Three Essential Strategies for Coding Excellence in the Era of Artificial Intelligence

As healthcare becomes increasingly data-driven, traditional health information (HI) roles are undergoing significant transformation—none more so than clinical coding. Once limited to health statistics, clinical coding evolved to also support reimbursement following the introduction of diagnostic related groups (DRGs) in 1983.

With the volume of global healthcare data growing significantly, coding accuracy has also become both more difficult and more essential to ensure accurate, meaningful, and effective medical decision-making. In fact, RBC Global Markets says that 30 percent of the world’s data volume is being generated by the health industry.

Yet HI leaders continue to face persistent obstacles in the clinical coding process, including workforce shortages, variable documentation practices, and inconsistent payer requirements.

While emerging technologies offer potential solutions, the broader imperative is clear: we must re-engineer the revenue cycle—clinical coding included—for long-term precision and efficiency.

Here are three strategies to successfully enter the new era of clinical coding performance while also acknowledging the pressures we face today.

Strategy #1: Prioritize the Human Element through Communication, Education, and Oversight

Effectively implementing the next generation of clinical coding practices hinges on integrating emerging artificial intelligence (AI) technologies with the domain expertise of seasoned coding professionals. Clinical coding teams must be prepared for an AI-integrated future.

However, implementation of this vision is far from straightforward. Human oversight remains essential, and the complexity of real-world coding environments demands a careful, evidence-informed approach to change management.

HI leaders must foster a culture that values both human judgment and algorithmic assistance. This starts with structured, data-driven discussions where coding staff can raise concerns, propose solutions, and question human and machine errors. Misinterpretation—by coders or AI—can be equally damaging.

To position AI as a support tool, not a threat, organizations must invest in staff development and promote a workplace where precision, compliance, and workflow efficiency are the shared goals. Continual professional development, including upskilling and reskilling, is no longer optional. It is mandatory. The following practices are critical:

• Define essential AI-era skills: Ensure coders stay fluent in core systems and regulations while equipping them to adopt emerging tools. These skills include understanding the differences between technologies such as ML, AI, NLP, and RPM. Coders should be able to recognize AI-generated bias, discrepancies, and inconsistencies.
• Strengthen revenue cycle literacy: Train coders to fully grasp the details of reimbursement, analytics, and compliance to broaden their strategic impact.
• Include coders in automation planning: Involve them early in tool evaluation and rollout to capture practical insights.
• Encourage ongoing industry engagement: Support professional learning through networking, forums, and credible nontraditional sources.

Strong human oversight, especially through audits, also remains critical. AI can inadvertently drive upcoding or downcoding without proper checks in place. Organizations must build or outsource audit expertise as AI enters the workflow. Consider the following guidance:

• Interrogate audit findings more deeply. Look more closely at errors and inaccuracies revealed in ongoing baseline audits and conduct targeted audits to uncover new opportunities for improvement. Routine audits may show a 95 percent accuracy rate in 1-2 percent of cases, but a targeted audit may reveal that 30-40 percent of the cases reviewed include some type of error leading to increased denial or audit risk. Targeted coding audits may include such topics as misclassifications of diagnosis or procedures, patterns of higher- or lower-level E&M codes, and context rules such as “history of” or “ruled out.” Analyzing how often coders override AI-suggested codes is another new area for coding audit consideration.
• Use audits to fix systems, not blame people. Errors often stem from flawed documentation or inefficient workflows, not coder incompetence. Instead of pointing fingers, focus on uncovering what is broken. Is the clinical documentation incomplete and the coder must make assumptions? What in the process may be causing the error? Is the proper documentation located in the wrong place? Sometimes reeducating the coder is not the issue. It could be the process itself.
• Target audits wisely. Use second-level reviews, the Office of Inspector General (OIG) target list, and modifier usage audits to zero in on high-risk areas. For example, determine where in the process the modifiers are applied (coders, billers, or clinicians), or if modifiers are hard coded into systems where there is a higher chance for errors.

Technology expedites the audit process. Systems can outperform human auditors in terms of speed and scope. However, technology does not outperform human judgment or clinical nuance. For example, a recent article in the Journal of American Medical Association’s Health Forum points out many systems’ inability to “interpret clinical nuance, contextual keys, patient preferences, and SDOH that influence care decisions.” Training is needed not only to use AI, but also to question the output.

HI leaders are encouraged to use technology to identify trends and points at risk, but tap human expertise for the evaluation steps—validate errors, uncover how the errors occurred, and establish a process for error resolution.  

Finally, coding is one of many interdependent steps in the revenue cycle. Weaknesses in one area ripple across the entire process. Ongoing education, open communication, and smart auditing help keep every team aligned and accurate.

Strategy #2: Build Your Data Analytics Prowess

The sheer volume of health data is rapidly exploding, and data-driven decision-making is a key skillset needed for the future in nearly every industry. Simply having more data is not useful. We need the ability to interpret that data to make accurate, meaningful, and effective decisions.

As provider organizations continue to amass large quantities of health data, HI professionals are uniquely positioned to bridge the communication gaps between revenue cycle and IT teams that often don’t speak the same professional language. Our skills in interdisciplinary collaboration are a strategic asset here.

Using expertise in data governance, patient safety, and ethical analytics, HI professionals can help guide responsible use of dashboards and metrics. Agility and commitment to ongoing learning are essential in this evolving landscape. AHIMA provides several specific training and credential programs.

Key competencies to develop include:

• Proficiency in data visualization tools
• Understanding of statistical and predictive analytics
• Ability to interpret complex datasets to inform policy and strategy
• Familiarity with interoperability standards and system integration
• Active participation in solution testing and validation with IT and services teams

Specifically related to clinical coding, HI should inform the machine learning process and consistently test system data. Analysis of outputs is important to ensure revenue, safety, and quality are not veering up or down in an unexplained or unexpected way. And if that happens, coders’ knowledge and experience can quickly identify the cause of deviations before they become detrimental.

Strategy #3: Recognize the Double-Edged Nature of AI

Legacy systems were never built for many of the coding and revenue complexities we face today. New technology tools are essential. However, new systems that are not kept in check can steer an organization off course. Human intervention, common sense interpretation, and continual testing are mandatory.

Another consideration for new automation tools is the complexity of coding in acute care settings. There are various case types that don’t fit AI coding models, and context is a tricky concept for AI systems to understand. This is especially true when clinical documentation is ambiguous or comes from disparate source systems.

Furthermore, AI cannot replicate ethical reasoning or replace the nuanced conversations among coders, clinical documentation improvement (CDI) specialists, clinicians, and auditors. These discussions remain critical for resolving discrepancies in coding, billing, and denials.

HI leaders avoid some of these downsides by setting realistic expectations about AI capabilities and limitations. Misplaced assumptions often result in costly errors. Proper alignment of staff roles with technology’s actual strengths fosters trust and effective collaboration.

When considering technology, ask vendors to provide explainable outputs and clear rationale resulting from the solution. As part of the implementation strategy, test outcomes to ensure expected results from the beginning and service by service.

Finally, HI professionals fear loss of work when technology promises new efficiencies. The reality is that technology-generated efficiencies provide an opportunity to reimagine jobs and roles as the nature of the work changes—but it doesn’t have to lead to loss of HI jobs. Jobs shift as new opportunities open for coding professionals to lead in oversight, strategy, and system optimization.

---

Stacey Sexton, RHIA, is Vice President at TruBridge, an Alabama-based healthcare solutions company that works with more than 1,500 healthcare organizations.