The data: Some 74% of physicians across specialties say patients have become more likely in the past six months to propose a self-diagnosis or treatment plan from an AI tool, according to a just-published Doximity survey. That includes 30% who say this behavior has increased dramatically. Nearly 3,400 US clinicians spanning 20 specialties were polled in September.

Why it matters: Consumers have traditionally relied on online health information to self-diagnose, with search engines ranking as the most-useful channel among those who do, per EMARKETER’s January US Digital Health Survey. But as search engines integrate AI, consumers can move beyond browsing separate articles to having detailed, back-and-forth conversations about their symptoms. This continuous exchange of personalized information should give consumers greater confidence in diagnosing themselves or developing treatment plans before—or even without—consulting a physician.

AI is helping some patients prepare for appointments and steering others toward the wrong conclusions. Some 55% of oncologists and 41% of internal medicine physicians in Doximity’s survey report that patients come to appointments asking more informed questions because of AI. But AI’s agreeable tone can also create false confidence: 27% of physicians say AI has made patients more confident in incorrect self-diagnoses. The stakes are even higher when patients use AI to develop treatment plans without physician oversight.

Implications for pharma and healthcare marketers: As more consumers use AI to self-diagnose or develop treatment plans, many of their queries will likely prompt AI to recommend or mention specific medications and treatments. That makes visibility in AI-generated health responses important for drugmakers.

These tools could shape patients’ awareness of treatment options before they consult a physician, much as pharma ads can prompt patients to ask doctors about specific medications. Pharma marketers will need to make accurate, up-to-date information on their medications, including indications, safety risks, and eligibility, accessible to AI platforms through unbranded disease awareness content rather than just promotional product pages.

Not everyone who turns to AI for medical advice goes on to see a doctor, though. If more patients bring AI-generated diagnoses and treatment plans to appointments, many others likely act on that guidance without consulting a doctor first.

Healthcare marketers can respond in three ways:

• Promote the dos and don’ts of using AI for health advice through patient education campaigns on social media, commonly used health websites, and patient portals.
• Use that messaging to highlight the risks of relying on AI-generated diagnoses and treatment plans without physician oversight.
• Work with physicians to incorporate guidance on evaluating AI health information into patient communications and appointment discussions.