Nearly one-third (32%) of people already say they would be comfortable with artificial intelligence leading a primary care appointment, according to a new survey from Outbreaks Near Me – although fewer (25%) would be comfortable with AI-led therapy
Twelve percent of adults in the U.S. say their healthcare providers currently use AI to help diagnose, treat or communicate with patients, but 66% expect AI to play a bigger role in healthcare in the next five years, the survey found.
Are these results surprising or expected? What do people expect from AI in healthcare? Healthcare IT News spoke with Autumn Gertz, project manager at the computational epidemiology lab at Boston Children’s Hospital and project manager of the Outbreaks Near Me team, to get answers and a deep dive into the survey results.
Outbreaks Near Me is a crowdsourced syndromic surveillance tool that supports longitudinal disease monitoring for a variety of illnesses including influenza and COVID-19. Outbreaks Near Me is the next generation of Flu Near You, a decade-running platform conducting syndromic surveillance for influenza and other respiratory diseases.
The data collected from Outbreaks Near Me complements traditional disease surveillance by capturing real-time data of illness that occur across the disease severity spectrum.
The computational epidemiology lab at Boston Children’s Hospital, led by Dr. John Brownstein, is the team behind Outbreaks Near Me. Founded in 2006, the computational epidemiology lab is an established global leader in using online informal sources for disease outbreak monitoring and real-time surveillance of emerging public health threats.
The computational epidemiology lab partnered with SurveyMonkey to administer this nationally representative, cross-sectional version of the Outbreaks Near Me survey. This survey, originally created to better understand the changing dynamics of the COVID-19 pandemic, also captures information on attitudes and practices around health- and disease-related topics, including artificial intelligence use in healthcare.
Q. Based on your research, nearly a third (32%) of people already say they would be comfortable with AI leading a primary care appointment, though fewer (25%) would be comfortable with AI-led therapy. Most people are not comfortable with AI in these cases. In your opinion, what will it take to get more people comfortable with AI if AI is to make progress in healthcare?
A. While the majority of respondents indicated they are not comfortable with AI leading their care, we still find it remarkably interesting that some are. One quarter are comfortable with AI-led therapy and one third are comfortable with AI-led primary care.
The reality of AI-led healthcare is still very novel, and having a subset, albeit a minority, of the population comfortable with that idea is not necessarily an expected result.
There are a few things that will lead to increased comfort and acceptance of AI in healthcare specifically. The first is trained and targeted AI for healthcare emerging on the market. Surveying people on hypothetical AI compared to human practitioners they are used to likely introduces bias.
The second is AI in general increasing its presence and utility within daily life. As AI becomes a part of all facts of life, people will likely be more comfortable with its inclusion in healthcare. Lastly, as individuals learn that people in their lives, whether it’s friends, family or their healthcare providers, are adopting AI for healthcare uses, they will likely gain comfort in trying it themselves.
Years ago, it would have been hard to believe you could connect with your doctor on your phone. Now patient portals, chat features and mobile versions of health records are fairly common practice. We could very likely be on the precipice of a similar level of change, but until the specific medical AI applications emerge on the market, we have to gauge comfort based on assumptions made by our survey respondents.
Q. Only 14% of healthcare professionals say they are currently using AI to help diagnose, treat or communicate with patients; the vast majority (83%) are not using AI for any of these tasks. In your opinion, what will it take to get providers to embrace AI if AI is to become mainstream in patient care?
A. A lot of the factors that will lead to the general population embracing AI will also lead to its increased acceptance among healthcare professionals. This combined with evolving medical practice guidelines will likely facilitate the integration of AI into medicine. Every new technology that has emerged, from robotic-led procedures to telemedicine, has been more accepted over time.
Health systems and facilities overall adopting AI, rather than uptake being dependent on individual providers, is a pathway to making AI use in patient care more mainstream. If systems and networks take the time and resources to research and invest in promising healthcare AI, the practitioners within that system are more likely to trail or adopt the AI as well.
Champions for AI within certain networks of medical professionals may also create a sense of confidence and intrigue, increasing their buy-in.
Similar to the general population, acceptance of AI in healthcare will come with the right AI products entering the market. Our results also found the majority of medical professionals think AI will play a larger role in healthcare in the next five years. There is a lot of potential in this space, but also a lot of hesitancy.
AI that fills existing gaps, improves patient care and makes providers jobs easier is more likely to be embraced than that that does not provide as much obvious utility. ChatGPT, for example, became popular very fast not only due to its ease of use but also the clear value it added to daily and professional life.
Q. 23% of healthcare professionals think the use of AI in healthcare will help more than it hurts, 33% think it will hurt more than it helps, and 42% think it will equally help and hurt. One way or another, that's a lot of hurt. And that goes directly against the Hippocratic Oath. In your opinion, what are caregivers' biggest fears and how does the industry overcome them?
A. Concerns and fears around the unknown, and lack of control, make us all hesitant. For healthcare providers, they also have to consider patient safety, quality of care and their oath as they factor in using new technology like AI.
Human practitioners are subject to bias and error in their work, and AI is no different. However, properly employed AI is a tool to assist medical practitioners, not a means to “replace them”; emphasis on this is important to contribute to provider trust and comfort in using AI.
Generally, healthcare providers are accustomed to changes in protocol coming from tested and trialed methods, and AI should be no different. Studies on patient outcomes and satisfaction while employing AI would likely help providers overcome concerns around using AI in their practice.
Integration of AI into the practice of healthcare could be sustainably and comfortably achieved by soliciting provider feedback within trial phases of its use.
Viewing AI not as a limit of their control over patient care, but an opportunity to expand access and equity within healthcare, could be beneficial. At the end of the day, providers want what is best for their patients.
Technology like this may allow more patients to get connected with care or types of care they previously were unable or unwilling to seek. AI in medicine is brimming with potential to address some of the accessibility and equity issues that exist within healthcare.
Q. Most people in the general population say they still prefer important healthcare-related tasks to be done by medical professionals rather than by AI. Does this ensure AI will only play a small supporting role in the years to come?
A. Not at all. Let’s ask these folks the same question in five years and see if they still think so. The idea of AI being used in their individual care is novel to most people, while human-provided care is the norm.
However, many of the individuals uncomfortable with AI for care may already use technology to supplement their healthcare. If they are web-searching symptoms, self-diagnosing online, or even using reviews to decide which provider to see, these are not solely “human-led” approaches to healthcare.
Our research also found that younger individuals are more comfortable with the idea of AI-led healthcare, especially in a primary care setting. If this trend continues, in years to come a larger makeup of the patient population will be ages that are comfortable with AI-led or supported healthcare.
While I am making some assumptions that attitudes will remain similar to our current results, this does seem to suggest younger individuals may drive more use of AI in healthcare, rather than small amounts, in the years to come.
Provider adoption will likely lead to patient adoption as well. If your healthcare provider asks or informs you that AI is being used during your visit, this may change how you feel about AI. Seeing AI in action would likely help warm folks up to its use.
Whether a patient gets a diagnosis quicker, has a scan effectively read, or is able to delineate whether to go to the ER or urgent care, real-life positive or neutral experiences will increase the role AI plays in patient care.