Improving Health Care Revenue Cycles Through AI

Malinka Walaliyadde, CEO, Co-Founder, AKASA

Can you share a brief overview of your professional history?

My name is Malinka Walaliyadde, and I am the CEO and a co-founder of AKASA, the preeminent provider of generative AI solutions for the health care revenue cycle. Before AKASA, I was a partner at Andreessen Horowitz (a16z), a Silicon Valley venture capital firm with $42B in assets under management. I helped start their health care investment team and worked on almost $250M of investments across 20 health care companies, including providers, payers, biotech, and health technology. 

My co-founders and I started AKASA because we all noticed a common trend: no one was solving the pervasive issues in the back end of health care with modern approaches. There was a lack of new technology in the field, and nothing was being purpose-built. That was the impetus for starting AKASA. 

Can you provide examples of how GenAI tools have been successfully implemented in hospitals to improve the revenue cycle?

Much of the revenue cycle's challenges revolve around the massive amount of unstructured clinical data teams deal with. That data is stored in clinical records, but prior technology approaches built on robotic process automation (RPA), bots, macros, and simpler natural language processing (NLP) simply haven’t been powerful enough to interpret that data in a useful way.

GenAI deeply understands the patient record. It enables software to rapidly and accurately understand large, complex clinical documents (such as chart records) that were previously opaque to computers, comprehend the clinical context, extract information from documents at scale, and use the data in meaningful ways.

Many major revenue cycle tasks are based on the core concept of leveraging patient clinical records in communications with payers. Prior authorization revolves around leveraging clinical records to validate with payers that a procedure is necessary. Coding converts records to a structured format (medical codes) so that payers can interpret, process, and remit payments. Denial management ensures payers do not miss critical clinical details when making decisions and appealing to them when they do. These are all areas where GenAI technology is having a massive influence.

At AKASA, we’re working with leading health systems to use GenAI to improve their biggest revenue cycle challenges, including prior authorization and coding. At one organization, GenAI has helped cut the time staff spend on prior authorization by up to 35 percent. 

What are some common pitfalls to avoid when utilizing AI in managing the health care revenue cycle?

The biggest obstacle to implementing this technology (or any, really) is a lack of trust. People don’t always understand new technologies and are uncertain about their viability in health care. But when it comes to AI, I think the tides are changing. I was recently at a major health care conference, and everyone was hungry to learn more. They recognize the benefits of GenAI and don’t want to get left behind.

When implementing it, however, you want to make sure you’re doing it the right way. Identify your problem areas within the revenue cycle and then find GenAI-powered solutions that can help solve them. What areas of your organization are the most significant bottlenecks? Where do you have the most open headcount? What’s one department that’s struggling right now? 

If you want to see success, look for a collaborative partner who specializes in GenAI and can train and fine-tune a model based on your data. And one who can grow across the revenue cycle with you — across payers and service lines. 

You also want to look for AI that isn’t a black box. For AI adoption to be successful, staff must trust the technology. It doesn’t matter how good the AI is if staff don’t trust or use it. This philosophy is how we’ve built our revenue cycle solutions. Our AI always explains itself in detail to build trust with staff. In our medical coding and prior authorization products, for example, the AI generates justifications and direct quotes for its recommendations and links back to underlying charts. We show our work so teams can see what’s happening and the rationale.  

How important is human involvement in conjunction with AI tools to ensure successful implementation in revenue cycle management?

GenAI is a magical tool that can help us be substantially more efficient and comprehensive. But having humans work together with GenAI is critical. You need a human-in-the-loop model where the technology performs the burdensome, time-consuming tasks, and the human supervises and audits the output. 

Can you discuss the benefits of using a health system's own clinical data to train AI models for revenue cycle improvement?

Most of the open-source large language models (LLMs) out there right now are built on a large corpus of publicly available data. Think ChatGPT, but using generic models like that for revenue cycle tasks often means low-performance metrics and inaccurate outputs. 

The health care revenue cycle is extremely nuanced, and each health system has unique processes. At AKASA, we train models for each of our customers to accommodate their specific data and nuances. This enables us to deliver higher-quality results, be more compliant with health system policies, update the AI more easily, have more control over how it performs, and create something purpose-built for the revenue cycle. 

How do GenAI tools specifically contribute to reducing staff burnout in health care organizations?

Health system leaders constantly say that staffing is their primary challenge. Revenue cycle teams are drowning in work, and hiring is difficult. The only way to actually impact burnout is to do more with less. As a society, we have primarily done that with technology.

A study out of MIT found GenAI increased the productivity of skilled workers by decreasing time-per-task by 40% while also improving the quality of output. It is rare to make things both better and faster — and showing substantial improvements in both is remarkable.

GenAI is simply better than people at certain tasks. For example, a machine can scan more documents faster than a person can, digging for details that are beyond the time constraints that teams face. GenAI can take certain work off staff plates (scanning through a mountain of records to submit a prior auth) and allow them to complete more tasks and focus on more complex needs. Giving teams access to tools that run on this technology empowers them to work better and faster.

We regularly chat with revenue cycle teams that are amazed by what GenAI can do. One veteran staffer said that this technology has been a long time coming and will help them focus on the work they actually want to do.

When was the last time your team was able to clear out their work queue at the end of every day? Imagine the dramatic improvement that would have — not only on productivity, but morale.

In what ways can GenAI tools contribute to improving patient satisfaction and fostering collaboration between payers and providers in the health care industry?

GenAI can improve the health care experience for all — providers and staff, payers, and patients. It’s about improving accuracy, comprehensiveness, and efficiency. This allows people to get back to what’s truly important: helping patients.

Right now, many revenue cycle teams often don’t have time to be proactive on the patient’s behalf. Patients suffer when it comes to high-dollar services or more complicated, lengthier inpatient stays. If RCM teams prioritized working on these cases and engaging insurance as soon as possible, it’d be easier and quicker for systems to understand what will be covered. For patients, it alleviates the burden and decreases the likelihood of bankruptcy stemming from medical care and surprise bills.   

Revenue cycle teams should really be zeroing in on the most challenging elements of the process. If GenAI is leveraged, these tasks can be managed more effectively. It’s shifting the entire industry to focus on the overall patient financial experience. 

There’s lots of talk about GenAI being an arms race between payers and providers. Everyone is jumping in right now, trying to figure out where and how to best make use of this technology. While payers are historically quicker to implement new technology, providers are moving quickly on GenAI. This is an industry-disrupting opportunity that we all need to get behind and take advantage of. We all need to use it responsibly in a way that eliminates friction, reduces costs, and results in better health care for all.