What You Should Know
- The Launch: Clinical AI lab Corti has released Symphony for Medical Coding, an agentic AI model available via API for healthcare software developers.
- Performance Benchmark: The model “claims” to outperform generalized LLMs from tech giants—including OpenAI, Anthropic, Amazon, Oracle, and Google—by more than 25% in clinical accuracy benchmarks.
Beyond the Back Office: The Human Cost of Bad Data
The problem is fundamental: generic LLMs are essentially highly advanced autocomplete engines. They treat medical coding as a simple “prediction” or “labeling” task. But medical coding—which requires navigating over 70,000 diagnosis codes in the ICD-10-CM alone—is not about prediction. It is a highly complex, hierarchical reasoning task that requires strict adherence to constantly evolving clinical guidelines.
When a human coder is working under massive time pressure, they miss the nuance buried in clinical notes. Corti recently ran its system against Danish patient data and discovered it identified three times as many suicide attempts as had actually been coded by humans. The clinical evidence was sitting right there in the medication records and physician notes, but it never made it into the structured data.
When health systems fail to code accurately, they cannot track disease trends, allocate proper resources, or design preventative interventions. “Medical coding has been treated as a back-office cost center for decades. It isn’t – it’s the data layer that healthcare runs on,” noted Andreas Cleve, CEO and co-founder of Corti.
