Are physicians ready for precision antibiotic prescribing? A qualitative analysis of the acceptance of artificial intelligence-enabled clinical decision support systems in India and Singapore

• Main Authors: Huang, Zhilian, George, Mithun Mohan, Tan, Yi-Roe, Natarajan, Karthiga, Devasagayam, Emily, Tay, Evonne, Manesh, Abi, Varghese, George M., Abraham, Ooriapadickal Cherian, Zachariah, Anand, Yap, Peiling, Lall, Dorothy, Chow, Angela
• Other Authors: Lee Kong Chian School of Medicine (LKCMedicine)
• Format: Article
• Language: English
• Published: 2024
• Subjects: Medicine, Health and Life Sciences; Antimicrobial resistance; Artificial intelligence
• Online Access: https://hdl.handle.net/10356/174050
• Institution: Nanyang Technological University

Objectives: Artificial intelligence (AI)-driven clinical decision support systems (CDSSs) can augment antibiotic decision-making capabilities, but physicians’ hesitancy in adopting them may undermine their utility. We conducted a cross-country comparison of physician perceptions on the barriers and facilitators in accepting an AI-enabled CDSS for antibiotic prescribing. Methods: We conducted in-depth interviews with physicians from the National Centre for Infectious Dis eases (NCID), Singapore, and Christian Medical College Vellore (CMCV), India, between April and December 2022. Our semi-structured in-depth interview guides were anchored on Venkatesh’s UTAUT model. We used clinical vignettes to illustrate the application of AI in clinical decision support for antibiotic prescribing and explore medico-legal concerns. Results: Most NCID physicians felt that an AI-enabled CDSS could facilitate antibiotic prescribing, while most CMCV physicians were sceptical about the tool’s utility. The hesitancy in adopting an AI-enabled CDSS stems from concerns about the lack of validated and successful examples, fear of losing autonomy and clinical skills, difficulty of use, and impediment in work efficiency. Physicians from both sites felt that a user-friendly interface, integration with workflow, transparency of output, a guiding medico-legal framework, and training and technical support would improve the uptake of an AI-enabled CDSS. Conclusion: In conclusion, the acceptance of AI-enabled CDSSs depends on the physician’s confidence with the tool’s recommendations, perceived ease of use, familiarity with AI, the organisation’s digital culture and support, and the presence of medico-legal governance of AI. Progressive implementation and continuous feedback are essential to allay scepticism around the utility of AI-enabled CDSSs.