AI Tackles Healthcare's $36B Problem: Orca Strikes & Billing Codes

NEW YORK, NY – April 07, 2026 – In the vast and intricate world of American healthcare, there is a medical billing code for being struck by an orca (W56.22XA), for burns sustained while one's water-skis are on fire (V91.07XA), and even for documented problems with in-laws (Z63.1). These seemingly absurd classifications are part of the ICD-10, a sprawling system of over 72,000 codes used to document every diagnosis, symptom, and procedure a patient experiences.

While they provide a fascinating, and often humorous, window into the myriad ways human life can go awry, these codes are the bedrock of the healthcare system's administrative functions. They determine how hospitals get paid, whether insurance claims are approved, and how public health trends are tracked. But the sheer complexity of this system has created a multi-billion-dollar crisis.

The System's $36 Billion Glitch

When a healthcare provider selects the wrong code, the consequences ripple across the system. A simple error can lead to a denied insurance claim, forcing hospitals to chase payments and disrupting cash flow. At scale, these mistakes are staggering. According to industry analyses published in outlets like the Journal of the American Medical Informatics Association (JAMIA), medical coding errors cost the U.S. healthcare industry an estimated $36 billion annually.

This figure accounts for lost revenue, the administrative cost of appealing and resubmitting claims, and potential fines for non-compliance. Some studies indicate that the error rate on medical bills can be as high as 75-80%. The American Medical Association (AMA) has estimated that up to 12% of all claims are submitted with inaccuracies, a figure that contributes to significant financial strain and administrative burnout. Even government programs are not immune; Medicare reported approximately $31 billion in improper payments in a single recent year, largely due to coding and documentation mistakes.

These errors stem not from negligence, but from the immense pressure placed on human coders to navigate a labyrinth of ever-changing rules and highly specific options. A coder must accurately translate a physician’s often-ambiguous notes into a precise code, distinguishing, for example, between an injury sustained during tackle football versus flag football, or a bite from a pig versus a peck from a chicken. The system's mandate to account for everything has made it nearly impossible to get everything right.

AI Enters the Administrative Arena

To combat this costly inefficiency, a new wave of technological innovation is sweeping through the healthcare industry. Artificial intelligence, powered by machine learning (ML) and natural language processing (NLP), is being deployed to read, understand, and accurately code clinical documentation at a speed and scale that humans cannot match.

Entering this burgeoning field is Corti Inc., a health-tech company with roots in Copenhagen and New York. The company recently announced its new model, Symphony for Medical Coding, an AI agent designed to assist healthcare professionals in navigating the coding maze. The goal is to reduce errors, streamline billing, and reclaim billions of dollars lost to administrative friction.

Corti joins a competitive and rapidly growing market. The global AI in medical billing sector is projected to expand from roughly $4.7 billion in 2025 to over $45 billion by 2035. Companies like CodaMetrix, Fathom, and Optum are already offering sophisticated AI platforms that analyze patient records and suggest the correct codes, with some claiming to automate up to 90% of the coding workload. These systems work by training on vast datasets of medical records, learning to identify diagnoses and procedures from clinical text and assign the appropriate ICD-10 codes with high accuracy.

Beyond the Bill: Data Integrity and Patient Health

The push for AI-driven accuracy extends far beyond the financial bottom line. The integrity of a patient's medical record hinges on correct coding. An error can attach an incorrect diagnosis to a patient's history, potentially affecting future care, insurance eligibility, and even life insurance premiums. For example, a mistyped code could erroneously label a patient with a chronic condition they do not have, creating a phantom diagnosis that is difficult to remove.

On a larger scale, this data forms the foundation of public health research. By analyzing the frequency and geographic distribution of specific codes, researchers can track disease outbreaks, identify at-risk populations, and measure the effectiveness of public health interventions. The "accidental anthropology" revealed by codes like "contact with a cow" or "prolonged stay in a weightless environment" becomes a powerful tool for understanding how people live and what health risks they face. When this data is clean and accurate, it enables more effective and targeted healthcare policy. When it is riddled with errors, the picture becomes distorted, hampering efforts to improve community health.

The Coder of the Future: Human-AI Collaboration

The rise of AI has understandably raised questions about the future for the tens of thousands of human medical coders in the workforce. However, the consensus among industry experts is that AI will augment, not replace, human expertise. The role of the medical coder is evolving from a production-focused data entry clerk to that of a highly skilled auditor and subject matter expert.

AI excels at handling high-volume, repetitive tasks with speed and precision. This frees human coders to focus on the complex, nuanced cases that algorithms struggle with—such as interpreting a physician's ambiguous handwriting, understanding the context of a complicated multi-symptom diagnosis, or navigating the unique rules of different insurance payers. The future role will involve reviewing AI-generated codes, managing exceptions, ensuring regulatory compliance, and providing feedback to improve the AI models themselves.

This symbiotic relationship leverages the best of both worlds: the analytical power of machines and the critical thinking and contextual awareness of human professionals. As technology handles the routine work, coders will be empowered to operate at the top of their license, focusing on clinical documentation improvement and ensuring the ethical application of codes. The demand for skilled professionals who can work alongside AI is expected to grow, requiring a new emphasis on data analysis, technology proficiency, and specialized knowledge. The goal is not to remove humans from the loop, but to equip them with more powerful tools to manage an increasingly complex system.