Artificial Intelligence

How leveraging artificial intelligence in utilization management can enhance your revenue cycle

July 27, 2023 1:38 pm

Today’s healthcare organizations strive to provide high-quality care that promotes optimal patient outcomes. It starts with ensuring the right level of care — inpatient, outpatient or observation. However, challenges associated with applying correct patient status are twofold. First, there’s considerable subjectivity in physician or case manager application of traditional inpatient admission criteria. The culprit? Lack of real-time data insights into each
patient’s ever-evolving clinical picture. Second, payers use clinical criteria unique to each organization. This lack of standardization leads to time-consuming, manual workflows. Together these challenges create a perfect storm for increased medical necessity denials that jeopardize the long-term financial sustainability of today’s healthcare organizations.

Rethinking denial prevention

A recent survey from the American Hospital Association (AHA) found that 89% of respondents experienced an increase in payment denials between 2017 and 2020, with half (51%) experiencing a ‘significant’ increase in denials. The AHA report outlining the survey results also stated, “Hospitals and health systems report a steep increase in short stay denials, even when clinical indicators and the severity of illness meet the standards for inpatient admission.”

In 2021 alone, 69% of healthcare leaders said they saw an increase in denials with an average increase of 17%, according to a recent poll conducted by the Medical Group Management Association (MGMA).

The denial rate has been particularly noticeable among Medicare Advantage plans. A recent industry survey found that the initial inpatient level of care claim denial rate for Medicare Advantage plans was 5.8% compared with 3.7% for all other payer categories. Hospitals participating in the survey wrote off 3.6% of their inpatient revenue as uncollectible in 2021. That percentage jumped to 5.9% in 2022.

“Throughout the industry, there’s a growing concern about the steady increase of denials,” said Tanya Sanderson, senior director of Revenue Integrity at XSOLIS, a company leveraging machine learning and artificial intelligence (AI) to generate real-time predictions that help case managers assign appropriate patient status. “However, it’s also an opportunity to change the trajectory of healthcare spending and be good stewards of financial resources while simultaneously ensuring hospitals receive appropriate reimbursement.” Changing that trajectory requires partnership between providers and payers. Both entities must align level of care with appropriate reimbursement while focusing on patient and member experience and outcomes. Physicians and case managers must also be able to quickly, accurately and consistently determine patient status based on real-time data (e.g., labs, vital signs, physician orders and clinical documentation) — not isolated diagnosis codes or the patient’s initial presentation that often changes over time. In addition, health plans and providers must align on medical necessity for inpatient level of care, and length of stay (LOS) using a transparent approach.

That’s where machine learning, an application of AI, can help. More specifically, machine learning continuously evaluates each patient’s profile against a vast database of situationally relevant cases. By recognizing clinical patterns, the AI generates highly precise predictive analytics that facilitate the determination of the most appropriate level of care and care setting.

“It’s about taking the true clinical picture into consideration,” said Sanderson. “That includes the patient acuity, anticipated level of services needed, medical conditions, risk of adverse events and more. AI brings in all this information to support the assignment of patient status and ensure appropriate reimbursement for the level of care provided.”

The cost of incorrect patient status

Once a patient is in a hospital bed, hospitals lose money if they don’t assign the appropriate patient status. The difference in payment between inpatient and observation claims can range from $1,500 to $5,000 or more per claim, depending on the diagnosis. Similarly, the diagnosis also affects the cost of care and the level of care provided. When extrapolated across thousands of claims, not assigning the correct patient status could result in significant financial loss. Recent XSOLIS’ analyses for one healthcare client revealed a conservative estimate of $68 million in lost reimbursement over a two-year period due to self-denied or inappropriately downgraded cases that were technically appropriate inpatient admissions based on their length of stay and other AI-driven clinical insights.

Ensuring revenue integrity, preventing avoidable denials

Subjective application of inpatient admission screening criteria had been a root cause of medical necessity denials at Hackensack Meridian Health, a 10-hospital integrated healthcare network in New Jersey. By taking a data- driven approach to case management, the organization was able to transform its processes and generate better outcomes.

“AI helps me evaluate cases quickly and update those cases in a timely and appropriate manner, so I don’t miss opportunities,” said Cinthia Neumany, RN, BSN, MAS, senior vice president of Clinical Operations and Population Health at Hackensack. “If I review a patient for observation, AI tells me if it looks like it’s moving into the inpatient realm. This allows me to review the case again and not lose a day to update a case that was appropriately inpatient status. It also helps me know when to shift cases to observation status before the bill goes out the door, so we don’t get denied for an inappropriate admission.”

“At its core, AI promotes revenue integrity and appropriate reimbursement,” said Sanderson. “Revenue integrity is proactively aligning providers and payers with the right level of care at the right time for the right reimbursement. An AI-driven approach is important.”

Armed with more data insights, for example, case managers can also approach physicians proactively to ensure thorough clinical documentation — again with the goal of safeguarding appropriate revenue.

AI also helps healthcare organizations track patient progress in real time and in the context of the optimal geometric mean LOS.

“This helps us know which patients are appropriately moving toward discharge,” said Neumany. “We use the anticipated discharge date as part of our multidisciplinary rounds, and the care coordinators use it as a target when working with patients and their families.”


The percentage of healthcare leaders, in 2021 alone, who said they saw an increase in denials with an average increase of 17%, according to the Medical Group Management Association

Why hospitals serving large Medicaid populations are at risk

Inpatient days affect whether disproportionate share hospitals (DSH) (i.e., hospitals that serve a large number of Medicaid patients) receive additional funding. When disproportionate share hospitals inappropriately shift cases from inpatient to observation status, they not only lose the reimbursement for the appropriate care provided, but they also negatively impact their DSH percentages. A recent XSLOIS analysis for one DSH organization revealed 3,000 potential missed Medicaid days over a two-year period due to incorrect patient status for cases that were technically appropriate inpatient admissions based on their length of stay and other AI-driven clinical insights.

Anticipating payer trends

Before engaging with AI-driven solutions, healthcare organizations have been limited to ad hoc remittance advice from each payer in their network to better understand that payer’s unique clinical criteria. However, AI can surface patterns to better understand trends across a provider’s full payer network — more accurately and thoroughly predicting future actions.

“It’s about trying to figure out each payer’s unique clinical criteria to better understand denials,” said Neumany. “We’re trying to predict those trends and address them proactively on the front end. Some denials are trending downward based on how we’re approaching payers and the strategies we’re putting in place.”

How XSOLIS’ CORTEX.UR promotes revenue integrity

The XSOLIS Care Level ScoreTM (CLS), a component of CORTEX.UR, is powered by XSOLIS’ predictive analytics and machine learning that continuously monitors the full spectrum of patient data directly from the electronic health record (EHR) and assesses patients in real time to identify potential underpayment or denial scenarios. This offers providers and payers consistent, real-time clinical insights and bidirectional communication in an EHR-agnostic tool to ensure appropriate reimbursement in the most efficient manner.

Don’t take it from us. Take it from XSOLIS’ customers who have achieved:

  • 25% improvement in reduced extended observation rates (i.e., patients who are discharged in observation status who stay longer than two midnights)
  • 36% reduction in medical necessity denials
  • 37% improvement in appropriate observation to inpatient conversion rate
  • 67% increased case review volume
  • 75% reduction in inaccurate patient status decisions

Promoting workforce optimization

Being able to identify patients tracking toward discharge in the next 24 hours also makes it easier to plan from a staffing perspective.

“It helps with our weekend workflows in particular,” said Neumany. “As with most organizations, we don’t have full staff on weekend days.”

Healthcare organizations can also leverage robust insights and reporting to maintain optimal revenue cycle operations, said Sanderson.

“From charge capture to final payment resolution, AI is changing how we do business,” she added. “Whether it’s identifying a change in patient acuity, or helping capture appropriate coding, AI can augment and automate certain functions to help staff focus on the complex issues that need attention the most. Leveraging the AI-driven analytics via dashboards and reporting can also help revenue cycle leadership identify opportunities for performance improvement such as shifting resources further upstream to prevent further rework downstream. This means organizations can also prioritize appeals and focus revenue cycle staff efforts on claims with the highest propensity for appropriate inpatient reimbursement.”

At Hackensack, AI increases productivity for three groups:

  • Utilization review (UR) specialists
  • Physician advisors (PA)
  • Revenue cycle staff

UR specialists spend less time reviewing cases and more time at the patient’s bedside. The same is true for PAs who only review cases by exception when necessary so they can focus on patient care or other clinically related denial prevention initiatives. For example, PAs only review cases of atrial fibrillation, chest pain, shortness of breath and abdominal pain when AI suggests an inpatient admission is highly likely. Finally, fewer back-end denials mean revenue cycle staff can focus on proactive denial prevention. When denials do occur, revenue cycle staff can prioritize denials based on predictive analytics — specifically whether cases were more or less likely to have truly warranted an appropriate inpatient admission.

Improving payer relations, fostering mutually beneficial cost savings

The resulting data transparency and insights also help healthcare organizations open lines of communication with payers.

 “We’ve been using AI to track the most consistent denial trends by payer,” said Neumany. “We then send this information back to our contracting team.”

Organizations can also use AI to identify opportunities for automation and administrative cost reduction. XSOLIS data indicates that hospitals can automate up to 40% of inpatient cases with 99% accuracy. In fact, success from those efforts has allowed many XSOLIS clients to expand automation efforts to also include observation approvals, further reducing administrative waste.

“When you can show a certain type of claim is almost always paid on first pass or even after peer-to-peer review or appeal, you should start having conversations about automating the authorization or approval process around those claims,” said Sanderson. “With an objective view of medical necessity, providers and payers can see unbiased data tied to final outcomes. This creates the opportunity to eliminate administrative burden for both the provider and payer.”

5 financial advantages of AI in case management/utilization review

Organizations often see the following five financial advantages:

  • Administrative cost savings
  • Data-driven payer contract discussions enhancing partnership
  • Enhanced revenue integrity
  • Insight into payer reimbursement trends and appropriate patient status determinations
  • Workforce optimization

Fostering collaboration between case management and revenue cycle

“AI-driven insights promote productive conversations,” said Sanderson. “This information brings teams together.”

Neumany agrees. At Hackensack, case management and revenue cycle staff meet twice a month and use XSOLIS’ insights to drill down into denials and identify best practices for proactive denial prevention.

For example, Neumany trends denials by payer, hospital site, DRG and more. Through objective patient status decisions, she eliminates one of the factors that can make it difficult to drill down into the cause (i.e., subjective application of inpatient admission criteria).

“By leveraging AI and helping the utilization management team partner with the revenue cycle team, we drive increased revenue through appropriate inpatient admissions,” said Sanderson. “We also strategically align these teams through insightful analytics. This alignment streamlines handoffs to ensure appropriate reimbursement and reduces back-end delays associated with avoidable denials.”

4 ways healthcare organizations can improve revenue integrity with the CLS

Healthcare organizations can improve revenue integrity with XSOLIS Care Level Score (CLS) in these four ways:

  1. Quickly identify missed inpatient reimbursement and shift processes and/or resources to address cases proactively.
  2. Easily track utilization review (UR)-related denial performance end to end, including denials occurring at the point of service.
  3. Gain visibility into level of care denials that UR and physician advisor (PA) teams can’t address or aren’t aware of to promote front-end denial prevention.
  4. Easily identify opportunities for payer engagement with full insights into final payer alignment as well as outlier behavior.


“To me, AI is the future,” said Neumany. “We need to figure out how to use it to support our teams and improve our workflows. We’re trying to reduce the time and resources required to complete a review.”

Sanderson agreed, saying: “We need to do things differently to make healthcare sustainable. That requires us to remove excess administrative waste and focus on patient care. AI will help us get there. Our goal is to reduce the friction between providers and payers and bring forth new collaboration efforts to reduce the administrative cost of healthcare.” 


XSOLIS is a platform, data science and solutions innovator serving health plans, hospitals, and payer organizations nationwide to create a more efficient healthcare system. Through its purpose-built solutions and industry-leading AI, XSOLIS breaks down healthcare silos to accelerate data-driven decision-making and collaboration across a connected network of providers and payers. CORTEX®, its AI-driven technology platform, is the first and only solution to use real-time predictive analytics to continuously assign an objective medical necessity score and assess the anticipated level of care for every patient. CORTEX eliminates waste through the science of data using automation, transparency, and objective insights to ensure appropriate care settings, enabling more efficiency across the healthcare system.

This published piece is provided solely for informational purposes. HFMA does not endorse the published material or warrant or guarantee its accuracy. The statements and opinions by participants are those of the participants and not those of HFMA. References to commercial manufacturers, vendors, products, or services that may appear do not constitute endorsements by HFMA.


googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text1' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text2' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text3' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text4' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text5' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text6' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-text7' ); } );
googletag.cmd.push( function () { googletag.display( 'hfma-gpt-leaderboard' ); } );