Preventing and Appealing Clinical Denials with Analytics, Automation, and AI



February 1, 2021

U.S. hospitals lose $262 billion due to denied claims each year, representing roughly 10% of the total claims paid.

Unraveling the root cause of denials costs an average of $118 for each hospital claim, with 60% to 80% of those denials typically recoverable on appeal. For an average hospital submitting 200,000 claims annually, that represents over $2M in potential rework. Hospitals typically lose 2-3% of net patient revenue from claim denials.

The reasons that insurers give for clinical denials vary widely, but commonly include “prior authorization missing,” “not medically necessary,” “experimental or investigational,” “length of stay,” and “treat in a lower level of care setting.” What is not commonly understood is how the reasons relate to the clinical root cause, what medical documentation is required to support the service, and how the payer’s clinical policies relate to the denial decision.

Facing the need to solve for denials from multiple payers with varied clinical policies, hospitals have traditionally hired nurses and other medical professionals to manually review cases, navigate systems, and liaise with departments to investigate and appeal denied claims. Hospitals frequently run into resource limitations with this conventional approach. Today, maybe 20% of clinical denials can be appealed with this method.

Successful clinical denial resolution requires a clinical skillset to research the root cause and to address the patient’s medical situation and appropriate payer criteria in an appeal. Applying modern technology — such as analytics, automation, and artificial intelligence (AI) — can help to increase the number of clinical denials investigated, boost appeal effectiveness by addressing true payer clinical policy root cause, and recover much-needed revenue.


Definition and context: Analytics is a mechanism for extracting meaningful insights from an organization’s existing data for better business decision making. Data are extracted from many source systems; cleansed, integrated and stored in a data lake; enriched and synthesized with the latest AI and analytics tools; and then rendered in reporting dashboards for decision-makers.

Potential analytics applications for clinical denials:

  • Indicate the propensity to overturn denials based on payer, denial error, procedure code, and diagnosis.
  • Identify trends in clinical documentation integrity by service and common payer clinical criteria.
  • Identify potentially uncollectible codes through prior appeal trends.


Definition and context: Robotic process automation (RPA) uses existing end-user software interfaces and enterprise applications to automate processes to reduce effort and bring in process efficiency. Full and hybrid/blended (with human intervention) automation approaches are possible. Automation leads to better capacity handling, faster processing, fewer errors, reduced penalties, better compliance, and cost optimization.

Potential automation applications for clinical denials:

  • Appeal templates for faster and accurate processing.
  • Automated prior authorization workflows to negate denials for missing authorizations (up to 10% of ACA plan claim denials relate to prior authorization).
  • Image recognition to facilitate automated routing and standard appeal follow-up on common clinical denials.
  • Clinical data integrity through automation using common denial sources from patient and provider data.

Artificial intelligence

Definition and context: While RPA has helped to automate rules-based, structured data and processes, cognitive automation is the process of identifying and processing unstructured and semi-structured data. Cognitive automation brings much needed artificial intelligence into rule-based RPA. Cognitive elements such as machine learning (ML) and natural language processing (NLP) can make sense of data that appears to lack patterns. AI goes one step further than RPA by enabling machines to execute subjective processes requiring decision making.

Potential AI applications for clinical denials:

  • Verify required medical documentation for frequently denied services and flag tagged portions for clinical review.
  • Digitize the request process for commonly required medical records when a patient’s file is missing outside physician records or other specialist encounters to support denied service.
  • Provide a summary (for a clinician) of the patient’s associated medical condition related to applicable payer clinical criteria.
  • Generate appeal letters based on medical records for clinician review for common conditions and denied services.

In clinical denials, a technology-led approach increases the number of appeals that the hospital is able to address, revenue collected, and effective clinician allocation. Analytics, automation, and AI can improve root-cause analysis based on underlying payer clinical criteria while both refocusing and minimizing processing tasks for staff.

Up to 90% of denied claims are preventable. Rework is preventable. With the help of technology, medical professionals can optimize research, decrease overall denials with key prevention strategies for medical documentation and billing, and reduce time spent on uncollectible accounts.

Get more ideas on how to use analytics, automation, and AI for revenue cycle management in our “Accelerating claims reimbursement with Sagility clinical denials advanced recovery” case study and our “Recover lost revenue: Leverage AI to automate your revenue cycle” whitepaper, which offer insights on how to use technology to maximize provider revenue.