CHICAGO — Deploying artificial intelligence in the dermatology practice will help bring more humanity back and lead to more engaged and satisfied patients, Justin Ko, MD, MBA, director and chief of medical dermatology at Stanford Health Care, said here at the American Society for Dermatologic Surgery annual meeting.
“This concept of thinking about job replacement when the topic of artificial intelligence comes up is really prevalent and is perpetuated by folks in Silicon Valley that have a funky world view that all of our problems can be fixed by technology,” Ko said. “Media reports are often reckless in their reporting of artificial intelligence with the man vs. machine narrative.”
Broadly, AI is a field that develops computer programs to perform tasks that are usually performed by humans.
“It’s not anything new. The recent confluence of a few factors have brought about the golden age of AI, like increases in computing power, new technology in machine learning and the advent of big data make it a powerful force,” he said.
In its present form, AI must be trained to perform specific tasks, based on a set of rules, Ko said.
“You can do this with any question, with any set of data,” Ko added. “Such as, ‘which patients on my schedule are most likely to miss an appointment or cancel?’ You can figure that out with all of the historic data you have. You can train a set of data to previous patients who have been late or missed an appointment. Let the AI figure out the characteristics that predict it and run it on future patients to see how it does.”
AI is powerful for two reasons: 1) it can uncover complex relationships that human computing power and the brain cannot; 2) it runs fast and does not fatigue. The system with run as fast the millionth time as it did the first, he said.
Augmented intelligence is a new conceptualization to enhance human intelligence not replace it.
“It works under the founding principle that we are the central actors working on behalf of our patients,” he said.
Ko highlighted multiple examples where augmented intelligence could work as a virtual clinical assistant:
- automatically translate a Mohs map into clinical documentation and billing;
- automatically compile visit summary during a patient consultation and filling out insurance paperwork;
- a system that alerts a doctor if a mole is new or that others have been stable since the last visit;
- flag patients that are at risk for post-operative infection or complications; or
- a system to engage patients after a procedure to monitor healing and progress in real-time that notifies the team if there was a problem.
“By deploying AI technology, can we actually help bring more humanity back to our practices?” he asked. “If we can do all of these things — all within the realms of possibility — would patients be more engaged and satisfied? Would we deliver a higher value of care?”
The quality of the data are crucial, Ko said. In addition to where it is from, how it is labelled is important.
Data gathering in an appropriate, ethical way is another concern.
“Algorithms have to be developed using high quality data that is representative and inclusive of populations upon which it is going to be used,” Ko said. “There are inherent social constructs that must be considered. Patients without access to care, those that would benefit from AI the most, are less likely to be in a dataset because they don’t have access in the first place. We must be thoughtful to ensure we are building bridges, no exacerbating these gaps.” – by Abigail Sutton
Ko J, et al. Technology disrupters in dermatologic surgery. Presented at: American Society for Dermatologic Surgery Annual Meeting; Oct. 24-27, 2019; Chicago.
Disclosures: Ko reports he is on a task force that deals with artificial intelligence.