Corti Unveils Advanced Medical Coding AI: Symphony
Corti, a health AI enterprise headquartered in Copenhagen, has developed Symphony, leveraging insights from a peer-reviewed investigation heralded as the most extensive medical coding study to date.
This innovative tool redefines medical coding as a cognitive task rather than a mere labeling exercise; it is now accessible via API.
Medical coding—the intricate process of transforming clinical documents, diagnoses, and procedures into standardized alphanumeric codes—serves critical functions in billing, reporting, and public health analytics.
Despite its significance, it remains one of healthcare’s most error-laden and impactful administrative endeavors.
In the American coding schema, ICD-10-CM, there are a staggering 70,000 diagnosis codes. Consequently, the prevalence of inaccuracies is a routine, costly issue, often eluding detection.
Corti has unveiled Symphony for Medical Coding, a novel solution purportedly outperforming offerings from tech titans such as OpenAI, Anthropic, Amazon, Oracle, and Microsoft by up to 25% in clinical accuracy assessments.
The system is available via API as of today.
Corti’s claims hinge upon a pivotal methodological differentiation. Conventional AI frameworks typically approach medical coding as a classification dilemma, predicting the most probable code from a training distribution based on clinical notes.
Such a strategy falters, as coding guidelines are perpetually evolving, rendering historically trained models insufficient.
Corti’s methodology, developed through a rigorously peer-reviewed framework titled Code Like Humans and acknowledged at EMNLP 2025—an esteemed machine learning symposium—regards coding as an intellectual reasoning exercise.
“The majority of AI solutions falter in the realm of medical coding because they conflate it with mere labeling rather than sophisticated reasoning. Accurate coding is contingent on the synthesis of evidence, contextual nuances, hierarchical structures, and interpretative guidelines.
Symphony for Medical Coding was crafted to replicate the decision-making processes employed by expert coders, which elucidates the substantial performance differential,” noted Lars Maaløe, PhD, CTO and co-founder of Corti.
Symphony employs a sequential process of four agents: an evidence extractor identifies conditions within clinical notes, an index navigator explores the ICD alphabetical index for potential codes, a tabular validator scrutinizes candidates against established guidelines, and a code reconciler ensures that final outputs are accurately sequenced and validated. Each phase mirrors the actions of a trained human coder.
This research was predicated on data from 1.8 million patient encounters, thus marking it as the largest peer-reviewed study of its kind.
The ramifications of inadequate coding extend beyond financial implications. Corti references a peer-reviewed examination of Danish patient data, which revealed that its system identified three times the number of suicide attempts compared to official coding records—cases present in clinical documentation and medication logs that were overlooked by overburdened coders.
“For decades, medical coding has been perceived merely as an administrative cost center; however, it constitutes the foundational data layer underpinning healthcare systems.
Accurate coding profoundly influences what health organizations can perceive, decide, and execute,” assertedAndreas Cleve, CEO and co-founder of Corti.
Omitting such instances hampers health systems’ capabilities to monitor trends, allocate resources, or devise effective interventions. The coding framework is not merely overhead; it fundamentally shapes how health systems comprehend their realities.
Symphony for Medical Coding stands as Corti’s inaugural system designed to perform across both US coding landscapes—ICD-10-CM for diagnoses, ICD-10-PCS and CPT for procedures—and European contexts without requiring localized retraining.
ICD-10 coverage for Europe, overseen by the WHO, is currently operational in beta as the company expands its reach to the UK, Germany, France, and Denmark.
The system generates auditable outputs: each code assigned is directly tied to the clinical evidence backing it, with any ambiguities flagged for human scrutiny.
Accessible through the Corti Console, Symphony integrates with the Corti Agentic Framework and accommodates both A2A and MCP standards. Solutions for enterprise and sovereign cloud deployments are also on offer.
Founded in Copenhagen, Corti maintains additional offices in New York and London. The company has raised a total of $100 million and serves more than 100 million patients yearly across various health systems, including the NHS.
The introduction of Symphony represents the commercial realization of the Code Like Humans research, aligning with Corti’s ethos of validating concepts through peer-reviewed discourse prior to their development into production-grade infrastructures.
Source link: Thenextweb.com.
