Majority of patients would trust AI more than human physicians for diagnosis, survey finds

© ipopba - stock.adobe.com

Artificial intelligence (AI) apparently is gaining health care credibility at a rapid clip.

A survey published this summer found 64% of patients would trust a diagnosis made by AI over that of a human doctor. But trustworthiness dips as health care issues and procedures become more complicated.

The results were part of “Technology and the Future of Healthcare,” a poll of 1,027 people across four generations by Innerbody, a health product and service testing company based in Palo Alto, California.

Younger people are more comfortable integrating AI and technology into their health care, while older patients are more skeptical of robot docs. The percentages likely to trust a diagnosis made by AI over a human physician are:

• Generation Z, born 1997 to 2012, 82%
• Millennials, 1981 to 1996, 66%
• Generation X, 1965 to 1980, 62%
• Baby Boomers, 1946 to 1964, 57%

But accuracy of diagnoses was the top concern about using AI in health care, based on 53.5% of responses overall. Data privacy (50.3%) and technical limitations (42.6%) were other top concerns, along with job losses for health care professionals (38.5%).

For specific procedures, 60% of respondents said they would be comfortable using AI to analyze medical images.

“The survey participants' observations hold merit, as numerous studies have been published demonstrating the ability of Deep Learning technology to identify cancer in radiology images,” report author Heather Schmidt wrote. Machine Learning (ML) uses algorithms to examine data and draw inferences, while Deep Learning is complex ML based on the human brain neural network.

Schmidt noted previous studies about a Deep Learning tool used in dermatology and radiologists using AI to diagnose breast cancer. But “it’s also clear that many AI studies have design flaws,” so more research is needed.

Just 3% of people said they are uncomfortable with any AI involvement in medicine.

Overall, 65% of patients said they were comfortable with use of robotics in medicine. Patients were more accepting of using robotics in X-ray work (86%), CT scans (82%), MRI scans (77%), and skin exams (75%).

Surgery is another issue, with less patient comfort about robots completing cesarian deliveries (47%), heart bypass surgery (46%), and hip replacements (45%).

Imaging also is the best proposed use of nanotechnology, with 55% of respondents confident in using tiny tech in medical imaging and 52% confident in using it for diagnosis. Drug delivery earned support of 45% of respondents, while 7% were not at all comfortable using it in health care.

Nanotech in health care is relatively new, so lower trust could be due to lack of understanding or uncertainty about how much research has been done to test it, Schmidt said.

“Studies researching the use of nanotechnology in cancer-fighting immunotherapy are promising, but this science is relatively new,” Schmidt wrote. “The same can be said for wound healing and tissue engineering; both of these fields are complex and offer significant challenges to medical professionals, and nanotechnology shows potential as a solution. However, the technology has yet to be widely used, possibly leading the public to hesitate to accept its usage in these types of applications.”