Win, Lose, or Draw—Study Finds AI Diagnosis With Same Accuracy As A Human Physician
Will machines beat mankind in evaluating the illness and disease states that afflict the human body? A new research effort says the jury is still out.
Artificial intelligence (AI), or deep learning, is much-hyped, and for good reason. Technology and AI are swiftly changing and challenging the way we live. Algorithms are used to hire employees, find suitable mates, remind us to take medication, and help people invest money wisely. Why can’t AI figure out what is wrong with you faster than your physician? AI also holds the long hoped for cure for medical mistakes—a day when machine learning eliminates or greatly reduces the number of serious medical errors made by humans that hurt other humans.
A study recently published in The Lancet Digital Health took a look at currently available research to make an assessment.
The Future is Not Quite Here
After undertaking a large meta-analysis of scientific literature relating to the diagnostic capability of AI, researchers from the University Hospitals Birmingham NHS Foundation Trust in the United Kingdom report, right now, humans and machines are about equal in their diagnostic capability.
Study lead Dr. Alastair Denniston notes, “We reviewed over 20,500 articles, but less than 1% of these were sufficiently robust in their design and reporting that independent reviewers had high confidence in their claims.” Overall, scientists believe there is not enough focused research to declare that the future of diagnosis is advanced algorithmic assessment.
Adds Dr. Denniston, “Within those handful of high-quality studies, we found that deep learning could indeed detect diseases ranging from cancers to eye diseases as accurately as health professionals. But it’s important to note that AI did not substantially out-perform human diagnosis.”
Machines can be trained and software written to digest enormous quantities of data to recognize disease or symptom patterns. Imaging studies can be sifted and screened to provide a more granular read than a human radiologist. The US Food and Drug Administration (FDA) has already approved 30 algorithms based on AI for healthcare.
No machine needs to beat a human at the practice of medicine. The hope is that AI can enhance and improve patient treatment—while reducing delayed or missed diagnosis, and a host of other medical errors. AI incorporated into the clinical setting may prove a potent aid to physicians in improving survival outcomes and reducing medical harm.
The future is not quite here, but it is on its way.
Speak with Experienced Attorneys
When the medical care you receive is negligent or below the expected standard of care, the legal team at Schochor, Federico and Staton, P.A. can help. Call us at 410-234-1000 to schedule a free, confidential consultation at our offices in Baltimore, Maryland, or in Washington, D.C.