02 Nov 2026
AI health assistants are becoming an increasingly common part of people's lives. Checking symptoms, asking about medications and receiving personalised advice can be quick and convenient. But healthcare is not an ordinary digital service scenario. When users open these platforms, they are often worried about their health, unsure whether to trust AI-generated advice, and aware that mistakes or data leaks could have serious consequences. For AI healthcare platforms, anxiety can therefore become a significant barrier to use.

But research published in the Journal of Business Research by Dr Wangshuai Wang of International Business School Suzhou (IBSS) at Xi'an Jiaotong-Liverpool University and his collaborators, titled "Alleviating User Anxiety towards AI Health Assistant Platforms: Integrating HAII-TIME and Perceived Control Perspectives," finds that anxiety is not a blanket obstacle, but has three distinct sources—each requiring a different design response.
Through two online experiments involving 540 Chinese users and 360 Spanish users across mental and physical health scenarios, and a field survey combining questionnaire responses from 880 users of a major Chinese AI healthcare app with their actual usage data), the study found a clear pattern: there is no one-size-fits-all interface solution. The most effective design depends on what is making the user anxious.
First: Technology-use anxiety — requires interactivity
This type of anxiety stems from a loss of behavioural control. Users worry they won't know how to operate the system, will be unable to correct mistakes, or will not know what will happen next.
The study shows that high information interactivity is the best antidote. Multi-turn conversations, remembering previous user inputs and offering relevant follow-up questions can make users feel that their actions are shaping the consultation process. For technology-anxious users, platforms should adopt progressive disclosure: start with a simple interface and only reveal advanced options when needed. Clear "show me how" guidance and conversations that refer back to users' earlier responses, can also build confidence without overwhelming them.
Second: Privacy-leakage anxiety — requires visible security
This type of anxiety is linked to a lack of cognitive control. Users cannot see how information is stored, processed, or shared, so they cannot predict what will happen to their data. The study notes that strong visual non-verbal cues—such as lock icons, professional clinical design, and authoritative security certifications—can make privacy protection more visible and easier to understandable. But the key is: more design is not always better. Field evidence shows that combining extremely high interactivity with abundant visual cues can actually heighten privacy vigilance by over-signalling the amount of data being collected.
For privacy-concerned users, security prompts should be placed next to sensitive data fields, with clear explanations of why the information is needed and how it will be protected. Generic privacy policies are far from sufficient; reassurance must appear at the moment of disclosure.
Third: Personalised-service anxiety — requires explanation + correction
This type of anxiety involves both behavioural and cognitive control. Users want evidence that the AI is capable, but also room to challenge or refine its recommendations. Therefore, the strongest response is a combination of interactive dialogue and explanatory visual cues. Platforms can support shared decision-making through features such as “why this recommendation?”, explaining which symptoms or data influenced the judgment, while convenient correction options allow users to add constraints or request alternatives.
Beyond the interface: AI healthcare needs deeper safeguards
The study emphasises that interface design alone cannot solve every concern surrounding medical AI. AI health services also require safeguards beyond the interface: recommendations should be clinically validated, serious cases should have clear human escalation pathways, and system language should reflect the seriousness of the context rather than relying on a generic "cheerful chatbot" tone.
Dr Wangshuai Wang notes: "The most common mistake AI healthcare platforms make is treating all users' anxiety as the same anxiety. Some people fear they won't know how to use it, some fear data leaks, and some fear the AI won't understand them. These three groups need completely different design languages. Diagnose the anxiety first, then prescribe the design—only then can AI healthcare truly become controllable, trustworthy, and suitable for high-stakes patient decisions."
About author
Dr Wangshuai Wang is an Associate Professor of Marketing at International Business School Suzhou, Xi’an Jiaotong-Liverpool University. His research has been published in leading international and Chinese journals, including Journal of Retailing, Journal of Service Research, Psychology & Marketing, Journal of Business Research, Marketing Letters, Tourism Management, Journal of Travel Research, Nankai Business Review, and Advances in Psychological Science. He currently serves as an Associate Editor of Current Psychology and as an Editorial Board Member of journals including Journal of Business Research and Marketing Intelligence & Planning.
About journal
The Journal of Business Research is an internationally renowned journal in the field of business research, rated 3 stars in the ABS Academic Journal Guide and A in the ABDC Journal Quality Guide. The journal covers a wide range of topics including consumer behaviour, marketing strategy, and organisational behaviour, and is dedicated to publishing high-quality research with theoretical contributions and practical implications.
02 Nov 2026