Fighting one’s way through the guideline jungle: GUIDE-AI develops intelligent therapy navigators

Berlin/Neuruppin, 4 March 2026

Guidelines collate recommendations for state-of-the-art treatment of particular diseases. They help doctors find the best possible therapy. The problem is that they are increasingly elaborate and difficult to comprehend. An alliance of public and industrial partners under the direction of the Charité – Universitätsmedizin Berlin plans to redress the situation with the project GUIDE-AI: the idea is to develop AI-based assistants that call attention to deviations from guidelines and in this way may improve medical treatment for millions of people. The EU Innovative Health Initiative provides funding of 9,5 million Euro for a four-year period.  

Medical guidelines are intended to help doctors in their choice of the best possible therapy for individual patients. However, medical knowhow doubles in volume every 73 days so that some guidelines have grown to more than a hundred pages, according to Matthias Gröschel, physician and head of a team of Charité researchers that combines Infectious Diseases, Respiratory Medicine and Critical Care Medicine: “And what’s more, recommendations are updated at regular intervals. It is a huge challenge to memorize all details and updates for a wide range of diseases.”

An example: cardiac insufficiency of the type HFrEF (heart failure with reduced ejection fraction). Physicians following the therapy guidelines need to consider several factors for the individual patient, such as blood pressure, heart rate, kidney values, electrolytes, comorbidities and many others. Since some medication relieves heart failure but at the same time affects kidneys and blood pressure, the physician must keep nephrological guidelines in mind for a patient with renal insufficiency. In short: it has become a complex job to decide on a safe and efficient therapy for individual HFrEF patients and costs time, which is precious in daily healthcare practice.

19 partners for a European guideline navigator

The GUIDE-AI consortium plans to use LLMs (large language models) for a rapid orientation in the guideline jungle. In the context of the project, research institutions work hand in hand with small and medium-sized companies, patient representatives and partners in the pharmaceutical industry from a total of seven European countries plus Israel. The guideline assistants will be drawn up in several European languages and allow for local specifications. Matthias Gröschel acts as coordinator, and AstraZeneca has the project management. The MHB is one of the project partners.

Dr. rer. medic. habil. Felix Mühlensiepen heads the MHB team of young researchers who focus on “User Experience in Digital Health”. He underlines the potential of AI-based guideline navigators to improve healthcare in Europe and promote evidence-based medicine in daily practice. He says a key factor is the systematic involvement of later users right from the start, and this is what the participation of his young team in the project GUIDE-AI is about.

The primary objective is to create navigators which – integrated into the hospital information system or medical office software – square the therapy prescribed on the basis of individual patient data with guideline recommendations and suggest amendments in case of discrepancies. A hospital information system, similar to medical office software, is a computer program for medical personnel displaying, among other things, individual patients’ data, from radiological images to lab results and lists of prescribed medications.

Back to cardiac insufficiency: The HFrEF navigator is to alert the doctor in the surgery or hospital if only three out of four recommended medications have been prescribed or a change of dosage for one of the drugs would be preferable. Matthias Gröschel: “The attending physician will ultimately determine the therapy based on all the relevant information. But the GUIDE-AI navigators are meant to suggest possibilities to optimize the therapy pursuant to guidelines.”

Focus on four common chronic diseases
The first focus of the GUIDE-AI team is on four chronic diseases that are widespread and involve considerable health burdens but are not always treated as recommended, despite established guidelines. These are: cardiac insufficiency of the HFrEF type, chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD) and asthma. Over 160 million people all over Europe are estimated to suffer from these complaints. The plan is to develop a specific navigator for each of the four diseases, with other disorders to follow.

The first step will be a survey to identify the needs for guideline assistants in medical practice. Various AI models will then be tested, and the best solution or solutions will be used as a basis for the respective navigator. The protection of sensitive medical data will be a priority; AI models are to be used on local servers and comply with the European data protection directive as the minimum requirement.  

Part of the project is a study to check whether physicians actually use the navigator to prescribe treatment according to guidelines more frequently. The program is also intended to provide easy-to-understand information and thus advise not only medical staff but the patients themselves on how to cope with their illness.

Background: GUIDE-AI
The GUIDE-AI consortium comprises 19 partner institutions from research, the health sector and the pharmaceutical industry based in Germany, Italy, the Netherlands, Estonia, Slovenia, Great Britain, Switzerland and Israel, with the Charité – Universitätsmedizin Berlin as the coordinating institution. The project receives funding from the Innovative Health Initiative (Grant Agreement No. 101253015). Funds for the IHI come from the EU research and innovation program Horizon Europe as well as from COCIR, EFPIA, Europa Bio, MedTech Europe and Vaccines Europe.

Scientific contact Charité:
Dr. Dr. Matthias Gröschel
Joint Department: Infectious Diseases / Respiratory Medicine / Critical Care Medicine
Charité – Universitätsmedizin Berlin
Phone: +49 30 450 570 400
E-Mail: matthias.groeschel@charite.de

Scientific contact MHB:
Dr. rer. medic. habil. Felix Mühlensiepen
Director, team young researchers “User Experience in Digital Health”
Medizinische Hochschule Brandenburg
E-Mail: felix.muehlensiepen@mhb-fontane.de

Links:
Factsheet GUIDE-AI listing all partners: https://www.ihi.europa.eu/projects-results/project-factsheets/guide-ai
Joint Department: Infectious Diseases / Respiratory Medicine / Critical Care Medicine,  Charité: https://infektiologie-pneumologie.charite.de/