Contact Sales

ePA: Why a Full-Text Search Will Fail and AI Is the Better Solution

May 29, 2026

The electronic health record is extremely confusing. A full-text search, however, will not offer any added value. Only integrated AI would truly help doctors and patients.

Ever since the electronic health record (ePA) became widely available, one of the main criticisms has been that it is too confusing, too time-consuming, and offers too little practical value. Currently, the ePA is merely a document repository, comparable to an overflowing folder on a computer. Anyone who wants to get a picture of their own health or the condition of the patient sitting in front of them has to click through countless documents. The problem has been recognized, and it’s clear to everyone involved: Only when information in the ePA is available in a clear and easily accessible format will the digital record become a valuable tool for everyone in the healthcare system.

A full-text search is now planned as a solution; it could be available by the end of 2026. In theory, it’s supposed to enable doctors and patients to find medical information quickly. In practice, however, this feature will neither provide clarity nor save time. Just think of the search function on your own computer: It only helps if you know exactly what you’re looking for—and even then, it often requires considerable effort.

The Problem with Full-Text Search in the ePA

A full-text search operates on purely syntactic principles. While it is effective for searching for specific character strings, it cannot search based on semantics—that is, the meaning of words or sentences. This has a number of negative consequences:

  • It can only find the exact terms that were entered in the query, not synonyms. However, medical documentation is not usually that consistent. One doctor’s note might say “myocardial infarction,” another “heart attack,” and a third might use only the abbreviation MI. Hardly any physician will enter every possible search term to ensure that nothing is overlooked.
  • The time-consuming work only begins after the search: Medical information is complex and context-dependent. Full-text search does not reliably recognize these connections. So when healthcare professionals go through all the documents found, they’ll discover that the vast majority of them have no relevance whatsoever to their patients’ current situation.
  • The enormous heterogeneity of the documents within the ePA compounds the problem: A full-text search across scanned PDFs, free-text entries, structured lab reports, and annotated X-rays inevitably generates noise. Doctors must laboriously work their way through hit lists, opening documents, scrolling through them, and interpreting the content.
  • For patients, the situation is even more complicated: They are often unable to determine whether a particular document is relevant to their current situation or not. The search provides data, but no insight.

The full-text search in the ePA will therefore fail to deliver on any of the promises associated with it: It will neither save time nor give doctors a better picture of their patients. It may even prove detrimental. Doctors are already fearing more malpractice lawsuits because the ePA contains all information about a patient—and, in theory, all of it could be taken into account. However, the reality on the ground is different. In the average seven minutes that doctors have for each patient, it’s nearly impossible to review all the available data. Full-text search could further increase the fear of liability lawsuits because it promises a better overview—but fails to deliver.

How Integrated AI Can Improve the ePA

For these reasons, generative AI is the far superior solution. This type of AI operates semantically, not just lexically. It understands questions, places them in context, and can summarize, prioritize, and explain information. Users—whether medical experts or laypeople—can ask for information via chat. Healthcare professionals can ask: “What relevant pre-existing conditions does this patient have?” “Have there been any abnormalities in liver function test results over the past two years?” The AI would provide a structured, understandable answer—including references to sources within the ePA.

Patients can ask: “When was my last X-ray?” “How many times a day do I need to take my blood pressure medication?” The AI can explain this information in an understandable way. This strengthens health literacy and self-determination—goals that remain unattainable with a simple full-text search.

Averbis Provides Examples of Real-World Applications

This kind of interaction with personal health information is by no means a distant prospect. Averbis has already implemented a similar system in several hospitals. The AI application “Medical Summary” draws on all data available within the hospital regarding a patient. With the help of this technology, the data can be organized so that the patient’s medical history is available in a unified database. A summary can then be generated from this database and made interactive. Users can ask questions or obtain more detailed information. Results from the Bosch Health Campus have shown that this approach allows medical professionals to review the data relevant to their diagnostic or treatment decisions five times faster.

In summary, it can be said that the full-text search in the ePA does not provide any substantial added value. It is limited in scope and inefficient. It takes time without providing any real understanding. Instead of devoting resources to integrating it into the electronic health record, we should take the leap toward AI right away. This technology recognizes context, can be used as a chatbot, and can transform information into knowledge—for both doctors and patients. Only in this way will the ePA realize its hoped-for potential and improve medical care. Real-world examples from various clinics show that this is already possible today with Averbis’s “Medical Summary.”