Global Study Reveals How Patients View Medical AI

How physicians feel about artificial intelligence (AI) in medicine has been studied many times. But what do patients think? A team led by researchers at the Technical University of Munich (TUM) has investigated this for the first time in a large study spanning six continents. The central finding: the worse people rate their own health, the more likely they are to reject the use of AI. The study is intended to help align future medical AI applications more closely with patients’ needs.

Acceptance by patients is essential for the effective use of artificial intelligence in medicine, regardless of whether AI is used as a diagnostic tool, to create individual treatment plans or for other applications. The international research network of the COMFORT study therefore surveyed around 14,000 patients in 74 clinics in 43 countries. In order to cover a wide range of diseases, the survey was conducted in radiology departments that carry out X-ray, CT and MRI examinations on behalf of other specialist disciplines.

A majority - 57.6% - viewed the use of AI in medicine positively. However, some differences emerged within the cohort: men were slightly more favorable at 59.1% than women at 55.6%. Approval also rose markedly with greater familiarity with technology and higher self-rated understanding of AI. Among respondents who described themselves as highly knowledgeable about AI, 83.3% were positive about using it in medicine.

The more severe a person’s own illness, the more negative their attitude toward AI. More than half of patients in very poor health rated medical AI "extremely negative" or "rather negative" (26.6% and 29.2%, respectively). Among those in very good health, the corresponding figures were 1.3% and 5.3%.

"The exact reasons for negative attitudes among seriously ill patients cannot be determined from our study," says Dr. Felix Busch, assistant physician at TUM’s Institute of Diagnostic and Interventional Radiology and the study’s first author. "We suspect that experiences with the health care system, illness burden, and psychological factors play a role."

Respondents expressed clear preferences for how AI should be used and designed. For 70.2%, it was important that medical AI be ‘explainable,’ allowing users to see the steps leading to its conclusions. In addition, 72.9% wanted these technologies to function as tools, with physicians making the final decisions.

Only 4.4% favored diagnoses made exclusively by AI. At the same time, just 6.6% wanted diagnoses to be made entirely without AI. These questions referred to hypothetical scenarios in which human clinicians and AI were equally accurate. "The results show that explainability must be considered from the outset," says Felix Busch.

One methodological limitation is the survey timing in 2023. "Since then, large language models in particular have advanced considerably. Patient attitudes may have changed," says Adjunct Teaching Professor Dr. Keno Bressem, senior lecturer and, together with Adjunct Teaching Professor Dr. Lisa Adams, senior author of the study. "Follow-up surveys are needed to test this and to align the development of medical AI with patients’ needs," adds Dr. Lisa Adams. A follow-up study by the COMFORT consortium using the same questionnaire is already underway.

Busch F, Hoffmann L, Xu L, Zhang LJ, Hu B, García-Juárez I, Toapanta-Yanchapaxi LN, Gorelik N, Gorelik V, Rodriguez-Granillo GA, Ferrarotti C, Cuong NN, Thi CAP, Tuncel M, Kaya G, Solis-Barquero SM, Mendez Avila MC, Ivanova NG, Kitamura FC, Hayama KYI, Puntunet Bates ML, Torres PI, Ortiz-Prado E, Izquierdo-Condoy JS, Schwarz GM, Hofstaetter JG, Hide M, Takeda K, Peric B, Pilko G, Thulesius HO, Lindow T, Kolawole IK, Olatoke SA, Grzybowski A, Corlateanu A, Iaconi OS, Li T, Domitrz I, Kepczynska K, Mihalcin M, Fašaneková L, Zatonski T, Fulek K, Molnár A, Maihoub S, da Silva Gama ZA, Saba L, Sountoulides P, Makowski MR, Aerts HJWL, Adams LC, Bressem KK; COMFORT consortium; Navarro ÁA, Águas C, Aineseder M, Alomar M, Al Sliman R, Anand G, Angkurawaranon S, Aoki S, Arkoh S, Ashraf G, Astri Y, Bakhshi S, Bayramov NY, Billis A, Bitencourt AGV, Bolejko A, Bollas Becerra AJ, Bwambale J, Capela A, Cau R, Chacon-Acevedo KR, Chaunzwa TL, Chojniak R, Clements W, Cuocolo R, Dahlblom V, Sousa KM, Villarrubia JE, Desai VB, Dhakal AK, Dignum V, Andrade RGF, Ferraioli G, Ganguly S, Garg H, Savevska CG, Radovikj MG, Gkartzoni A, Gorospe L, Griffin I, Hadamitzky M, Ndahiro MH, Hering A, Hochhegger B, Huseynova MR, Ishida F, Jha N, Jiang L, Kader R, Kavnoudias H, Klein C, Kolostoumpis G, Koshy A, Kruger NA, Löser A, Lucijanic M, Mantziari D, Margue G, McFadden S, Miyake M, Morakote W, Ngabonziza I, Nguyen TT, Niehues SM, Nortje M, Palaian S, Pentara NV, de Almeida RPP, Poma G, Purwoko M, Pyrgidis N, Rafailidis V, Rainey C, Ribeiro JC, Agudelo NR, Sado K, Saidman JM, Saturno-Hernandez PJ, Suryadevara V, Schulz GB, Soric E, Soto-Pérez-Olivares J, Stanzione A, Struck JP, Takaoka H, Tanioka S, Huyen TTM, Truhn D, van Dijk EHC, van Wijngaarden P, Wang YC, Weidlich M, Zhang S.
Multinational Attitudes Toward AI in Health Care and Diagnostics Among Hospital Patients.
JAMA Netw Open. 2025 Jun 2;8(6):e2514452. doi: 10.1001/jamanetworkopen.2025.14452

Most Popular Now

AI also Assesses Dutch Mammograms Better…

AI is detecting tumors more often and earlier in the Dutch breast cancer screening program. Those tumors can then be treated at an earlier stage. This has been demonstrated by...

AI could Help Emergency Rooms Predict Ad…

Artificial intelligence (AI) can help emergency department (ED) teams better anticipate which patients will need hospital admission, hours earlier than is currently possible, according to a multi-hospital study by the...

RSNA AI Challenge Models can Independent…

Algorithms submitted for an AI Challenge hosted by the Radiological Society of North America (RSNA) have shown excellent performance for detecting breast cancers on mammography images, increasing screening sensitivity while...

Head-to-Head Against AI, Pharmacy Studen…

Students pursuing a Doctor of Pharmacy degree routinely take - and pass - rigorous exams to prove competency in several areas. Can ChatGPT accurately answer the same questions? A new...

NHS Active 10 Walking Tracker Users are …

Users of the NHS Active 10 app, designed to encourage people to become more active, immediately increased their amount of brisk and non-brisk walking upon using the app, according to...

Brain Imaging may Identify Patients Like…

By understanding differences in how people’s brains are wired, clinicians may be able to predict who would benefit from a self-guided anxiety care app, according to a new analysis from...

Unlocking the 10 Year Health Plan

The government's plan for the NHS is a huge document. Jane Stephenson, chief executive of SPARK TSL, argues the key to unlocking its digital ambitions is to consider what it...

AI can Find Cancer Pathologists Miss

Men assessed as healthy after a pathologist analyses their tissue sample may still have an early form of prostate cancer. Using AI, researchers at Uppsala University have been able to...

How AI could Speed the Development of RN…

Using artificial intelligence (AI), MIT researchers have come up with a new way to design nanoparticles that can more efficiently deliver RNA vaccines and other types of RNA therapies. After training...

AI, Full Automation could Expand Artific…

Automated insulin delivery (AID) systems such as the UVA Health-developed artificial pancreas could help more type 1 diabetes patients if the devices become fully automated, according to a new review...

MIT Researchers Use Generative AI to Des…

With help from artificial intelligence, MIT researchers have designed novel antibiotics that can combat two hard-to-treat infections: drug-resistant Neisseria gonorrhoeae and multi-drug-resistant Staphylococcus aureus (MRSA). Using generative AI algorithms, the research...

AI Hybrid Strategy Improves Mammogram In…

A hybrid reading strategy for screening mammography, developed by Dutch researchers and deployed retrospectively to more than 40,000 exams, reduced radiologist workload by 38% without changing recall or cancer detection...