Corti’s AI Outperforms Rivals in Medical Coding
▼ Summary
– Corti’s Symphony for Medical Coding is an AI system that treats coding as a reasoning task, not a classification problem, and claims to outperform major AI models by up to 25% in clinical accuracy.
– Medical coding, which converts clinical details into standardized codes, is a highly error-prone administrative task with significant financial and public health consequences.
– The system uses a four-agent sequential process that mirrors human coders, extracting evidence, searching codes, validating against guidelines, and reconciling the final output.
– Errors in conventional coding can obscure critical public health data, as demonstrated by a study where Corti’s system identified three times more suicide attempts than official records.
– Symphony is available via API, operates across US and European coding systems without retraining, and provides auditable outputs linked to supporting clinical evidence.
A new approach to medical coding automation is challenging the status quo, with a system that treats the task as a complex reasoning process rather than a simple classification problem. This shift is yielding significant gains in accuracy, addressing one of healthcare’s most persistent and costly administrative burdens. The process of translating clinical notes into standardized billing and diagnostic codes is notoriously error-prone, with the American ICD-10-CM system alone containing over 70,000 potential diagnosis codes. Mistakes are common, expensive, and often go undetected, undermining both financial operations and public health data.
Corti, a clinical AI company based in Copenhagen, has introduced a solution designed to tackle this issue head-on. Its new product, Symphony for Medical Coding, is an agentic system that the company claims outperforms competing models from major tech firms by as much as 25% on key clinical accuracy benchmarks. The system is now available via API.
The claimed performance advantage stems from a fundamental difference in methodology. Most AI systems frame medical coding as a classification task, where a model predicts the most likely code based on its training data. This approach falters because coding guidelines are constantly updated, rendering models trained on historical data structurally obsolete. Corti’s framework, developed through peer-reviewed research presented at the EMNLP 2025 conference, redefines the challenge. Their “Code Like Humans” paradigm treats coding as a reasoning task that requires interpreting evidence, context, and complex guidelines.
“Most AI systems fall short because they treat coding as labeling, not reasoning,” explained Lars Maaløe, CTO and co-founder of Corti. “Correct coding depends on evidence, context, hierarchy, and guideline interpretation. We built Symphony to follow the same decision process expert coders use, and that is why the performance gap is so meaningful.”
The system operates using a sequence of four specialized agents, mirroring a human coder’s workflow. An evidence extractor first isolates conditions mentioned in a clinical note. An index navigator then searches the ICD alphabetical index for candidate codes. A tabular validator checks those candidates against official guidelines, and a final code reconciler sequences and validates the output. This research was built upon an analysis of 1.8 million patient encounters, forming the largest peer-reviewed study of its kind in this field.
The impact of inaccurate coding extends far beyond revenue cycle management. Corti points to a peer-reviewed study of Danish patient data where its system identified three times as many suicide attempts as were officially recorded. These cases were documented in clinical notes and medication records but were missed by human coders working under time constraints. When such critical data is lost, health systems cannot accurately monitor public health trends, allocate resources effectively, or design targeted interventions.
“Medical coding has been treated as a back-office cost center for decades,” said Andreas Cleve, CEO and co-founder of Corti. “It isn’t. It’s the data layer that healthcare runs on. Getting it right changes what health systems can see, decide, and do.”
Symphony for Medical Coding is designed to operate across both U. S. and European coding environments without requiring local retraining. It handles the complex U. S. systems for diagnoses and procedures while also supporting WHO-maintained ICD-10 coverage for Europe, with beta availability currently expanding into the UK, Germany, France, and Denmark. A key feature is its auditable output; every assigned code is linked directly to the supporting clinical evidence, and any ambiguities are flagged for human review.
The product is accessible through the Corti Console, integrates with the Corti Agentic Framework, and supports industry standards for interoperability. Enterprise and sovereign cloud deployment options are available. Founded in Copenhagen with additional offices in New York and London, Corti has raised $100 million in total funding and serves over 100 million patients annually through partnerships with major health systems, including the NHS. The launch of Symphony represents the commercial realization of the company’s research-driven philosophy, translating validated academic concepts into production-grade infrastructure.
(Source: The Next Web)
