AI as Opportunity and Challenge in Hypertension Management

AI as Opportunity and Challenge in Hypertension Management

Artificial intelligence is increasingly shaping medicine and is already widely used—though sometimes met with skepticism. Around 50% of internet searches are now AI-related, with expectations of over 80% by 2027, reflecting changing user and patient behavior. Especially in potentially serious illnesses, patients often seek extensive information online, and about a quarter of all ChatGPT queries are health-related. In response, ChatGPT Health (for patients) and ChatGPT Healthcare (for doctors) were launched in the U.S. this spring to support health questions, visit preparation, and case comparison — but without crossing the border of providing diagnoses.

What are the opportunities and risks?

Today, AI systems are no longer just search engines but rather answer engines capable of providing tailored responses to the questions of an individual and/or a patient. At the same time, thanks to their semantic programming and speech recognition, they can identify people’s concerns, needs, and emotional states and respond appropriately. They adapt their responses to the user’s mood, are always patient, explain things repeatedly if necessary, and are available 24/7. Physicians etic communication style appeals to and even spoils users and patients and raises expectations toward physicians, underscoring the importance of patient-centered communication in healthcare and medical education. At the same time, however, these systems have clear limitations; they can trigger false alarms, overinterpret information, or provide incorrect results due to biased training data or improper use.

How can AI be applied effectively?

Research on trust in AI shows mixed results: A 2022 JAMA Internal Medicine study by Ayers et al. found that 80% of patients rated AI responses as better than those of physicians, mainly due to greater length and empathy¹. In contrast, a Munich-based study involving U.S. patients identified skepticism and concerns about AI competence, particularly when patients themselves or their doctors use AI, alongside fears of overreliance². But at the same time, AI can be a valuable tool, for example, in preparing patients for consultations by helping them formulate questions in advance. It can also support patient education through videos or chatbots in waiting rooms, addressing routine questions and enabling physicians to focus on more complex issues. Or physicians can draw on the phrasing used by chatbots to more easily transition to a higher level of explanation.

How can AI be used in treatment of hypertension?

AI excels at recognizing patterns, processing large datasets, and continuously assessing risks, making it well-suited for managing hypertension, which can involve dynamic disease courses, multiple influencing factors and various contexts. It is already used to analyze everyday data—such as pulse waves, sleep, and activity from wearables—to support early detection. AI also enhances the evaluation of 24‑hour blood pressure monitoring and ECGs by identifying patterns efficiently. In addition, digital hypertension assistants (e.g., from Omron) use apps to interpret readings, detect medication errors, explain warning signs, visualize trends, and support patient adherence.

Estimation of Treatment Success

Hospitals and research centers can use AI to make predictions and calculate probabilities for stroke and heart attack risks, but also to assess which type of patient is most likely to respond to which medications and who might develop which side effects (Clinical Decision Support Systems (CDSS) or Predictive Analytics in Healthcare).
To do this, large electronic health databases can be analyzed in a very short time. An important part of modern planning is therefore estimating treatment success: ‘How likely will the patient benefit from the treatment?’ To this end, patient data such as blood pressure profiles, heart rates, medication prescriptions and intake logs, imaging data, and laboratory results can be automatically consolidated and analyzed together to identify patients who are expected to experience a significant reduction in blood pressure or those who may not respond to treatment. Until now, this has often been done manually on a case-by-case basis. If sufficient data is available, screenings could also be conducted quickly within a patient group. AI-based image analysis, including vascular assessments and measurements of vascular anatomy, is also likely to play a role in facilitating procedure planning for renal denervation.

So, what might a realistic application of AI be?

The future could be a combination of both:

• AI that analyzes data, identifies risks early on, and prioritizes anomalies
•  that answers patients’ standard questions and prepares them for their consultation with the doctor
•  and that creates greater effectiveness and satisfaction for both parties.

And a doctor who assesses context, verifies statements, and makes final clinical decisions – decisions that AI-driven data might support more effectively and quickly. AI is here to stay in everyday (clinical) practice, but it can offer great opportunities if used correctly.

Ressources:

1. Ayers, 2023: DOI: 10.1001/jamainternmed.2023.1838
2. Reis, 2025: DOI:10.1001/jamanetworkpen.2025.21643