Improving The Healthcare Revenue Cycle With AI And RPA
Health care billing with stethoscope, bottle of medicine.
Imagine that you are the CEO of a large healthcare company and wondering what to do with AI. You’ve heard of some of the fascinating findings from various research laboratories on how AI can match or surpass human doctors in diagnosing cancer, retinal disease, and even Covid-19. You are drooling a bit on the dollars you may bring in from funding organizations and wealthy donors who want to be associated with these sexy developments.
But then break out of that dream and focus on the true value of AI for your hospitals. First of all, you know that what works in the laboratory doesn’t always work at the bedside. Second, remember that the FDA has to approve these AI treatments, and that can take years. The final lesson that will ruin your dream is that any of these AI diagnostic features are unlikely to cut your costs. You will likely still need the same number of human radiologists, ophthalmologists, and so on.
Finally, what you really need, comes to mind, is AI to cut administrative costs. You remember a conference at a beautiful resort where Professor David Cutler, a renowned Harvard health economist, spoke. He said that administrative costs in the US account for a quarter to a third of total health care spending. He came up with several ideas for reducing the costs associated with the revenue cycle, prior approval, and data clearinghouses. The CEO wasn’t sure, but believed the professor had mentioned that AI could help with some of these approaches.
This CEO’s sudden insight into administrative AI was successfully implemented at Baylor Scott & White Health, a large academic medical system of 52 hospitals that is the largest nonprofit in Texas. Sarah Knodel, the senior vice president of Revenue Cycle there, is not the CEO, but she oversees many administrative functions and has 2,500 employees reporting to her.
- Reduce your collection costs
- Optimize net sales
- Enhance the patient’s financial experience
Knodel and BSWHealth have been working on an approach for these areas for eight years. With a focus on improving price transparency for its patients, BSWH implemented an automated, machine learning-based price estimation tool from Waystar, a healthcare technology provider. The tool makes estimates of patients’ own costs before receiving treatment. While this may seem normal in many industries, it is very unusual in healthcare to be able to accurately estimate costs in advance. Before the tool was implemented, making the estimates was a very manual process that combined different information from numerous systems at BSWHealth. It took a sales cycle employee 5 to 7 minutes to produce a limited accuracy estimate. However, now 70% of the estimates are calculated without human contact. The system automatically pulls real-time eligibility and benefits data from the patient’s health insurance company and combines this with fees and contracted rates to produce an estimate of expenses that are unique to a particular patient. The technology collects and learns from insurance claims to improve the accuracy of the estimates over time.
No outside organizations rate hospitals based on patients’ financial experience, but BSWHealth has had positive feedback since the tool was implemented. Payment options are discussed prior to treatment, resulting in 60-100% improvements in point-of-service collection in various clinics and hospital departments. Doctors are also happy to receive a preliminary assessment, as it leads to fewer abortions on the day the service is provided. Five years ago, Knodel recommended making the appraisal system available in an online, self-service format so that price buyers can get their own appraisals when assessing where to get treatment. Last year the US government made such forecasts mandatory for care, but BSWHealth was way ahead of the requirement (and many hospitals are not yet meeting them).
In addition, BSWHealth uses intelligent technologies for “damage recording” in the insurance collection department of the office in order to automate the process of checking the status of outstanding insurance claims. In the past, a human collector had to log into or go to multiple payer websites. Now Robotic Process Automation (RPA) and screen scraping technology mimics the user who logs into the payer’s website. Because the RPA system receives the claim status from the payer, the data is integrated into the collector’s workflow so that it never ends up in the collector’s worklist if it is accepted and scheduled for payment. Conversely, accounts that are rejected and require immediate action are expedited for review. The Status RPA results in an exception-based workflow where only accounts that really require human intervention are submitted for review by a collector.
Sarah Knodel said her organization runs many projects of this type and uses machine learning, or RPA, in almost every department in the sales cycle. In areas such as usage monitoring, a new technology reads the documentation of the medical record in real time and predicts whether a patient should be in the inpatient or the observation status in order to ensure compliance with regulatory and chargeable requirements. As a result of these efforts, BSWHealth has reduced the number of full-time positions in the occupancy checking department by over 20%, while the number of refusals to pay has been reduced by the same percentage. Looking ahead, Knodel’s goal is to leverage these technologies to develop more collaborative and innovative partnerships with payers. It hopes to eliminate the time-consuming and inefficient back-and-forth of treatment approvals and appeals in favor of automation and efficiency.
The Waystar perspective for a smarter sales cycle
To understand what is happening across the broader industry to AI and RPA to lower the revenue cycle and administrative costs, I spoke with Matt Hawkins, CEO of Waystar. This company is also a good example of how transaction processors can create new offerings from their data depletion. He said Waystar processes 2.5 billion healthcare billing and collections transactions annually for approximately 40% of US patients, all on a single data platform. This data enables the company to use machine learning, other forms of AI, and RPA to take costs out of the system and use them to improve patient care.
No one has ever blamed the U.S. healthcare system for being simple, and that is reflected in the data on it. Waystar works with 500,000 service providers, 1,000 health care providers and 5,000 payers and health insurers. With all of these sources, Waystar provides a clearinghouse for billing and collections data, but to analyze it and apply AI, they need to use rule modules to pull, match, standardize, merge and transform the data. After that, they can use all sorts of algorithms, such as the one, to predict how much a procedure will cost at will in one of the major healthcare systems with their technology. Their AI and RPA platform is called “Hubble”. It does not allow a glimpse into other galaxies, but a glimpse into the future of your insurance claims.
Some Waystar offerings offer propensity modeling of one type or another – predicting the likelihood that a claim will be judged correct by the payer and paid, predicting whether or not a patient will be able to pay their bill, and so on Learn.
Waystar uses RPA to automate things like the status of claims, appeals against rejection of claims, and other administrative processes that involve reviewing and communicating data from one system or another. You have robots pinging or calling payer companies, and increasingly, payer companies are letting robots process the response. Soon we humans will mainly be watching robots talk to each other.
Perhaps it may be tempting for our hypothetical CEO to see the machine learning projects in action at Baylor Scott & White Health and the AI and RPA applications that Waystar is offering with Revenue Cycle trying to figure out what to do with AI technology is to be done. It seems unlikely that a large donor would sponsor a large initiative in this area, but then again, the work on the shooting cycle would likely more than pay off in a short amount of time.