Healthcare, Lay Expertise, & AI: Why We Shouldn’t Use One Without The Others
Fun fact: cockroaches enjoy burrowing within the tubing system of in-home nebulizer respiratory devices. A mother of an asthmatic child in a research study I managed in Baltimore, Maryland was aware of this — she spent years observing the habits of the infestations in her home.
This is not a common factor some would consider, however, and many doctors at her local hospital were not aware. So in, rightly, prescribing home nebulizer treatments to asthmatic children in Baltimore, many of which allergic to and living in homes with cockroaches, the physicians were essentially prescribing the delivery of irritants directly to the children’s airways. The example from my past research illustrates how lay groups can contribute useful knowledge unknown to the greater scientific community. This information exchanged is generally referred to as “lay expertise”, a concept described in Nature’s 2011 article “The Local Perspective” as knowledge,“acquired from experience, observation, and previous generations,” rather than structured (academic) scientific training.
There is a divisive debate within the health and scientific communities regarding the merit of lay expertise when it comes to patient care. It is the standard that a medically trained professional should be charged with determining serious health diagnoses. Naturally, an individual is consulted about their health history and symptoms when they go in for treatment; however, over the past decade even this level of lay expertise has been minimized due to the advent of interconenctive health databases between hospitals, such as Epic, and other artificial intelligence technologies becoming staples in modern healthcare.
Artificial intelligence, however, is a divisively debated topic as well. Innovations that can improve wellness and save lives are beneficial advancements for all of us. PathAI, an artificial intelligence startup for breast cancer detection, introduced preliminary technology in 2016 able to detect breast cancer with only a 7.5 percent error rate (only 4 percent higher than human error rates). In 2014, Researchers at Yale successfully implemented ‘artificial pancreas’ in humans controlled by iPhones. Even country wide epidemics, such as the flu, are now swiftly trackable and diagnosed using Google Flu’s artificial intelligence algorithms that analyze our search queries.
While these technological advances are making positive impacts on many lives, going back to the issue of lay expertise, should not personal knowledge now start to be given higher esteem in patient care as well? As artificial intelligence’s wide integration into the realm of healthcare continues to rise, there is a widening gap between self and personal health in medicine.
Personal experiences and circumstances not quantifiable by algorithms can play a vital part in our health profiles. While there are wearable technologies that can track our glucose levels and heart rates, and even baby diapers from the company Pixie Scientific that can observes a baby’s health by analyzing his/her poop, there are subjective elements unique to each of us that could impact our health outcomes.
Going back to the cockroach and nebulizer discovery above, a child could have on a wearable as she takes her treatment. The wearable could detect levels of allergens within her system at a given time and determine that her cockroach level was increasing as her medication level increased. It could then advise that the machine be replaced, as it must be malfunctioning and allowing environmental irritants into the device. The device is replaced, however, the pattern continues. This is not due to the artificial intelligence not performing its function — it monitored allergens and returned actionable data — but rather to a personal knowledge that cockroaches themselves burrow in nebulizer tubing. So replacing the device without remedying the socio-environmental problem known to the person leaves the algorithm’s purpose moot.
Of course, artificial intelligence continues to grow more ‘intelligent’ and there could be (and may already be) a device able to offer doctors recommendations on potential contributing factors to their findings; however, with the diversity of our personal situations, this may not be enough. A baby wearing a Pixie Scientific diaper could eat high levels of a culturally isolated food that misrepresents lines in stool samples. A flu predicted by Google Flu could be the result of simultaneous students writing papers during a school health week. A patient, still, can be misdiagnosed with cancer.
In Yuval Harari’s book on the rise of AI, A Brief History of Tomorrow, he asserts that the future of humankind will be dependent on, and controlled by, the artificial intelligence we have created. Acknowledging how dependent we are on our smartphones and how deeply we already live within algorithmically dictated spaces (social media), there are glimmers of possibility in this predicated future.
So if this is true, if we are progressing closer to an artificially dictated intelligence, we must start considering who is dictating the information it’s learning from.
Scientists, doctors, and engineers will be involved in the creation of healthcare artificial intelligence that will go on to ‘self-learn’ from its initial programming, which is why lay expertise is more necessary now than ever.
Whatever your health equivalent of a cockroach in a nebulizer may be, personally experienced and observed information must play a tandem role with artificial intelligence in navigating healthcare. We do not know what we do not know about our health, and then they are the things that we do know. We should be considering and learning from both. Striking a balance between lay experts, scientists, and technology can empower and secure the beneficial progress of health wellness for our futures.
