What health plans need to know about provider usage of AI

As the name implies, autonomous coding software requires minimal human intervention. It uses AI to assign billing codes to medical services. While autonomous coding may currently be used on a limited scale, as it is further developed questions are arising about its effectiveness and neutrality. If the software is accurate, it could reduce errors and improve the precision of provider coding.  

Autonomous coding software should mean the coding on claims is more likely to be correct when submitted to health plans. As a result, health plans may find fewer opportunities for cost savings through edits or medical record reviews. Given the questions that still surround autonomous coding, health plans are encouraged to continue monitoring codes. 

Providers and revenue cycle management teams are developing AI models to predict when claims might be denied. These models prompt providers to correct claims before submission, with the goal of increasing the likelihood of approval. 

This technology could lead to more accurate claims which should reduce the number of corrections health plans need to make. The potential downside, however, is that coding could be manipulated to avoid denials. Health plans must remain vigilant and watch for potential abuse. 

AI-powered nudging tools, often paired with ambient listening and practice management systems, suggest care options to providers. These prompts encourage providers to ask specific questions or perform certain tests. 

While these tools don’t affect the health plan’s decisions about care appropriateness, they could increase medical spend if providers deliver additional, potentially unwarranted care. Health plans must be aware of this possibility and monitor trends in utilization. 

AI also provides health plans with many tools to handle increased provider AI usage: 

• AI can monitor for behavioral anomalies, such as sudden spikes in appeals volume, drops in payment integrity true positive rates and unusual or templated communications that may indicate AI-generated content.  
• Stylometric analysis of provider communications can help flag automated documentation. 
• Claims pattern analysis can reveal changes in coding practices or increased consistency that suggest AI-driven optimization. 

To prevent losses from provider AI usage, health plans should invest in advanced analytics and machine learning to proactively detect improper payments, fraud and abuse. Automating claims review processes with AI supports accuracy before payments are made, while adapting policies and renegotiating rates can help maintain financial stability in response to more accurate billing.  

Most importantly, proactively educating providers on AI-enabled tools and establishing human oversight frameworks for explainable AI are essential for transparency and fairness. Ultimately, collaboration between health plans and providers will be the crucial component to balance innovation with payment accuracy.