Clinicians equipped with machine learning can, in theory, apply what works for one patient to the care of another—and another, and another—and so on.
Ideally, the subsequent patients end up receiving care that is both personalized to them and driven by ever-bigger big data. Meanwhile administrators and payers get care protocols with less variation and thus less cost.
Stated another way: “The radical potential of AI is that health systems no longer need to choose between personalization and scale.”
That’s the thinking of Benjamin Fels, who has a vested interest in seeing the technology from the perspective of an AI enthusiast: He’s CEO of the Seattle-based healthcare AI company Macro-Eyes.
And yet Fels has written a balanced and thoughtful piece on healthcare’s need to make better use of data if it is to get the biggest return on its inevitable AI investment.
Financial Times published the piece May 18.
“What could go wrong? A great deal,” Fels writes. “Bias in medical machine learning is deadly. The most accessible data to train models for healthcare do not reflect the global burden of disease. Likewise, in clinical trials, participants do not accurately reflect the diversity of patients.”
Fels adds that bias in machine learning tends to throw cold water on AI’s potential whenever algorithms are trained on insufficiently broad data.
“The machine’s reality is what you show it,” he points out. “Realign the training data and the algorithm learns to correct former tendencies.”
The best AI, he suggests, uses massive datasets to continually remind provider organizations that their patients are individuals, not groups or even subgroups.
The technology’s ability to pull this off is counterintuitive, Fels acknowledges. However, if properly deployed, healthcare AI “could reinforce the humanity of medicine by emphasizing what a provider notices about a patient,” he writes, “making the interaction between patient and provider ever more central.”
Financial Times has posted the piece in full for free.