A white paper from UK digital health company Aide Health has found that patients are more likely to disclose sensitive health information to AI systems than to clinicians. The report, “Building Patient Trust in AI,” analyzed data from Aide Health’s NHS programs alongside peer-reviewed studies.
In NHS primary care, Aide Health’s conversational platform found that 26% of asthma patients admitted to not taking their medication as prescribed – a figure higher than typically reported in face-to-face consultations.
The report found that while 48% of people are comfortable with AI identifying health risks through wearables, only 35% believe it improves care quality. Trust levels vary across communities, reflecting broader health inequalities.
“Healthcare AI sits at the crossroads of trust and empathy,” said Ian Wharton, Founder and CEO of Aide Health. “People trust technology to listen without judgment, but they still want to feel understood. Our research shows we can design systems that do both – helping patients speak honestly while supporting clinicians to respond with compassion.”
The report introduces a design principle called adaptive neutrality, which determines when AI should be neutral and when it should show empathy. According to the report, patients often prefer AI for tasks such as medication tracking or lifestyle monitoring, while human empathy remains essential for diagnosis or treatment decisions.
Despite rapid growth in AI investment – projected to reach $148.4 billion by 2029 – 60% of patients remain uneasy about its role in healthcare, according to the report.
Aide Health has announced Mirror, an AI-powered app designed for patients that listens during healthcare consultations and produces summaries in plain English. Research cited in the report shows up to 80% of medical advice is forgotten immediately, and nearly half of what’s remembered is inaccurate.
Aide Health uses conversational AI to help patients adhere to treatment, track their health, and build health literacy, while providing clinicians with real-time patient data.
