A “hype cycle” devised by industry and innovation research firm Gartner places AI and deep learning at the “peak of inflated expectations” phase of the emerging technology life cycle. The next step in the cycle is the “trough of disillusionment”, which is a steep downturn in usage and adoption. Some innovative technologies never survive this downturn, while others emerge stronger than ever.
Artificial Intelligence is more nebulous than previous disrupting innovations like the iPhone. Algorithms designed to emulate human levels of thinking, problem solving, and memory are referred to as AI systems. These programs utilize machine learning and act more autonomously than a simple calculator or computer.
Deep learning is the most advanced subset of machine learning. It allows machines to be nearly independent in their decision making, once initially coded and given new sets of information. These more self-reliant machines make decisions in a “black box”, meaning that programmers and users may not know why or how a machine using AI came to its conclusion.
These systems have many real-world applications and can be a huge resource in the medical field. Disease detection, diagnosis, treatment, virtual medical assistants, and robot assisted surgery are all being explored with the help of AI using machines. ABI Research estimated that in 2021, AI applications could save the global healthcare industry $52 billion.
Of course, the issue with a black box machine making sophisticated medical decisions means that there is a possibility of mistakes and malpractice. Conversely, AI could also be used to reduce error and fraud, so the question remains: will it predominantly be a force for good or not?
Perhaps the flashiest use of AI is in disease care. Diseases like cancer have plagued humans for centuries, and even though concerted efforts like Richard Nixon’s 1971 National Cancer Act have made some ripples, a 100-percent effective cure still eludes oncologists and researchers. Machines using AI may make a huge difference in cancer care for years to come. In cancer care, time is everything. If machines can diagnose a patient sooner and with better accuracy than an oncologist, they can save lives or increase life expectancy.
Research teams from all around the world have developed algorithms to detect cancers before the human eye could, reduce misdiagnosis, and even develop treatment plans. Even rare cancers like mesothelioma have research teams developing AI systems specifically for diagnosis in all its intricacies. Dr. Jack Kriendler, founder of the Centre for Health and Human Performance, reflected optimistically on AI as a weapon against cancer in an interview with Forbes.
“I would sooner today trust computer scientists and data scientists to tell me how to treat a really complex system like cancer than my fellow oncologists,” said Kriendler. “I would not have said that two to three years ago.”
Besides promising applications for patient care, AI also has the potential to save time and money in other aspects of healthcare. Some of the most beneficial uses of AI come in at the administrative and assistant levels, saving time and money for the healthcare system.
Voice assistants can help with many administrative tasks that leach doctors’ time away from patients. Medical billing is incredibly complicated and frustrating, and using smart machines to check for errors and fraud could streamline this system. Other administrative uses include patient check-in and checkups, to reduce the time nurses spend on these tasks.
However, like any emerging technology, AI has its snags and drawbacks. First, treatment by machine may not be as advanced as Dr. Kriendler indicated. Even advanced and highly-publicized AI systems, like IBM’s Watson, have been slow to develop successful treatment plans, and the ones they do develop have been subpar, bordering on dangerous.
Another issue could be data bias. Medical reporting and data is not free from human error and influence, which means the data these machines are trained on is also not unbiased. Training machines on this data could have consequences, especially when it comes to minority medical care, which has historically faced problems of racism and discrimination due to poor reports.
Third is the exorbitant cost of some of these systems. Though AI is projected to save the industry money globally, the price tag for smaller hospitals may not be achievable. This could result in growth of treatment disparities between urban and rural populations. Additionally, using the simple law of supply and demand, healthcare providers and private practices will eventually be able to charge patients a higher premium for their own health if AI becomes the saving grace is could be.
These uses of AI seem poised to make waves in the healthcare industry. Though a Skynet situation is unlikely to emerge, there are still issues that must be ironed out before widespread implementation can be achieved.
The next two stages in Gartner’s hype cycle offer an insight into how AI will progress over the next five years. After the “trough of disillusionment” comes an upward slope and then the “plateau of productivity.” Keep an eye out for these advances in the next few years to see how this buzzword innovation will affect healthcare.
Emily Walsh is a health advocate and enthusiast. She spends her time researching healthcare innovation, drinking too much coffee, and spreading awareness to her readers. She welcomes questions, criticisms, and vegan eggnog recipes sent to her here.