Pulmonology Coding Alert

Remember to review the tech’s code choices.

Revenue cycle teams handle thousands of medical claims each year to ensure their providers receive accurate reimbursement. However, factors beyond a manager’s control can affect an organization’s workflow, which could lead to an accumulation of unprocessed claims. Luckily, you can use artificial intelligence (AI) to tackle this backlog.

Read on to learn three helpful tips on using AI to efficiently move through your buildup of medical claims.

Tip 1: Identify How Your Unprocessed Claims Accumulated

Backlogged claims have a domino effect on healthcare organizations, providers, and patients, as well as the revenue cycle management (RCM) practice. Unprocessed medical

claims can build up in an RCM organization’s workflows over time and for several reasons.

• Staffing shortage: When your practice has fewer employees than it needs to operate at full capacity, efficiency can slow down, which can cause claim backlogs to grow.
• Patient numbers: Claims can pile up when providers are seeing more patients than they normally would, such as during the COVID-19 public health emergency (PHE) or during cold and flu season.
• Network outage: As exhibited by the Change Healthcare cybersecurity attack, unprocessed claims will build up when the coding and billing services are unable to contact payers and clearinghouses.

When unprocessed claims build up in your practice’s workflow, you won’t receive reimbursement in a timely fashion. This will affect your bottom line and practice expenses.

By identifying how your unprocessed claims accumulated, you can plan accordingly how your practice is going to tackle the backlog. Claims that have more complex cases may be better served by your senior staff while the simpler cases could be handled efficiently by an AI system.

“AI use is abundant in healthcare facilities, whether explicit or discreet. AI is making headways in revenue cycle management, such as with prior authorizations, charity care screening, and claim status verification,” says Jessica Miller, MHA, CPC, CPC-I, subject matter expert at MediCodio in San Ramon, California.

Tip 2: Delegate Certain Claims to AI

After understanding how the claims built up, you’ll want to explore which ones would be the best assigned to your AI system. Using AI technology, such as machine learning (ML) or robotic process automation (RPA), can help streamline your RCM workflow. These technologies can swiftly move through simple cases or assess claim denials, so human coders can give more attention to the cases and claims that require it.

Assigning the simpler cases to AI will also help speed up the reimbursement process for any missing revenue.

For example, say your backlog contains 100 claims that would receive $200 apiece in reimbursement. The AI system can evaluate the claims, recommend codes for reporting, and require a simple quality assurance (QA) review before submission — possibly handled within a couple of days. If all 100 claims are approved for reimbursement, your practice will receive $20,000 with minimal effort.

Similar to assessing the strengths and skills of your human employees, you can use AI as a tool to enhance your practice’s efficiency in tackling your claims backlog.

Tip 3: Review AI’s Code Selections

QA is vital after AI completes its code recommendations. Human coders must review the suggestions before submitting the claim for reimbursement. Their knowledge, subjectivity, and critical thinking skills are crucial to making sure the AI system suggested the codes based on the medical report it scanned. By reviewing the AI’s code choices, you’ll ensure the codes accurately represent the services provided, the patient’s diagnoses, and any supplies used. At the same time, you’ll also provide useful feedback to the technology.

“There has to be an auto feedback loop in an AI tool that continuously learns from coder actions and improves the accuracy of the AI tool on an ongoing basis,” says Devanshu Yadav, vice president of strategy at Raapid Inc. in Louisville, Kentucky.

Even though AI reviews and generates outputs quickly, humans will consistently be needed for claims processing. The technology is helpful but should be used as a tool to assist revenue cycle teams rather than act autonomously.

“Regardless of AI’s abilities, human intervention will always be necessary to verify and validate the AI-generated information,” Miller says.