How AI Is Revolutionizing Medicine
There isn’t a field AI won’t touch, but one exciting trend is AI for medicine. It turns out — medical professionals have similar problems as data scientists, lots of tedious tasks, not enough time to get to higher-order ones. AI could change that. Medicine’s need for diligence and detail benefits significantly from using machines as support for human labor and expertise. You could be seeing increased use of AI in medicine as soon as your next doctor’s visit. Let’s take a look at how it might show up.
Diagnosing and Managing Disease
Doctors are overworked and strapped for time. Doctors visits have decreased in time, and specialists are increasingly required to spend a lot more time dealing with day to day details than they are exploring more extensive treatment plans or finding solutions.
Machines trained on medical data can help ease some of that load. Researchers at Google recently trained a data set specifically for diagnosing one of the most common causes of adult vision loss, diabetic retinopathy. It had a sensitivity score of 87–90%.
In another advancement, researchers from China developed a system to diagnose colon health, improving polyp detection using a combination of AI and expert diagnosis. Machines were better able to detect tiny polyps, something challenging for specialists to catch, especially when strapped for time.
Another possibility could be aiding doctors in another time-consuming activity, disease maintenance. For example, Mabu, a small robot designed to interact with a patient in his or her home, has shown promise in limited trials. Patients answer questions and prompts from the robot based on data training from medical best practices. If a patient is short of breath, she can ask targeted questions designed to assess the level of severity and make recommendations for when to call a human doctor.
Enhancing Radiology
Nearly a third of Radiologists will be sued for malpractice during their careers, a number caused by the sheer number of films radiologists process during a single day. That kind of consistent attention to detail is derailed by a number of external factors, including fatigue and distraction.
Researchers training algorithms to detect anomalies in films have achieved impressive accuracy ratings, consistently higher than radiologists themselves. Detecting micro deposits of cancer can be challenging for any human to see in a film, but a machine can detect deviations with no trouble.
So will AI replace radiologists? Not in the short term. Standard sensitivities suggest that the expert final opinion of a human specialist is still needed to corroborate a machine’s findings and then recommend a treatment path in consultation with the patient and other doctors. Machines can’t quite handle that level of interaction yet, but they could help facilitate one on one time with patients.
Gene Sequencing
We already know that the answers to some of our most persistent health questions lie in our genes and the genes of organizations that infect us. Machines have the power to process large amounts of data quickly and efficiently, drastically improving our understanding of complex gene sequencing.
Increased computing power allows us to study larger and large populations and sample sizes for minute deviations that could unlock answers to diagnoses or cures. NVidia, for example, works with Parabricks to deliver the massive computing power necessary to analyze a whole genome, a requirement in clinical research. It also has programs to examine molecular “docking” or how a potential treatment will bind to a target protein.
Machines also make it easier for us to explore cures based on formulations of compounds and simulating high-level trials between compounds faster than any human in a lab could do. What previously took months to find out doesn’t work now takes weeks, maybe even less. Machines can also run simultaneous experiments, computing in a way that humans can’t.
The result? Humans get answers faster and have more time for higher-order tasks. Researchers are better equipped to filter out failures, reducing cost, and time-intensive labor.
Augmented Intelligence in Medicine
The American Medical Association is a wholehearted adopter of Augmented Intelligence, a reimagining of AI that places machines in complementary roles to humans, not as replacements. This combination of machine intelligence and human expertise could create a new wave of medical care that’s both more accurate and more caring. Where medicine has been lacking in the “care” department, machines could bring out that side again.
Read more data science articles on OpenDataScience.com, including tutorials and guides from beginner to advanced levels! Subscribe to our weekly newsletter here and receive the latest news every Thursday.
