there,” says Miller Olafsson, with the potential ability to hack into centralized systems as part of a ransomware attack or other threat.

Even in its more basic uses, the potential of AI and machine learning is enormous. But the tricky part of using it in the health care sector is the need to have access to incredible amounts of data while at the same time understanding the sensitive nature of the data collected.

“For artificial intelligence to be used in systems, procedures, or devices, you need access to data, and getting that data, particularly personal health information, is very challenging,” says Carole Piovesan, managing partner at INQ Law in Toronto.

She points to the developing legal frameworks in Europe and North America for artificial intelligence and privacy legislation more generally. Lawyers working with start-up companies or health care organizations to build AI systems must help them stay within the parameters of existing laws, says Piovesan, and provide guidance on best practices for whatever may come down the line and help them deal with the potential risks.

Risk can take many forms, including the “human risk factor” behind these AI systems and whether they have the right talent trained to use them. “And then there is the governance structure needed to ensure that the system operates as intended. So I think lawyers have a huge role to play in the process of using AI wisely.”

Piovesan’s partner at INQ, Mary Jane Dykeman, agrees. “There must be confidence in the data itself. Is it clean? Is it usable? Is the data biased? And one assumes the data is legally obtained and meets privacy obligations.” 

Bayne at Osler points out that often the organization holding data grants access based on its non-commercial use. “But as soon as it crosses into the realm of potentially being used commercially, that could open up the need to go back and get informed consent.”

Bayne says that, as a lawyer who works with start-ups and venture capital funds, he focuses his questions on what the data is, how much there is, and whether it is the correct data. He also wants to make sure that there is a clear understanding of how the health care system works in the intended market.

He also asks clients how much any proposed innovation will improve the standard of care. “Will it increase life expectancy significantly? Will it improve the quality of life? Will it make a patient more comfortable?”

For organizations that hold the data needed for AI in health care settings – hospitals, long-term care homes, and primary care facilities – Dykeman says they also need to understand the potential benefits of AI and the challenges that can arise in using it.

In fact, in collaboration with Canada Health Infoway, INQ has developed the Toolkit for Implementers of Artificial Intelligence in Health Care to assist health care organizations across Canada that want to explore and develop potential uses. This toolkit explains critical concepts and issues related to the use of AI in health care, along with operational suggestions to help deploy AI.

It also guides AI governance to maximize the benefits of AI while minimizing foreseeable risks. The toolkit has six modules that provide checklists to help organizations plan their AI governance activities more effectively, tips on best practices, and case studies demonstrating real-world Canadian examples of successful AI solutions.

Through the six modules, the toolkit explains the fundamentals of AI and highlights unique benefits for the health care sector. It also provides an overview of the industry’s critical risks when deploying AI (algorithmic bias, challenges to privacy and security), Canada’s regulatory landscape, and a roadmap for investment opportunities in AI. The toolkit also looks at responsible AI governance and risk-assessment components.

Says INQ’s Piovesan, “There are a number of health care facilities very interested in adopting the use of artificial intelligence to improve efficiency and patient outcomes. In Canada’s health care sector, where we’ve got universal health care, we have an amazing opportunity to looking at the technology portfolio and start meaningful investments.”

However, those investments should also include building trust and good governance of AI “because when you put all this together, we’re more likely to have a successful outcome than if you think of this as just a technology issue and invest in technology alone.”

She adds that “if you choose good use cases and make the right investments, it’s all worth it. Because what we’re finding is return on investment is amazing when it comes to the use of AI and health care.” 


Clinical applications

  • improving patient experience and health care outcomes
  • Using AI in select procedures to improve patient care

Detection and Diagnosis 

  • Improve speed and accuracy of disease detection and diagnostics

Research and Development

  • Drug discovery
  • Medical Devices
  • System Efficiency