AI Breakthrough Enhances Medical Coding Accuracy, Outperforming Physicians

AI Breakthrough in Medical Coding Accuracy

Researchers at the Mount Sinai Health System have developed a novel approach to improve the accuracy of artificial intelligence (AI) in assigning medical diagnostic codes. This simple yet effective tweak could potentially outperform physicians in coding accuracy, reducing paperwork time for doctors, minimizing billing errors, and enhancing the quality of patient records 1

1

2

The "Lookup-Before-Coding" Method

The study, published in NEJM AI on September 25, 2025, introduces a "lookup-before-coding" method. This approach first prompts the AI to describe a diagnosis in plain language and then select the most appropriate code from a list of real-world examples. Dr. Eyal Klang, Chief of Generative AI at Mount Sinai, explains, "We gave the model a chance to reflect and review similar past cases. That small change made a big difference" 1

1

Study Methodology and Results

The research team utilized 500 Emergency Department patient visits from Mount Sinai Health System hospitals. They tested nine different AI models, including small open-source systems, using the following process:

1. AI models generated initial ICD diagnostic descriptions based on physician notes.
2. A retrieval method matched each description to 10 similar ICD descriptions from a database of over 1 million hospital records.
3. The model then used this retrieved information to select the most accurate ICD description and code 2

   2

   .

The results were impressive: models using the retrieval step outperformed those without it and even surpassed physician-assigned codes in many cases. Surprisingly, even small open-source models performed well when allowed to "look up" examples 1

1

Related Stories

Revolutionary AI System InfEHR Transforms Medical Diagnosis Through Advanced Pattern Recognition

Mount Sinai Study Reveals Cost-Effective AI Strategy for Healthcare Systems

AI Models Stumble on Medical Ethics Puzzles, Revealing Cognitive Biases

AI Diagnosis Outperforms Emergency Room Doctors in Harvard Study, Raises Questions on Patient Care

Implications for Healthcare

Dr. Girish N. Nadkarni, Chair of the Windreich Department of Artificial Intelligence and Human Health at Mount Sinai, emphasizes that this approach is about "smarter support, not automation for automation's sake." The potential benefits include:

1. Reduced time spent on coding by physicians
2. Fewer billing errors
3. Improved quality of patient data
4. More time for direct patient care 2

   2

Future Directions

While the retrieval-enhanced method is not yet approved for billing and was tested specifically on primary diagnosis codes from emergency visits discharged home, it shows promising potential for clinical use. The researchers are now integrating the method into Mount Sinai's electronic health records system for pilot testing and plan to expand it to other clinical settings and include secondary and procedural codes in future versions 1

1

Dr. David L. Reich, Chief Clinical Officer of the Mount Sinai Health System, highlights the broader implications: "The big picture here is AI's potential to transform how we care for patients. When technology relieves the administrative burden of our physicians and other providers, they have more time for direct patient care" 2

2

This innovative approach to AI-assisted medical coding represents a significant step forward in leveraging technology to improve healthcare efficiency and quality, potentially benefiting patients, clinicians, and health systems alike.