Corti AI introduces "Symphony" for automated medical coding

Contents

Medical coding forms the basis for billing in the healthcare system, as it translates the clinical service provided into a standardized language that can be processed by health insurance companies. At the same time, however, medical coding is very time-consuming and prone to errors. The correct assignment of diagnosis codes – especially using the International Classification of Diseases (ICD) – from unstructured medical documentation is therefore a very important but also difficult task that requires trained, highly qualified personnel.

Corti AI, a technology company from Copenhagen specializing in healthcare, has now released “Symphony for Medical Coding,” a system designed to make the medical coding process for billing more efficient and precise using artificial intelligence.

Adapted for German Systems

What distinguishes Symphony from other systems, according to the manufacturer, is a more comprehensive understanding of medical documentation through the use of multiple AI agents and a systematic, multi-stage approach to performing the coding. At the same time, the software is adapted nationally and regionally. In recent months, the manufacturer has further developed the software specifically for the German market. According to the manufacturer, the implementation now includes complete diagnosis and procedure coding as well as the recognition of medical codes based on the widely used SNOMED-CT (Systemized Nomenclature of Medicine – Clinical Terms) coding system.

According to CortiAI, this covers approximately 90 percent of the requirements of German hospitals. The challenge in Germany lies particularly in the complexity of the DRG system (Diagnosis Related Groups) used for billing. Here, the smallest nuances in documentation (and thus in coding) can significantly influence the amount of financial reimbursement.

In parallel with the release, Corti has summarized its analyses and results in a scientific paper and published it on arXiv.

First Data in Real Applications

One of the biggest challenges in using generative AI in medicine is “AI hallucinations” and, in general, the generation of erroneous data and inaccuracies. According to Corti, “Symphony for Medical Coding” has shown robust and secure performance since its introduction on April 1, 2026.

Within the first week, approximately 8000 complex medical coding requests were processed, with only 16 errors registered. This corresponds to an error rate of 0.2 percent of cases, which is significantly below the usual error rates for manual coding by specialists. It appears that specialized AI systems have now reached a level of performance with which they are capable of taking on complex administrative tasks in the healthcare system without compromising safety.

Strategic Partnership with Dedalus

To roll out the new system for AI-assisted medical coding widely, Corti has entered into a strategic partnership with Dedalus, among others, one of the world's leading providers of hospital information systems (KIS).

The cooperation, which already includes about 20 partners in Germany, is now to be extended beyond national borders throughout Europe. The goal is to embed Symphony directly into the familiar software systems of doctors and coding specialists. Instead of laboriously transferring data manually, the AI suggests suitable codes in the background, which only need to be validated by the specialist.

Efficiency Gains Through AI Against Skilled Labor Shortage

If hospitals could automate or at least partially automate their coding almost error-free with the help of AI, this would be a decisive contribution to economic stability.

In fact, medical coding currently ties up considerable personnel resources, while errors in coding cause significant financial damage. Also, considering the growing shortage of skilled workers, the automation of coding offers the opportunity to make processes more efficient and save costs.

(dmk)