From Dr. Google to Dr. AI: How AI is reshaping patient behaviour

5 actions healthcare leaders should take as AI reshapes patient behaviour

What follows are the five themes that ran through the discussion. 

1. Understand where AI is already entering your patient journey

AI adoption is no longer a future scenario. In Germany, 58% of people already use AI, with ChatGPT used by 71% of AI users, followed by Gemini at 50% and Microsoft Copilot at 43%, according to Bitkom’s 2026 survey (Bitkom Research, 2026).

Health-related usage is also becoming mainstream. Bitkom found in 2025 that 45% of people in Germany had already used AI chatbots to clarify symptoms or ask general health questions (Biktom, 2025). International data shows how quickly this can translate into action: Rock Health’s 2025 U.S. Consumer Adoption Survey found that 32% had used AI chatbots for health information, 64% of those users engaged with them at least weekly, and 81% had taken at least one relevant action following an AI interaction (Rock Health, 2026).

The panel discussion reflected the same behavioural shift. Dr. Ina Lucas described patients increasingly arriving at the pharmacy with AI-informed questions, suspected diagnoses and suggested treatments.

What healthcare leaders should do now: Map where AI already influences your patient journey, from initial symptom research and product discovery to consultation, treatment and follow-up. Identify the moments where patients may arrive AI-informed and determine where your organisation needs to provide trusted information, validation or human guidance.

2. Design the role of healthcare professionals around interpretation, not information delivery

AI gives patients unprecedented access to health information, but access does not equal understanding.

Dr. Markus Frühwein described patients who have already explored their symptoms with AI as potentially better oriented when they enter the consultation. The opportunity is therefore not to compete with AI on information retrieval, but to focus professional expertise where it adds the most value: contextualising information for the individual, identifying risks and translating knowledge into appropriate action.

Dr. Ina Lucas described a similar spectrum in pharmacies, ranging from highly informed patients to people who have received an AI-generated answer but cannot assess what it means for them.

What healthcare leaders should do now: Redesign patient interactions around an AI-informed patient. Train teams to ask what information patients have already received, validate or correct it efficiently and focus consultation time on interpretation, individual context and the appropriate next step in care.

3. Make trusted health information discoverable by AI

As patients increasingly use AI at the beginning of their health journey, visibility can no longer be understood only in terms of Google rankings, websites or traditional search.

Lutz Boden highlighted the growing importance for manufacturers of ensuring that reliable product, medicine and safety information can be found, understood and correctly contextualised by AI systems. Technology platforms are increasingly becoming an intermediary between healthcare organisations and patients.

For healthcare and pharma companies, this creates a new commercial discipline: AI visibility and discoverability.

What healthcare leaders should do now: Audit how your organisation, products and health information appear across major LLMs. Identify information gaps, improve the structure and authority of your content and ensure that trusted medical and product information is machine-readable, consistent and accessible across the digital ecosystem.

This should not remain an isolated marketing initiative. Connect AI visibility with CRM, customer intelligence, commercial activation and broader AI use cases to move from individual pilots to scalable value creation.

4. Build connected patient pathways before debating individual channels and responsibilities

The panel repeatedly returned to one structural challenge: patients still move through a fragmented system.

Dr. Ina Lucas described the potential for pharmacies to become accessible local health hubs that help patients navigate towards the right point of care. But this model depends on something more fundamental than expanding individual services: interoperability and connected information flows across pharmacies, medical practices and other healthcare providers.

Without reliable access to relevant information such as medication, vaccination status and existing conditions, every interaction risks starting again from zero.

What healthcare leaders should do now: Start with the patient journey rather than organisational boundaries. Identify where patients currently repeat information, experience unnecessary handovers or lose continuity between channels and providers. Prioritise use cases where shared data, clear escalation paths and interoperable processes can remove those frictions.

The question should first be “What should the optimal patient pathway look like?” and only then “Which organisation or profession owns each step?”

5. Treat security as an architecture decision, not a final compliance check

Greater connectivity and AI adoption create new value, but also new attack surfaces.

Arwid Zang highlighted how dramatically the security environment has accelerated. Vulnerabilities that once gave organisations months to react can now be exploited within minutes. AI introduces additional risks because systems increasingly consume external data, documents and instructions automatically.

That makes traditional security reviews at the end of a project insufficient.

What healthcare leaders should do now: Bring cybersecurity expertise into AI and digital-health initiatives at the design stage. Define which data enters the system, where it originates, which systems and partners are trusted, what an AI agent is allowed to execute and how compromised or manipulated inputs would be detected.

For every AI use case involving health data, ask three questions before implementation:

1. What data can enter the system and from where?
2. What actions can the AI take based on that data?
3. What happens if either the data or the AI output is compromised?

Security by Design should be a prerequisite for scaling AI in healthcare, not a review conducted after go-live.