Blackmores Group is starting to use knowledge graph technology to ground AI tools that could help healthcare professionals recommend supplements to patients.
Head of the company’s year-old Ai Health Innovation Hub, Warren Mackay-Smith, told the Google Cloud Summit Sydney that knowledge graph technology is well suited to the health and life science space, because the technology is able to handle a number of different data points and permutations that may be encountered in these settings.
“When you think about the amount of products that we have across our portfolio, when you think about the amount of ingredients that are within those products, when you think about the number of conditions that those ingredients could treat, when you think about the number of interactions an ingredient could have with someone’s existing condition, whether they’re taking aspirin or not, and think about how it interacts with the body, there are multiple permutations that are difficult to understand and difficult to represent,” Mackay-Smith said.
Knowledge graphs are useful for mapping complex relationships between data points.
Queries of the graphs are easily explainable, and this can also make graphs a good foundation to run AI atop of.
That appears to be the plan at Blackmores Group.
The company already provides “interactions” guidance – primarily how vitamin, mineral or herbal remedies interact with prescription medicines – and other clinical support tools through a ‘Blackmores Institute’.
Through the Ai Health Innovation Hub, agentic AI technology as an information delivery mechanism is on the radar.
Mackay-Smith suggested that the knowledge graph technology could provide the data foundational layer for AI tooling.
“In healthcare, the challenge is finding results that exactly match the [patient’s] conditions and understand the context of what is required,” he said.
“That’s why we moved to a graph approach. It gives us the scale across the breadth of ingredients, conditions, way things work within the body, interactions. It understands the relationships between those, and that’s really key for us because then we’re able to build reliable solutions on top. It makes it explainable.
“You can actually ask the AI, “Why have you recommended X?” And it can traverse the graph and understand why it’s come up with the solution that it’s done. And that means every recommendation is traceable.
“For our industry, that’s extremely important. If we don’t get a recommendation right, we can lose a doctor’s trust, we could compromise a consumer’s health, and we could have non-compliance [with] a regulatory body.
“These are things that we need to tackle head-on if we want to build agentic solutions [and] … drive the agentic future for our organisation and hopefully for our industry.”
Ry Crozier attended Google Cloud Summit Sydney as a guest of Google Cloud.
