Healthcare Research Program Achieves 100% ICD-10 Code Assignment with Governed Agentic AI

Discover how Aimpoint Digital built a governed agentic AI and AgentOps framework on Databricks to increase ICD-10 auto-coding coverage from 15% to 100%.

Key takeaways
Achieved 100%
ICD-10 code assignment coverage compared to 15% baseline
50%
reduction in manual review time
TECH STACK
Company Logo Icon
Industry
Healthcare
Location
MD, USA
SERVICES
Artificial Intelligence
Artificial Intelligence
Empower your business with pragmatic applications of AI
Data Engineering & Infrastructure
Data Engineering & Infrastructure
Deploy analytics at scale with analytical infrastructure modernization
Product
AgentOps Brickbuilder
AgentOps Brickbuilder
Deploy, manage and observe any agent system on Databricks
TECH STACK
Databricks
MLflow
LangChain

The Challenge

In a large-scale healthcare research program, accurate ICD-10 coding ensures medical conditions are classified consistently across years, enabling reliable cost modeling, disease prevalence tracking, and trend reporting.

With only 15% of new conditions being automatically coded by the existing system, all cases required multiple manual reviews to ensure accuracy, consistency, and alignment. While necessary, this approach constrained scalability, delayed the process, increased operational cost, and introduced risk in maintaining up-to-date ICD-10 standards. The organization needed a solution that could improve coverage and reduce manual labor while maintaining strict governance and oversight.

Our Approach

Aimpoint Digital designed and deployed a governed, agentic AI retrieval solution on Databricks to automate ICD-10 code assignment with accuracy and scalability. By combining intelligent retrieval techniques with a structured agentic workflow, we ensured the system could consistently interpret medical condition descriptions, reference the most current ICD-10 standards, and return reliable, context-aware results.

To accelerate delivery and ensure enterprise readiness, we leveraged Aimpoint's AgentOps Accelerator. Rather than building evaluation, governance, and deployment processes from scratch, we applied a standardized lifecycle framework that embeds automated testing, model validation, human-in-the-loop approvals, and controlled production deployment from the outset. This allowed the team to move from development to production more quickly while maintaining strict oversight, auditability, and compliance with enterprise AI standards.

By using the AgentOps Accelerator, the organization benefited from faster implementation, improved model reliability, and a scalable foundation for future AI initiatives — all without sacrificing governance or control. A streamlined user interface within Databricks provided transparent access to model outputs, enabling business users to trust and operationalize the solution immediately.

Results

RESULT #01
100% Code Assignment Coverage

Automated ICD-10 code assignment increased assignment from 15% to 100% of new medical conditions.

Healthcare Research Program Achieves 100% ICD-10 Code Assignment with Governed Agentic AI
RESULT #02
50% Reduction in Manual Review

Human effort decreased from two coders per condition to one, significantly reducing operational workload.

Healthcare Research Program Achieves 100% ICD-10 Code Assignment with Governed Agentic AI
RESULT #03
Enterprise-Grade AI Governance with a Future-Ready Foundation

The AgentOps framework ensured human-in-the-loop approval for model promotion, supporting compliance and controlled AI lifecycle management and the organization now has a scalable, governed AI architecture capable of supporting additional healthcare research automation initiatives.

Healthcare Research Program Achieves 100% ICD-10 Code Assignment with Governed Agentic AI

Key Takeaways

Aimpoint Digital delivered an enterprise-grade agentic AI framework that increased ICD-10 code assignment coverage from 15% to 100%, ensuring scalable, accurate automation. Simultaneously, a governed and standardized AgentOps framework on Databricks reduced manual coding effort by 50%, establishing a secure, human-in-the-loop AI lifecycle ready for production.

Achieved 100%
ICD-10 code assignment coverage compared to 15% baseline
50%
reduction in manual review time

Let’s talk data.
We’ll bring the solutions.

Whether you need advanced AI solutions, strategic data expertise, or tailored insights, our team is here to help.

Meet an Expert