Improving Revenue Cycle Management with True AI and Machine Learning

Artificial Intelligence and Machine Learning are two vast topics of discussion as we work on implementing these tools into a multitude of areas in the healthcare space. So lets narrow that down a bit and focus on Revenue Cycle Management. We reached out to our brilliant Healthcare IT Today Community and asked them – where is true AI and machine learning being used to improve revenue cycle management? The following are their answers.

Steve Roberts, CEO at Vyne
As organizations are confronted by rising costs and staffing shortages, they will increasingly look to artificial intelligence and automation to modernize and optimize their revenue cycle practices and operations. Specifically, providers will expect technology that delivers real-time claims and payment resolution to help them accomplish the objectives of accelerating payments, reducing patient bad debt and write-offs, and furnishing patients with a convenient, transparent consumer experience.

Malinka Walaliyadde, Co-Founder and CEO at AKASA
AI (generative AI, in particular) is dramatically improving revenue cycle management by enabling staff to operate with unprecedented efficiency and comprehensiveness. We’re moving away from a world where people need to spend countless hours combing through medical documents to manually extract key information to put into a prior auth or other encounter. GenAI completes these processes in seconds and provides justifications with direct links to the relevant medical text for critical reliability. Automating tasks like data entry eases the administrative burden on staff, empowering them to operate at the top of their license and focus on higher-value activities. Ultimately, the integration of true AI and machine learning in RCM contributes to greater accuracy, reduced denials, and a better patient experience.

Aasim Saeed, CEO at Amenities Health
I believe the most important and interesting question that revenue cycle managers should be asking is how we can use AI to provide more meaningful transparency into something that patients currently have no visibility into, which is healthcare pricing. It’s a mistake to focus solely on automation and how bills are processed. After all, the goal of any inward-facing technology is cost reduction, which often means lowering headcount. Instead of operational efficiency, we should instead be looking at how to get revenue cycle innovations and advanced technologies to provide some type of direct-to-consumer benefit, such as helping patients know what they are going to pay ahead of time and offering ‘no surprise billing’ guarantees. Health systems still struggle to show their true pricing everywhere within a system. We need to change that dynamic.

Kali Durgampudi, President and CEO, Automation at Apprio
There is scope for machine learning and artificial intelligence (AI) to improve key problems in revenue cycle such as prior authorizations and denials, but those technologies are not being used to their full potential due to myriad challenges such as access to data and interoperability. However, that is the next frontier that must be unlocked due to acute shortage of labor, increasing workloads in the midst of increasing complexity, and interoperability challenges between payers and providers.

Tannus Quatre, PT, MBA, SVP & Chief Business Development Officer, Therapy at Net Health
Industry trends shape how revenue cycle management processes evolve, and artificial intelligence (AI) and machine learning (ML) are two areas that will eventually influence the revenue cycle management (RCM) process. With ongoing regulatory changes being made to coding and billing practices by CMS, manual processes will become even more tedious, with increased room for error for rehab therapy practices of all sizes. This leaves significant room for AI/ML to automate processes and assist providers in their daily operations—saving time and money lost due to mistakes.

Neville Zar, Senior Vice President, Revenue Cycle Management at athenahealth
AI and ML technologies can be leveraged to enhance RCM by automating routine tasks, streamlining critical workflows, and removing administrative burdens and are already making an impact in RCM. Examples of this include automating insurance package selection, improving claims processing by using predictive models to prioritize follow-ups and help physicians get paid faster, and predicting denials to prevent practices from submitting claims that will ultimately be denied. The best RCM work is eliminating the work itself, and if the work cannot be eliminated then it’s about automating the work through AI and machine learning models.

Additionally, AI can play a critical role in making prior authorizations more efficient, a challenging and time-consuming process for practices, that can oftentimes delay care, by using language modeling to predict whether or not a pre-authorization is needed. AI also identifies missing information for prior authorizations, reducing delays and associated costs, and reduces administrative burden for providers.

Rick Stevens, Chief Technology Officer at Vispa
True AI and machine learning are being increasingly integrated into various aspects of healthcare revenue cycle management to streamline processes, improve efficiency, and optimize revenue. While many people confuse AI with automation, true AI/ML uses past data to make predictions of future outcomes. For example, a healthcare organization can now predict future revenue trends, denial trends, and fraud. In the area of claim follow-up, ML models can predict the expected date of payment based on the details of a claim, incorporate an automated claim status check when that expected date has passed, and then only present the claim to a human knowledge worker when a clear exception has occurred. By leveraging ML in ways such as these, organizations can target the most important work at the right time with limited human resources.

Joel Rydbeck, Senior Vice President, Health, Finance & Insurance – Sales at DMI
Reduction of revenue leakage and reduction of denials are top priorities for RCM teams at health organizations. AI is showing promise in its ability to address these draining challenges. By exponentially increasing both the scope of documentation and the depth of detail, AI enables much more comprehensive scanning and attention to and consistency of detail than RCM teams have historically been capable of. This, in turn, means faster, more comprehensive, and more accurate submission of claims, powering an increase in reimbursement amounts and speed. Solutions like Clinithink are enabling Boston Children’s Health, Norwell Health, and many others to make this transformation without increased staffing costs.

John Squeo, Senior Vice President & Market Head, Healthcare Providers at CitiusTech
True AI differs from Ordinary AI as it closely mimics the creative, interpretive, and processing capabilities of human cognitive processes. This includes various forms of content, workflows, and processes such as s natural language understanding of structured and unstructured data, pattern recognition, anomaly detection, and generative tasks such as drafting clinical notes, consumer correspondence, ads, emails, articles, research papers, video descriptions, titles, business plans, and even code software applications. The advanced ability to replicate some of the qualities of human cognitive activities can radically enhance productivity and completeness in some notable areas of the revenue cycle:

1. Predictive Analytics and Denials Management: ML algorithms can analyze vast datasets of historical claims data to predict which claims are likely to be denied by insurers; this allows Healthcare Providers to proactively address potential issues before submission, significantly reducing denials and ensuring faster reimbursements
2. Intelligent Coding and Charge Capture: ML can analyze patient charts, procedures performed, and medical codes to identify optimal coding for each case; this not only improves accuracy but also helps uncover missed charges or under-coding, ensuring Healthcare organizations capture the revenue they are entitled to
3. Automated Claim Review and Payment Posting: ML can automate the review of claims received from Payers; by identifying discrepancies and errors, the system can flag potential issues for manual review, streamlining the process and expediting accurate payments
4. Personalized Patient Communication and Payment Plans: ML can analyze patient data and payment history to predict their ability to pay; this allows Healthcare Providers to offer personalized payment plans and communication strategies, improving patient satisfaction and reducing collection costs
5. Cognitive Document Processing: ML can be used to extract data from complex medical documents like EOBs (Explanation of Benefits) and remittance advice; this eliminates manual data entry, reduces errors, and speeds up the revenue cycle process

It is important to note that True AI in RCM is still evolving. While ML offers significant advantages, it requires robust data sets and ongoing training to function effectively. Healthcare organizations need to ensure they have the necessary infrastructure and expertise to leverage these powerful tools.

Andy Adams, Managing Director, Performance Improvement and Advisory Services at Nordic Consulting
True AI and ML are still underutilized in revenue cycle processes. However, AI is revolutionizing parts of the revenue cycle, such as the accuracy and efficiency of coding and documentation processes via Natural Language Processing (NLP) algorithms that interpret clinical notes with high precision, suggesting appropriate codes and ensuring compliance. Machine learning models also automate or prioritize the review of claims, identifying anomalies and flagging them for further investigation—like denial overturn propensity or recommendations for patient engagement.

The challenge lies in operationalizing AI and ML at the staff level. The history of AI and ML in revenue cycle is filled with examples where the technology’s potential wasn’t fully realized, such as predicted no-show encounters that weren’t acted upon. Success comes from deploying AI and ML into workflows with clear objectives, ensuring staff understand how to use the data to enhance their work and have procedures for handling exceptions or errors. This builds trust in the technology and prevents fallback on old habits.

John Garcia, Chief Product Officer at Janus Health
AI and machine learning are being applied in several impactful ways within RCM, despite varying levels of adoption and AI maturity among health systems and vendors. A prime example is the management of claim denials, which have been a persistent challenge. Historically, the sheer volume of denial data made it difficult to analyze and act upon. However, machine learning algorithms, trained on extensive datasets, can now identify patterns and factors that predict denials. This allows health systems to flag high-risk claims proactively, enabling intervention before denials occur – putting health systems in the driver’s seat.

Steve Albert, Chief Product Officer at R1
In healthcare, the revenue cycle management process holds great potential for automation and simplification by using AI. For example, large language models like ChatGPT have made significant strides and continue to improve in their ability to create structure from pools of unstructured data. Providers can use these tools to generate summaries of text-heavy interactions, such as calls, emails, and account notes, enabling staff formerly devoted to those activities to spend time directly with patients.

Thomas Thatapudi, CIO at AGS Health
Healthcare organizations are using AI in areas such as clinical documentation, patient communication and payments, scheduling, prior authorization, and medical coding. Coding, in fact, has been utilizing true AI and machine learning for about a decade, particularly in computer-assisted coding (CAC). As these CAC applications reach a plateau in coding accuracies of approximately 70-75 percent, newer market entrants are leveraging deep learning models and generative AI to truly increase autonomous coding rates. As a result, I anticipate that coding will be one of the areas within RCM that will be most impacted by true AI, machine learning, and deep learning.

So much to consider here! Huge thank you to Steve Roberts, CEO at Vyne, Malinka Walaliyadde, Co-Founder and CEO at AKASA, Aasim Saeed, CEO at Amenities Health, Kali Durgampudi, President and CEO, Automation at Apprio, Tannus Quatre, PT, MBA, SVP & Chief Business Development Officer, Therapy at Net Health, Neville Zar, Senior Vice President, Revenue Cycle Management at athenahealth, Rick Stevens, Chief Technology Officer at Vispa, Joel Rydbeck, Senior Vice President, Health, Finance & Insurance – Sales at DMI, John Squeo, Senior Vice President & Market Head, Healthcare Providers at CitiusTech, Andy Adams, Managing Director, Performance Improvement and Advisory Services at Nordic Consulting, John Garcia, Chief Product Officer at Janus Health, Steve Albert, Chief Product Officer at R1, and Thomas Thatapudi, CIO at AGS Health for taking the time out of your day to submit a quote in to us! And thank you to all of you for taking the time out of your day to read this article! We could not do this without your support.

Where do you think true AI and machine learning is being used to improve RCM? Let us know either in the comments down below or over on social media. We’d love to hear from all of you!