Artificial intelligence (AI) is increasingly incorporated into healthcare through diagnostic support, risk prediction, image interpretation, clinical decision support, personalized treatment, and administrative automation. Its potential value lies in processing large volumes of heterogeneous clinical information and identifying patterns that can complement conventional medical reasoning. However, clinical performance alone does not determine whether an AI-enabled system will be successfully adopted. Patients remain central stakeholders because they provide health data, receive AI-informed recommendations, and ultimately decide whether technology-supported care is acceptable. Systematic reviews show that patients and the public often recognize the potential benefits of clinical AI while simultaneously expressing reservations about autonomy, reliability, accountability, privacy, and the continuing role of clinicians [1]. Patient familiarity with AI varies substantially. In a German hospital survey, general awareness was high, yet detailed knowledge was considerably lower, and older adults and individuals with lower educational or technical affinity expressed greater caution [2]. Nationally representative data from the United States similarly indicate that attitudes toward AI in diagnosis and treatment are heterogeneous and influenced by the type of clinical task and the degree of physician involvement [3]. Qualitative work has also demonstrated that patients are concerned about errors, reduced human contact, bias, and uncertainty regarding responsibility when AI recommendations are wrong [4]. These findings suggest that acceptance is not a simple reflection of technological enthusiasm; rather, it reflects patients’ understanding of AI, perceived benefits, perceived risks, and trust in the healthcare professionals who use it. Data governance is another major determinant of public confidence. Patients generally support the use of health information for AI research when governance, consent, confidentiality, and institutional accountability are clear, but willingness to share information declines when commercial use or uncertain data stewardship is involved [5]. Experimental research further shows that privacy concerns, trust, communication, transparency, liability, and perceived usefulness vary according to the clinical context in which AI is introduced [6]. More recent patient studies have reinforced the preference for physician oversight rather than autonomous AI decision-making [7,8]. Radiology research has similarly demonstrated that patients value efficiency and diagnostic support but continue to expect human supervision and clear responsibility for final decisions [9,10]. Internal medicine is a particularly relevant setting because patients often present with multimorbidity, chronic disease, repeated investigations, and complex therapeutic decisions. These circumstances create opportunities for AI-supported risk stratification, monitoring, and decision support, while also increasing the importance of trust and communication. Evidence describing patient attitudes in small-island Caribbean settings remains limited, creating a need for context-specific assessment before wider implementation. Therefore, the present study aimed to assess patients’ awareness and knowledge of AI, examine their perceptions of its benefits and concerns in internal medicine, determine their acceptance of AI-assisted healthcare, and identify demographic and knowledge-related factors independently associated with acceptance among patients attending internal medicine services in Dominica.

Study design and setting: A hospital-based cross-sectional observational study was conducted among adult patients attending internal medicine services at All Saints University School of Medicine and DCF Hospital, Dominica, from January 2025 to December 2025. The study was designed to characterize patient awareness, perceptions, concerns, and acceptance of artificial intelligence (AI) in clinical care at a single time point. The questionnaire domains were informed by previously published patient-centered surveys and reviews examining AI awareness, trust, physician oversight, data sharing, and acceptance in healthcare [1,2,5,9,11,12]. Study participants and sampling: Patients aged 18 years or older who attended internal medicine outpatient or inpatient services during the study period and were able to understand the survey questions were eligible. Patients who were critically ill, cognitively unable to provide reliable responses, unable to provide informed consent, or who returned substantially incomplete questionnaires were excluded. A total sample of 250 participants was included. Consecutive eligible patients were approached during routine clinical attendance until the target sample was achieved, reducing investigator selection of individual participants. Data collection instrument: Data were collected using a structured questionnaire comprising four sections: demographic and clinical characteristics; awareness and knowledge of AI; perceptions of benefits and concerns; and acceptance of AI-assisted care. Items addressed AI-assisted diagnosis, clinical decision support, treatment selection, chronic disease monitoring, privacy, erroneous recommendations, physician-patient interaction, and preference for clinician oversight. The wording was kept nontechnical, and participants were provided with a brief standardized explanation of medical AI before completing opinion items when clarification was required. Variables and scoring: Perception and acceptance statements were recorded using a Likert-type response format. For descriptive reporting, “agree” and “strongly agree” responses were combined. An overall acceptance score was derived from acceptance-related items; scores reaching at least 60% of the maximum attainable value were categorized as moderate-to-high acceptance, whereas lower scores were categorized as low acceptance. Age, sex, education, chronic disease status, previous awareness of AI, and basic understanding of medical AI were evaluated as potential correlates of acceptance. Statistical analysis: Data were entered and analyzed using standard statistical procedures. Continuous variables were summarized as mean and standard deviation, while categorical variables were expressed as frequencies and percentages. Associations between categorical participant characteristics and acceptance category were examined using the chi-square test. Variables of clinical relevance or demonstrating bivariate association were entered into a multivariable binary logistic regression model. Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) were reported, and a two-sided p value <0.05 was considered statistically significant. Ethical considerations: Participation was voluntary, and written informed consent was obtained before questionnaire administration. No personally identifying information was included in the analytical dataset, and responses were handled confidentially. The study was conducted in accordance with internationally accepted ethical principles for human-participant research. Necessary Permissions were obtained before starting the study.

A total of 250 patients attending internal medicine services in Dominica were included in the study. The mean age of the participants was 45.8 ± 16.7 years, and 138 (55.2%) were female. Most participants had secondary or higher education, while 108 (43.2%) reported at least one chronic medical condition. Hypertension was the most frequently reported chronic condition, followed by diabetes mellitus. The demographic and clinical characteristics are summarized in Table 1. Table 1. Demographic and clinical characteristics of the study participants (n = 250) Characteristic n (%) / Mean ± SD Age, years 45.8 ± 16.7 Age group, years 18–29 48 (19.2) 30–44 65 (26.0) 45–59 72 (28.8) ≥60 65 (26.0) Sex Male 112 (44.8) Female 138 (55.2) Educational level Primary education or less 38 (15.2) Secondary education 91 (36.4) Technical/vocational education 43 (17.2) University education 78 (31.2) Employment status Employed/self-employed 143 (57.2) Unemployed 38 (15.2) Retired 44 (17.6) Student/other 25 (10.0) At least one chronic medical condition 108 (43.2) Hypertension 72 (28.8) Diabetes mellitus 51 (20.4) Cardiovascular disease 24 (9.6) Chronic respiratory disease 19 (7.6) General awareness of AI was relatively high, with 171 (68.4%) participants reporting that they had previously heard of AI. However, only 122 (48.8%) reported a basic understanding of medical AI. Approximately half were aware of AI-assisted diagnosis, while 96 (38.4%) knew about AI-based clinical decision support. Internet and social media were the most frequently reported sources of AI-related information. Awareness and knowledge findings are presented in Table 2. Table 2. Awareness and knowledge of artificial intelligence in healthcare among participants (n = 250) Awareness/knowledge characteristic n (%) Previously heard of artificial intelligence 171 (68.4) Aware that AI can be used in healthcare 153 (61.2) Reported basic understanding of medical AI 122 (48.8) Aware of AI-assisted diagnosis 128 (51.2) Aware of AI use in medical imaging 114 (45.6) Aware of AI-based clinical decision support 96 (38.4) Previously encountered/used an AI-enabled healthcare application 61 (24.4) Primary source of information regarding AI Internet/social media 93 (37.2) Television/newspapers/news platforms 54 (21.6) Healthcare professionals 39 (15.6) Family/friends 25 (10.0) Academic/workplace exposure 18 (7.2) No previous source/exposure 21 (8.4) Participants generally expressed favorable perceptions toward the supportive use of AI in internal medicine. Most believed that AI could improve diagnostic accuracy, accelerate diagnosis, support treatment selection, and improve healthcare efficiency. Nevertheless, concerns regarding privacy, incorrect recommendations, and reduced physician–patient interaction were common. Overall, 211 (84.4%) participants believed that physicians should retain responsibility for final clinical decisions, and 196 (78.4%) preferred combined physician–AI decision-making. The principal perception and acceptance findings are shown in Table 3. Table 3. Patients’ perceptions and acceptance of artificial intelligence in internal medicine (n = 250) Statement Agree/Yes, n (%) Positive perceptions AI can improve diagnostic accuracy 197 (78.8) AI can facilitate faster diagnosis 186 (74.4) AI can assist physicians in treatment selection 181 (72.4) AI can improve personalized patient care 174 (69.6) AI may reduce avoidable medical errors 169 (67.6) AI may improve healthcare efficiency 191 (76.4) AI may improve access to specialist-level decision support 176 (70.4) Concerns regarding AI Concerned about privacy and security of medical information 180 (72.0) Concerned about incorrect AI recommendations 171 (68.4) Concerned that AI may reduce physician–patient interaction 154 (61.6) AI may make healthcare less personal 146 (58.4) Acceptance of AI-assisted care Comfortable with physician using AI to assist diagnosis 188 (75.2) Accept AI-assisted treatment recommendations reviewed by a physician 179 (71.6) Willing to undergo AI-supported risk assessment 182 (72.8) Willing to use AI for chronic disease monitoring 170 (68.0) Trust AI recommendations when explained by the treating physician 181 (72.4) Comfortable with AI diagnosis without physician review 71 (28.4) Willing to follow AI-generated treatment without physician confirmation 64 (25.6) Prefer combined physician and AI decision-making 196 (78.4) Physician should make the final decision when AI is used 211 (84.4) Overall moderate-to-high acceptance 164 (65.6) Low acceptance 86 (34.4) Overall acceptance of AI-assisted healthcare differed significantly by age, educational status, previous awareness of AI, and basic understanding of medical AI. Participants younger than 45 years demonstrated greater acceptance than those aged 45 years or older. Acceptance was also higher among participants with technical or university education and among those with previous awareness or understanding of AI. In multivariable analysis, basic understanding of medical AI, previous awareness, and higher educational attainment remained positive predictors of acceptance, whereas age ≥45 years was independently associated with lower acceptance. Sex and chronic medical condition status were not independent predictors. These findings are summarized in Table 4. Table 4. Factors associated with acceptance of AI-assisted healthcare and multivariable logistic regression analysis (n = 250) Characteristic/Predictor Moderate-to-high acceptance, n (%) Low acceptance, n (%) Adjusted OR (95% CI) Adjusted p-value Age Bivariate p=0.004 <45 years (n=113) 85 (75.2) 28 (24.8) Reference ≥45 years (n=137) 79 (57.7) 58 (42.3) 0.58 (0.34–0.98) 0.043 Sex Bivariate p=0.499 Male (n=112) 76 (67.9) 36 (32.1) Reference Female (n=138) 88 (63.8) 50 (36.2) 0.91 (0.54–1.53) 0.716 Education Bivariate p=0.005 Secondary or less (n=129) 74 (57.4) 55 (42.6) Reference Technical/university (n=121) 90 (74.4) 31 (25.6) 1.78 (1.01–3.15) 0.047 Previous awareness of AI Bivariate p=0.001 No (n=79) 40 (50.6) 39 (49.4) Reference Yes (n=171) 124 (72.5) 47 (27.5) 1.89 (1.05–3.40) 0.034 Basic understanding of medical AI Bivariate p<0.001 No (n=128) 69 (53.9) 59 (46.1) Reference Yes (n=122) 95 (77.9) 27 (22.1) 2.18 (1.22–3.90) 0.009 Chronic medical condition Bivariate p=0.301 Absent (n=142) 97 (68.3) 45 (31.7) Reference Present (n=108) 67 (62.0) 41 (38.0) 0.83 (0.49–1.42) 0.502 Note: Bivariate p-values were calculated using the chi-square test. Adjusted odds ratios were obtained from multivariable binary logistic regression. OR = odds ratio; CI = confidence interval; AI = artificial intelligence. In summary, 65.6% of participants demonstrated moderate-to-high acceptance of AI-assisted healthcare. Acceptance was considerably greater when AI functioned as a decision-support tool under physician supervision rather than as an autonomous substitute for clinicians. Although patients recognized potential benefits in diagnostic accuracy, efficiency, and personalized care, concerns about privacy, erroneous recommendations, and diminished human interaction remained prominent.

The present study provides a patient-centered assessment of awareness, perceptions, and acceptance of AI within internal medicine services in Dominica. Approximately two-thirds of participants demonstrated moderate-to-high acceptance, and favorable views were particularly evident for AI-supported diagnostic accuracy, faster decision-making, healthcare efficiency, and physician-assisted treatment selection. These findings are consistent with systematic reviews showing that patients generally recognize the clinical potential of AI but prefer its use within a human-supervised model rather than as an autonomous substitute for professional judgment [1,8,12]. The strong preference in this study for combined physician–AI decision-making therefore reflects a broader pattern of conditional acceptance rather than unconditional trust in technology. Awareness was reasonably high, but detailed knowledge was more limited. This difference has been observed in previous patient surveys, including the German study by Fritsch et al., where exposure to the concept of AI was more common than confidence in understanding it [2]. Miró Catalina et al. likewise identified gaps between general familiarity and knowledge of specific healthcare applications [11]. In the present study, both previous awareness and a basic understanding of medical AI were independently associated with higher acceptance. This relationship suggests that patient education is an important component of implementation. Clear explanations of what an AI system does, what information it uses, how its recommendations are interpreted, and where clinician responsibility remains could reduce uncertainty without overstating technological capability. The findings also demonstrate that trust is closely linked to physician oversight. Participants were substantially more comfortable with AI-assisted diagnosis and treatment when recommendations were reviewed or explained by a treating physician, whereas acceptance declined sharply for AI-generated diagnoses or treatment recommendations without physician confirmation. Similar patterns have been reported by Khullar et al., Esmaeilzadeh et al., and Robertson et al., whose studies showed that patients distinguish between AI as a supportive instrument and AI as an independent decision-maker [3,6,7]. Qualitative work has further emphasized apprehension regarding accountability, communication, and the possibility that technology could weaken the human component of care [4,10]. Recent evidence also indicates that introduction of AI can influence patient–physician trust, reinforcing the need for transparent clinical communication [13]. Privacy and accuracy were the dominant concerns in this study. Nearly three-quarters expressed concern about security of medical information, and more than two-thirds were concerned about incorrect AI recommendations. These concerns closely parallel findings from data-sharing research, in which patient support for AI depends on confidence in confidentiality, consent, institutional governance, and responsible secondary use of health information [5]. Privacy challenges associated with AI are amplified by the large datasets required for algorithm development and by uncertainty regarding data access and control [14]. The higher acceptance observed among younger and more highly educated participants also resembles demographic gradients reported in earlier surveys [2,7]. For Dominica, implementation strategies should therefore combine technical governance with patient-facing communication, accessible consent processes, clinician supervision, and educational approaches that accommodate different levels of digital and health literacy. Limitations This study has several limitations. Its cross-sectional design prevents assessment of changes in attitudes over time and does not establish causal relationships. Recruitment from internal medicine services at selected institutions restricts generalizability to all residents of Dominica. Self-reported responses are susceptible to recall and social desirability bias. In addition, acceptance was assessed using hypothetical AI-assisted scenarios rather than direct experience with deployed clinical AI systems.

Patients attending internal medicine services in Dominica demonstrated awareness of artificial intelligence and generally favorable acceptance of AI-assisted healthcare. Acceptance was strongest when AI was positioned as a tool that supports, rather than replaces, physician judgment. Participants recognized potential gains in diagnostic accuracy, efficiency, personalized care, and chronic disease monitoring, while expressing concerns about privacy, inaccurate recommendations, and reduced human interaction. Younger age, higher educational attainment, previous AI awareness, and better understanding of medical AI were associated with greater acceptance. Successful implementation should prioritize patient education, transparent communication, secure data governance, explainable clinical use, and physician oversight to preserve trust and support responsible integration of AI into internal medicine practice.

1. Young AT, Amara D, Bhattacharya A, Wei ML. Patient and general public attitudes towards clinical artificial intelligence: a mixed methods systematic review. Lancet Digit Health. 2021;3(9):e599-e611. doi:10.1016/S2589-7500(21)00132-1. 2. Fritsch SJ, Blankenheim A, Wahl A, Hetfeld P, Maassen O, Deffge S, et al. Attitudes and perception of artificial intelligence in healthcare: a cross-sectional survey among patients. Digit Health. 2022;8:20552076221116772. doi:10.1177/20552076221116772. 3. Khullar D, Casalino LP, Qian Y, Lu Y, Krumholz HM, Aneja S. Perspectives of patients about artificial intelligence in health care. JAMA Netw Open. 2022;5(5):e2210309. doi:10.1001/jamanetworkopen.2022.10309. 4. Richardson JP, Smith C, Curtis S, Watson S, Zhu X, Barry B, et al. Patient apprehensions about the use of artificial intelligence in healthcare. NPJ Digit Med. 2021;4(1):140. doi:10.1038/s41746-021-00509-1. 5. Aggarwal R, Farag S, Martin G, Ashrafian H, Darzi A. Patient perceptions on data sharing and applying artificial intelligence to health care data: cross-sectional survey. J Med Internet Res. 2021;23(8):e26162. doi:10.2196/26162. 6. Esmaeilzadeh P, Mirzaei T, Dharanikota S. Patients’ perceptions toward human-artificial intelligence interaction in health care: experimental study. J Med Internet Res. 2021;23(11):e25856. doi:10.2196/25856. 7. Robertson C, Woods A, Bergstrand K, Findley J, Balser C, Slepian MJ. Diverse patients’ attitudes towards artificial intelligence (AI) in diagnosis. PLOS Digit Health. 2023;2(5):e0000237. doi:10.1371/journal.pdig.0000237. 8. Beets B, Newman TP, Howell EL, Bao L, Yang S. Surveying public perceptions of artificial intelligence in health care in the United States: systematic review. J Med Internet Res. 2023;25:e40337. doi:10.2196/40337. 9. Ongena YP, Haan M, Yakar D, Kwee TC. Patients’ views on the implementation of artificial intelligence in radiology: development and validation of a standardized questionnaire. Eur Radiol. 2020;30(2):1033-1040. doi:10.1007/s00330-019-06486-0. 10. Haan M, Ongena YP, Hommes S, Kwee TC, Yakar D. A qualitative study to understand patient perspective on the use of artificial intelligence in radiology. J Am Coll Radiol. 2019;16(10):1416-1419. doi:10.1016/j.jacr.2018.12.043. 11. Miró Catalina Q, Femenia J, Fuster-Casanovas A, Marin-Gomez FX, Escalé-Besa A, Solé-Casals J, et al. Knowledge and perception of the use of AI and its implementation in the field of radiology: cross-sectional study. J Med Internet Res. 2023;25:e50728. doi:10.2196/50728. 12. Moy S, Irannejad M, Manning SJ, Farahani M, Ahmed Y, Gao E, et al. Patient perspectives on the use of artificial intelligence in health care: a scoping review. J Patient Cent Res Rev. 2024;11(1):51-62. doi:10.17294/2330-0698.2029. 13. Zondag AGM, Rozestraten R, Grimmelikhuijsen SG, Jongsma KR, van Solinge WW, Bots ML, et al. The effect of artificial intelligence on patient-physician trust: cross-sectional vignette study. J Med Internet Res. 2024;26:e50853. doi:10.2196/50853. 14. Murdoch B. Privacy and artificial intelligence: challenges for protecting health information in a new era. BMC Med Ethics. 2021;22(1):122. doi:10.1186/s12910-021-00687-3.