Reality Check: Augmented Intelligence is the Future of Medtech
Product Creation Studio is pleased to bring you exclusive content by Heather Thompson, a senior editor for Medical Design & Outsourcing and former editor-in-chief of MD+DI magazine. With 15 years of experience covering medical technologies and FDA regulations, Heather specializes in delivering the latest trends and news in the medical device industry.
Augmented intelligence has the potential to disrupt the medical device world. And for some, augmented intelligence is a far more attractive option than its counterpart, artificial intelligence. The topic is one that has captured the attention of designers at Product Creation Studio. Chrissy Glaister, Lead Usability Engineer, participated on a panel facilitated by IDEO on this timely topic at MD&M West in Anaheim, CA. In the session, they explored some of the implications, responsibilities, and adoption challenges augmented intelligence could present.
This got the team at Product Creation Studio thinking about how the technology will fit into medtech’s future. I asked PCS to explore some questions that people in the medtech space might have about augmented intelligence. Here are some of the ideas they shared:
1. What is the difference between augmented intelligence vs. artificial intelligence?
Augmented intelligence positions itself as assistant to our own internal decision-making—not to make decisions for us, but to give us the critical information necessary to make the best decisions possible for a situation.
2. Why does medtech need augmented intelligence?
Caregivers are trying to solve complex problems, but are really only using generic tools to get there. Right now, there are many errors that occur because professionals cannot determine the root cause of an illness, which results in “treatment hunting” and “treatment stacking.” Humans are still required to be binders and puzzle masters. Augmented intelligence could lower the level of mastery and help doctors string together solutions faster and more accurately. If medical devices could raise the percentage of accuracy on any human assumption (e.g., right diagnosis, right therapy), we’d see vast improvement in the healthcare system.
3. What are the risks of augmented intelligence?
We have to ask how safe the intelligence is with humans (doctors). We may need keep-outs to ensure the intelligence is a tool and not something more. There are also ethical risks. Consider an ethical dilemma for an autonomous car scenario: A human jumps out in front of an autonomous car, and the AI must make decision with no safe outcome option. The car must either hit the pedestrian or swerve into a tree. How does that translate to a medical technology scenario and who is responsible?
Human-centered design is the likely response to such challenges. Designers will be tasked to draw a line on such questions and own the reasoning behind such decisions. The decision may not always be right, but we need people who are documenting the best-known answer at any time and professionals who can make empathic decisions at the cost of stakeholder desires.
4. How do we put augmented intelligence data in the context of emotional and empathetic design?
As systems grow in complexity, both quantitative and qualitative data have value. We need to marry the numeric and binary data with ideas of ranges and mapping. Mapping provides emotional context to those hard data numbers.
5. How will it influence the medical regulatory environment?
There is no doubt augmented intelligence will complicate a regulatory landscape. FDA is filled with smart people who want smart regulations and control (which is positive). Augmented intelligence systems only work if individuals trust the system output. FDA may not be able to trust this system immediately to do what it does best—solve difficult problems—because of the unknowns.
Augmented intelligence will also move too fast for FDA. Long development cycles might dissuade developers from using the technology. However, augmented intelligence could be used to improve validation processes as well as the final product.