Customers, Partners, and Friends -
Last month we held an all-hands meeting on Digitize.AI’s 2019 roadmap. I wrote this on the whiteboard to spark the discussion: "In 2018 we brought A.I. to the healthcare revenue cycle—what did we learn?"
What came from this was a series of retrospective and future-facing insights that we realized later would be valuable to share with others. I’ve bulleted the key takeaways below, and would love to hear thoughts from our network (you can email me at email@example.com)
Best of luck to you all in 2019.
- Pat Morrell
In 2018 we brought A.I. to the healthcare revenue cycle—what did we learn?
- ROI > Potential. Our customers, CFOs and Chief Medical Officers, care about practicality, cost-savings, and outcomes (whether in the revenue cycle or the operating room). Although much of the industry talk about A.I. is headline-catching, A.I.’s “potential” is irrelevant to leaders solving near-term problems—for us, and our customers, the “job to be done” and the “ROI” are paramount. As technologists here at Digitize.AI, we need to see 10 years ahead and build solutions for today’s use (both of which we do).
- Silver bullets… aren’t. Given the variability/breadth/complexity of RCM processes and rules, an A.I. solution that claims to automate the ENTIRETY of revenue cycle processes and challenges as an “all-in-one” silver bullet is impractical and non-credible. The best way to create real and immediate value in the revenue cycle with A.I. is to initially focus on one core problem and build best-in-class models and expertise around measurable results. For our company, that one core problem is prior authorization.
- Insight is in the trenches. To really understand that core prior auth problem (and how A.I. can solve it), learn how it’s handled today. We spent 2018 embedded with the revenue cycle team of a top 10 academic medical institution—what we learned informed many of the insights you see below.
- The real crisis. Prior auth is not just a financial issue—it’s a patient treatment crisis, and it’s growing.
Prior auth is expensive: docs spent 20 hours/week prior auth tasks; avg cost $83k/year/docPrior auth creates treatment crises: 92% of docs say prior auths negatively impact treatmentPrior auth is growing: volumes are up 54% in 4 years
- Data, not process. Most providers mischaracterize prior auth as a “process problem" and throw bodies and workflows at the issue. But prior auth is a “data problem”—the delays, denials, miscommunication, and costs most commonly result from providers failing to anticipate and meet payers’ data demands.
- An untapped ace-in-the-hole. But, we at Digitize.AI know that Providers have the data to level the playing field—most just lack the pragmatic technical vehicle to employ it in the right way to improve authorization rates, lower denials, and level the playing field.
- Dream-scenario A.I. prior auth machine. We’re an interdisciplinary team of data scientists, software developers, business strategists, and entrepreneurs. We’ve spent the past year with development partners and cross-functional leaders across the healthcare industry to zero-in on what a best-in-class solution should deliver to be seen as valuable. To meet our provider customers’ prior auth needs, that solution must do the following:
- work 24/7, behind-the-scenes (no new interfaces/tools for teams, clinicians)
- execute flawlessly (no data errors that cause denials)
- execute in real-time (no batch upload delays)
- integrate with existing workflows (no change management) and EHRs
- deliver financial results (400%+ ROI) and remove clinical burdens (less burnout)
- anticipate payer behavior (predict auth/denial likelihood, timetable for auth, etc)
- deliver real-time reports on payer trends (extract insights valuable to negotiations)
- show ROI i.e. fast authorizations, higher auth rates, and fewer denials (at less cost)
- grow “smarter” over time i.e. use predictive intelligence to continually improve ROI
(We built “Lia”—our prior auth “system of intelligence"—based on this dream-scenario solution. If you haven’t already, take a look at this video on our website to see Lia in action).
- Predictions for 2019. Based on what we learned in 2018, here’s what we expect in the coming year:
- The “old playbook” becomes increasingly obsolete in the revenue cycle. Historically revenue cycle leaders and CFOs could solve problems like prior authorization by adding FTEs, outsourcing to third parties, or retooling workflows. Those legacy levers are losing their efficacy, and as we see more CFOs capping internal headcounts and trimming operating budgets—while simultaneously increasing clinical volumes (perhaps via acquisition)—revenue cycle leaders will need alternative means of automated execution (i.e. A.I.).
- Increasing openness from EHR vendors to A.I. partners to solve problems like prior auth. We see this already (e.g. Epic’s App Orchard), and believe the trend will increase given the pervasiveness and recognition of the issues surrounding prior auths.
- Ongoing rule changes and enhanced requirements from payers. No surprise here. The variability inherent to prior authorization sequences is unlikely to change, but an adaptable and predictable A.I. asset can help providers anticipate changes and evolve to protect patients and revenues.
- The potential for increased clinician burn-out. Unfortunately, as the prior auth growth trend continues, we anticipate the potential for more and more clinicians to burn-out due in part to the concentration of administrative tasks that fall on them and their support staff.
- Wider A.I. adoption in healthcare (back office moreso than clinical space). As comfort with new technologies like Lia grows, and as solutions continue to show results like Lia’s, the financial benefits and understanding will far outweigh the change aversion that comes with any technological wave. This is corroborated by recent reports that project the annualized spend on A.I. solutions in healthcare to grow 10X (or more). Of course, the first-movers will be the ones to reap the windfall benefits.
- 50-70%+ of prior auths can be handled by A.I. (and growing). Lia is already operating within this range, and can take on increasingly complex cases as our A.I. models continue to evolve. The immediate productivity and accuracy benefits are compelling, and will continue to improve through 2019. And by empowering tools like Lia to take this workload off of existing teams’ plates, those team members are liberated to focus on more value-added and patient-supportive tasks.