Research review: Patients have understandably mixed feelings about AI in healthcare

A new review of the relevant scientific literature suggests patients beholding AI-aided healthcare feel as though they’re standing before a complex and shifting landscape. Many are aware that emerging benefits exist side-by-side with persistent uncertainties.

The researcher behind the observation, Osnat Bashkin, PhD, of Ashkelon Academic College in Israel, reviewed 38 studies from 13 countries. She found more than 75% of patients recognize worthwhile benefits in healthcare AI. Yet 50% to 70%—surely including a broad swath overlapping with the 75% cohort—entertain profound concerns. 

The concerned subgroup worries about losing human connection with clinicians, having to trust AI for reliability and safety, and risking personal privacy and data security. 

Bashkin’s paper, published June 3 in the International Journal of Medical Informatics, offers several insights for provider organizations hoping to better understand patient perceptions of AI in healthcare. Here are five. 

1. Patient acceptance of AI encompasses a complex interplay of individual characteristics.

These include technology design features, healthcare contextual factors, relational elements and systemic considerations, Bashkin reports. She remarks:  

‘Successful implementation strategies should maintain human oversight, ensure transparent communication about AI capabilities and limitations, tailor approaches to specific contexts, and engage patients as stakeholders in development and implementation.’

2. Patients view AI as a transformative technology.

Many believe it can improve diagnostic accuracy and streamline healthcare processes, Bashkin found.

‘However, they often struggle to grasp its complexities, particularly regarding AI applications in personal health.’

3. Patients are becoming more open to the use of AI in healthcare. 

However, findings depend on the group surveyed and the AI tool examined, Bashkin points out. More:  

‘Despite the challenges associated with the varying levels of understanding of AI, its novelty leads patients to desire greater individual and institutional control over AI.’

4. The perceptions and reactions shaping patient experiences while interacting with AI for health purposes are incompletely understood. 

Understanding these perceptions is “crucial as AI continues to revolutionize healthcare,” the researcher writes.  

‘For AI to fulfill its potential, it must not only be adopted by physicians and health providers but also by patients and other members of the public.’

5. Patient acceptance of AI is not merely—or even mostly—determined by the technical performance of AI systems.

It’s also shaped by “a dynamic interaction between individual perceptions, healthcare relationships, organizational contexts and broader societal factors,” Bashkin concludes. “In healthcare settings, factors such as trust, privacy concerns and the preservation of human connections play critical roles.” 

‘An integrated approach accounting for all these factors enables a more nuanced understanding of the multidimensional nature of patient acceptance of AI in healthcare.’

The paper is posted in full for free. 

• In other research: 
   
  • Acute GI Bleeds in ICU: AI algorithm predicts transfusion need, mortality risk
     
  • Vanderbilt: AI researchers develop new technologies for cancer care
     
  • MIT: Scientists argue for more FDA oversight of healthcare AI tools
     
  • Northwestern: New AI transforms radiology with speed, accuracy never seen before  
     
  • Institute of Science Tokyo: AI model diagnoses cancer via laptop computer 
     
  • York University: Specific learning strategies can enhance AI model effectiveness in hospitals
     
  • Florida Atlantic University: AI remotely detects dizziness and balance disorders 
     
• Regulatory:
   
  • FDA approves pediatric indication for SONU wearable device to treat nasal congestion
     
• Funding: 
   
  • Sprinter Health raises $55M To expand access to preventive care and close critical care gaps