Artificial Intelligence in Medication Adherence: A National Assessment of Knowledge, Attitudes, and Perceptions Among Chronic Disease Patients in Jordan

Abstract

Purpose: To assess knowledge, attitudes, and perceptions toward AI-based medication adherence tools in a national cross-sectional survey among Jordanian adults with chronic diseases and to identify factors associated with favorable acceptance. Patients and Methods: A national cross-sectional online survey was conducted between January and July 2025 using convenience and snowball sampling and a self-developed, pilot-tested, and content-validated Arabic questionnaire. The questionnaire captured sociodemographic characteristics, digital literacy, chronic disease information, and KAP toward AI-based medication adherence tools. Knowledge was assessed using eight statements describing AI capabilities and rated on a five-point Likert scale, which were recoded dichotomously for scoring (score range 0-8; ≥5 indicating good knowledge). Attitudes and perceptions were measured using five-point Likert scales (mean scores ≥3.5 indicating positive or high levels). Descriptive statistics, chi-square tests, multivariable logistic regression, sensitivity analyses, and polypharmacy subgroup analyses were performed to identify factors associated with favorable acceptance. Results: Among 552 participants (mean age 52.7 ± 12.6 years; 56.3% female), the most prevalent chronic diseases were hypertension (213, 38.6%) and diabetes mellitus (179, 32.4%), with 154 (27.9%) reporting polypharmacy (≥5 medications). 59.2% demonstrated good knowledge (mean score 5.3 ± 1.8). Moderately positive attitudes were observed in 48.6% (mean 3.6 ± 0.9), while 52.7% reported high perceptions (mean 3.7 ± 0.9). Strong support was reported for AI-based reminders (80%) and educational functions (76%), whereas endorsement of predictive features was lower (49-58%). Major concerns included privacy (71%), technical reliability, and reduced human interaction. Chi-square tests showed significant associations with age and digital literacy (p < 0.001). Multivariable logistic regression confirmed that younger age, higher education, advanced digital literacy, and smartphone ownership independently predicted favorable KAP (p < 0.001). Polypharmacy was associated with greater receptivity in unadjusted analyses but was not an independent predictor after adjustment. Findings remained robust in sensitivity analyses. Conclusion: Jordanian patients with chronic diseases show moderate-to-good knowledge and cautiously positive attitudes toward AI-enabled medication adherence tools. Acceptance is shaped mainly by digital literacy and trust, underscoring the need for governance frameworks, clinician oversight, Arabic-language design, and targeted digital literacy initiatives to support equitable integration into national platforms such as Hakeem.

Ähnliche Arbeiten