Patient acceptance of healthcare digitalization through an AI Chatbot: system usability and patient satisfaction

Abstract

AI chatbots in healthcare represent a significant advancement that enhances service delivery while reducing workload for healthcare professionals. Yet empirical research in low- and middle-income countries remains limited. This study evaluated patient satisfaction and usability of the NH Smart Health Assistant, a specialist referral chatbot developed inhouse and deployed at Nawaloka Hospitals, Sri Lanka. To evaluate patient satisfaction and usability of the chatbot and to explore the relationship between these two constructs, a cross-sectional survey of 400 out patients assessed demographics, usability (UMUX-Lite), satisfaction, confidence in recommendations, and privacy concerns. Chi-square tests, correlations, and regressions identified predictors of satisfaction, reuse intent, and usability. The study revealed high overall satisfaction (85.5%) and strong intent to reuse (90.25%). Mean usability score was 82.0(±16.2). We identified data privacy(OR=3.14) and confidence in chatbot’s recommendations (OR=1.96) as the strongest predictors of satisfaction. Intention to reuse was positively associated with higher education, prior chatbot use, ease of use, and data privacy (OR 4.1, 4.0, 7.4, 2.9). Viewing data privacy as secure boosted UMUX-Lite by ~18 points, while being educated up to A/L added 11 more (all p < 0.05). We demonstrated that healthcare chatbots achieve high user satisfaction. Successful integration depends on building user trust rather than demographics or prior experience of chatbots. Multilingualism, voice interaction support, and data transparency are recommended to sustain engagement when scaling AI nationally. Future research should broaden demographic sampling beyond urban populations to inform generalization strategies.