New AI Learns from Thousands of Medical Records to Advise Physicians
- By John K. Waters
Engineers at the USC Viterbi School of Engineering have developed a new artificial intelligence (AI) algorithm that taps into, and learns from, what amounts to the recorded experiences of thousands of doctors to suggest the best diagnostic strategies.
Led by Gerald Loeb, a professor of biomedical engineering, pharmacy, and neurology, the engineers went beyond Bayesian Inference, which evaluates a probability from a specific set of facts, to create an algorithm that learns from the collective electronic records of millions of patients to "think like a doctor" and suggest better and more cost-effective diagnosis and treatments.
"The algorithm works just like a doctor," Loeb said in a statement, "thinking about what to do next at each stage of the medical work-up. The difference is that it has the benefit of all the experiences in the collective healthcare records. This is a new form of artificial intelligence."
Loeb, who is a trained physician, was a pioneer in the field of neural prosthetics and one of the original developers of the cochlear implant. He believes the new AI-driven algorithm will lead to faster, better and more efficient diagnoses and treatments.
In the abstract of the report on the new model, published in the Journal of Biomedical Informatics, Loeb writes:
The currently available approaches of knowledge-based and pattern-based artificial intelligence are unsuited to the iterative and often subjective nature of clinician-patient interactions. Furthermore, current electronic health records generally have poor design and low quality for such data mining…. Because the software is tracking the clinician’s decision-making process, it can provide salient suggestions for both diagnoses and diagnostic tests in standard, coded formats that need only to be selected…. As the database of high-quality records grows, the scope, utility and acceptance of the system should also grow automatically, without requiring expert updating or correction.
By inverting Bayesian Inference, the algorithm allows identification of the diagnostic actions that are most likely to disambiguate a differential diagnosis at each point in a patient’s work-up.
By drawing on the diagnostic and testing experiences of thousands of physicians, this novel approach can suggest options the doctor using it might never have considered. But Loeb has said that the algorithm isn't meant to replace the judgment of individual physicians, just "complement and support them."
Speaking to Marc Ballon for the USC online newsletter, Loeb acknowledged the challenge of implementing the algorithm in current, balkanized American health care system. He said he believes promise of the new model lies with tech giants, such as Google, Amazon, and Microsoft, that have the resources to disrupt that system.
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at email@example.com.