Determinants of AI Adoption in Healthcare: Insights From a Unified Theory of Acceptance and Use of Technology (UTAUT) Study Among Doctors and Nurses in a Tertiary Care Hospital in North India

Background Artificial Intelligence tools are increasingly entering clinical practice, yet their adoption depends on how healthcare professionals perceive, experience, and intend to use them. Understanding factors associated with adoption is critical for designing targeted AI education. Objectives To assess acceptance of AI-based tools among doctors and nurses in a tertiary hospital in North India using a tool based on the Unified Theory of Acceptance and Use of Technology (UTAUT), and to identify early and late adopters based on prior AI exposure, training, and behavioral intention. Methods A cross-sectional mixed-methods study was conducted among 256 healthcare professionals (116 doctors, 140 nurses) at a multispecialty tertiary-care hospital. Prior to the administration of the questionnaire, an informative video introducing key concepts and applications of artificial intelligence in healthcare was shared with all participants to ensure a common baseline understanding. The structured UTAUT-based questionnaire captured demographics, AI exposure and training, perceived usefulness, accuracy, helpfulness, workload impact, behavioral intention, and barriers/enablers, along with open-ended questions. Quantitative analysis compared perceptions between those with and without prior AI experience, stratified by profession, and explored early adopters based on pre-specified predictor items. In an additional exploratory step, composite early-adopter definitions were constructed using AI exposure, formal training, tool use, and consistently high intention scores. Results Only about one-fifth of doctors (21.6%) and nurses (19.3%) reported prior AI experience in healthcare, and most of those with experience had used AI-enabled tools such as clinical decision support or drug-interaction checkers. Doctors reported the highest likelihood of using AI for drug-drug interactions, drug dosing and patient follow-up, whereas nurses rated AI as most useful for monitoring, documentation and nutrition support. Among doctors, prior AI experience was associated with a significantly higher likelihood of using AI for diagnosis, follow-up and triage, and with greater perceived helpfulness in triage. Across all participants, prior AI experience increased intention for daily clinical activities and clinical decision-making, although not for workflow tasks. When early adopters were defined broadly (any exposure, training or high intention), over 85% met criteria; however, under a strict definition requiring training, exposure, tool-use and uniformly high intention, only 3.5% qualified as AI champions, predominantly nurses. Trust, medico-legal concerns and limited awareness were the most frequently reported barriers by both; however, nurses more frequently expressed fear of job displacement. Training, institutional support and integration with electronic medical records were identified as key enablers. Conclusions Doctors and nurses showed different patterns of AI readiness. Doctors with prior AI exposure showed greater willingness to use AI, reflecting higher trust in AI for high- risk clinical decision support tasks. Nurses demonstrated high baseline acceptance, with less variation by prior AI use; however, formal training clearly distinguished early adopters. In conclusion, targeted, profession-specific training would serve as a key enabler to support effective and responsible AI adoption in healthcare.