The promise of artificial intelligence and machine learning for healthcare delivery is bright – but significant challenges around bias, equity, and information development and delivery still need to be addressed head-on for AI’s promise to become reality.
Speaking Monday at HIMSS21 Digital, Dr. John Halamka, president of Mayo Clinic Platform, shared his insights about AI’s ability to improve health system efficiency, patient outcomes and physician workflows.
Additionally, he addressed the growing concern of AI bias and discussed the safeguards necessary to guarantee equity in AI-supported healthcare.
Halamka sees an environment in which AI provides large, diverse datasets of past patient information that physicians can leverage to augment their own personal experience and knowledge when treating patients.
Such a system could, for example, give an emergency physician important facts that prompt the physician to look beyond an initial diagnosis (erratic behavior due to the presence of cannabis) for an issue he or she hadn’t considered (the patient has meningitis).
“AI augmentation of human decision-making is going to be greatly beneficial,” said Halamka.
However, one key issue that must be solved first is ensuring equity and combating bias that can be “baked in” to AI. “The AI algorithms are only as good as the underlying data,” said Halamka. “And yet, we don’t publish statistics describing how these algorithms are developed.”
The solution, he said, is greater transparency – spelling out and sharing via technology the ethnicity, race, gender, education, income and other details that go into an algorithm.
“A perfect storm for innovation usually takes place when there is urgency to do it, policy that suggests we should do it and industry saying it’s the right thing to do. And that’s exactly what is happening right now. It’s a top of mind issue,” he said.
Halamka points to what he calls the four grand challenges to AI adoption in healthcare:
- Gathering valuable novel data – such as GPS information from phones and other devices people carry as well as wearable technology – and incorporating it into algorithms.
- Creating discovery at an institutional level so that everyone – including those without AI experience – feels empowered and engaged in algorithm development.
- Validating an algorithm to ensure, across organizations and geographies, that it’s fit for purpose as well as labeled appropriately as a product and for being described in academic literature.
- Workflow and delivery – getting information and advice to physicians instantly while they’re in front of patients.
The good news: Starting this year, Halamka predicts increasing public-private collaborations to tackle concerns about AI bias, equity and fairness.
“Let us hope government, academia and industry work on those four challenges – and we’ll all be in a better place,” he said.