Intelligent curation: Improving access to quality health information online

“The biggest problem with technology for older adults, however useful it may be to them, is that it’s not designed with them in mind,” says Dr. Julie Robillard. “If there are too many hurdles to adopting assistive technologies, older adults just won’t and they may miss out on important benefits.”

In general, while older adults do tend to adopt novel technologies at a slower rate than other user groups, they are keen to engage with digital resources; as many as four in ten Americans over the age of 65 are smartphone users, for example, and according to Pew Research data, 67 per cent use the Internet.

In Canada, 564,000 people are living with dementia but the number is expected to grow to nearly one million by 2030 as the population ages. Alzheimer’s disease is the most common form of dementia but there is a lot of uncertainty about what causes the disease and how to protect oneself from it. Previous research has shown that about 80 per cent of people, and half of older adults, turn to the Internet for health information.

Dr. Robillard’s earlier work found that a majority of physicians report having seen patients who have sought specific health information online, so it stands to reason that older adult web users are searching online for information related to their own health and aging. For those with neurodegenerative diseases such as Alzheimer’s and dementia, existing algorithms for web searches are problematic as results are not organized in a way that matches the needs and identities of end-users. 

“There are a number of assistive technologies available to people with dementia, but often those tools don’t align with user values and emotions,” says Dr. Robillard. “Technology needs to be adaptive, and artificial intelligence can help us design web browsing tools that are more responsive to users’ subtle cues. As a result, older adults will be more likely to use online health information to make decisions that will lead to positive outcomes.”

Dr. Robillard and her team, including collaborator Dr. Jesse Hoey from the University of Waterloo, presented an artificial intelligence-powered version of the QUEST at the 2017 Association for the Advancement of Artificial Intelligence Joint Workshop on Health Intelligence. QUEST is a seven-item checklist developed in the Robillard lab to evaluate online health information and to assist users in finding accurate, unbiased information about Alzheimer's disease and dementia online. Using machine learning and a computational model of identity, Robillard and colleagues are automating the QUEST tool and making it responsive to users' needs and emotions.

“In speaking with older adult web users, one of the challenges frequently reported is difficulty in assessing the quality of information they’re finding,” says Dr. Robillard. “It can be hard sometimes to distinguish between educational and advertorial content online, especially for users with possible cognitive impairment.”

Because online health information is not regulated, the harms of misrepresenting commercially driven content as scientific or clinically relevant can be plentiful, with consequences ranging from negative health decision-making to vulnerability to fraud and predatory marketing. Dr. Robillard’s solution, an ongoing effort to improve the way we assess the validity of information presented online, aims to arm web users with clear criteria for what is and isn’t good content.

To Dr. Robillard, “good content” is simple:

  • Does the website cite its sources?
  • Are the sources current?
  • Is there an author associated with the content whose credentials you can verify?
  • Does the site eschew strong, misleading language like “cure,” “guarantee,” or “miracle”?

Earlier research from Dr. Robillard’s lab found that high-quality online information focused on risk factors and presented balanced, verifiable information while lower quality websites emphasized nutrition as prevention and were more likely to present conflicts of interest, such as selling unproven items for the prevention or treatment of dementia.

In short, if it sounds too good to be true … it’s probably not true. With QUEST, researchers hope artificial intelligence will take the guesswork out of the equation.

“As our population ages and unregulated online health information proliferates, it’s urgent that we find ways to assist older adults with dementia in avoiding misleading or harmful information that could deter existing treatment plans or cause personal harm,” says Dr. Robillard. “It is our hope that we can effectively harness the utility of artificial intelligence in curating content based on user needs, even as those needs change and evolve over the course of a disease.”

For more information on QUEST, email Dr. Robillard at julie.robillard@ubc.ca