Explainable AI Is the Future of Payment Integrity

Healthcare leaders, take note: AI must evolve beyond denial-focused workflows and become a transparent, forward-looking partner in payment integrity that supports both payment accuracy and streamlines clinical innovation.

This is the compelling vision of Meghan Dierks, MD, FACS, former Chief AI Officer at the U.S. Department of Health and Human Services and a faculty member at Harvard Medical School.  Dierks, who recently spoke at the AI for PI Summit 2025, urged payers, providers, and healthcare executives to consider this critical shift. Not only would AI adoption with advanced analytics transform healthcare industry cost savings as we know it today, but it also could dramatically improve trust and value-based care outcomes.   

The Payment Integrity Paradox

AI and generative AI has undeniably improved fraud detection, coding validation, and claims processing and automation. Yet many providers now view payment integrity solutions as rigid systems that prioritize short‑term financial control over patient outcomes and innovation. That double‑edged sword—financial gains over administrative costs with growing provider distrust—cannot persist.

Dierks pointed to several areas where traditional payment integrity programs fall short.

  • Precision medicine: Genomic testing and biomarker-driven therapies, now the standard of care in oncology and other specialties, are often flagged as “experimental” because legacy AI models rely on outdated utilization patterns.
  • Digital and hybrid care models: Remote monitoring, the use of AI-driven diagnostics, and advancements in digital therapeutics are reshaping patient care, yet conventional payment integrity solutions still treat them as anomalies.
  • Value-based care: Providers pursuing long-term outcomes often face denials for interventions that don’t fit fee-for-service frameworks, even when these interventions reduce total cost of care over time.

This tension underscores the need for cutting-edge payment integrity processes that evolve as quickly as modern medicine—and incorporate human expertise that enhances, rather than obstructs, patient care.

Reimagining AI in Payment Integrity

Dierks urged the industry to move toward a “validation paradigm”: one where the use of AI dynamically assesses treatments against real-time clinical evidence, patient biology, and emerging guidelines. Rather than reflexively blocking coverage, next-generation payment integrity should validate appropriate innovation and reduce the time-consuming lag between new medical advances and payer policies.

Achieving this vision requires a strong foundation:

Unified Data Architecture

PI must shift from siloed claims data to a comprehensive data orchestration model. By integrating electronic health records, specialty guidelines, FDA updates, payer policies, and real-world evidence, AI systems can deliver a single, current source of truth for coverage decisions.

Continuous Model Evolution

Static, rules-based models cannot keep pace with modern healthcare. Continuous learning pipelines, capable of ingesting new guidelines, appeals outcomes, and clinical evidence, are essential for maintaining both accuracy and relevance.

Explainable AI (XAI)

Transparency is non-negotiable. Dierks emphasized that explainable AI, particularly through counterfactual reasoning, can transform payment integrity from a “computer says no” gatekeeper into a system that tells providers why a decision was made—and what would change the outcome. This approach not only accelerates human review by 3–5x but also builds trust by turning denials into actionable guidance.

The Path to Adoption: From Theory to Practice

Transforming health plan payment integrity is not about deploying technology overnight just for operational efficiency; it’s about leveraging artificial intelligence for targeted, incremental progress. Dierks recommended starting with high-impact, high-complexity areas such as precision oncology, high-cost specialty drugs, and complex surgical interventions. These are the reimbursement areas where outdated PI approaches create the most friction, financial risk, and regulatory scrutiny.

A key element of adoption is medical record workflow integration. AI-driven payment integrity should enhance existing EHR-centric workflows rather than add administrative burdens. Embedding decision support at the point of care—before claims are submitted—can help providers ensure claims accuracy in real time, reducing downstream denials and appeals.

Dierks also highlighted the convergence of clinical decision support (CDS) and payment integrity teams. Historically, CDS optimized patient outcomes, while PI focused on administrative cost control. Advanced and generative AI can now handle both functions, recommending evidence-based care while ensuring cost-effective coverage decisions. This convergence will eliminate adversarial dynamics and replace them with shared understanding between healthcare payers and providers.

Trust, Transparency, and the Flywheel Effect

Explainable AI is the linchpin for building stakeholder trust in the next generation of payment integrity. Counterfactual narratives—answering questions like “What’s the smallest change needed to approve this claim?”—empower both providers and reviewers. These narratives accelerate decision-making, reduce administrative friction, and foster an ecosystem of AI-powered learning on both sides.

Over time, the adoption of XAI creates a flywheel effect:

  • Continuous learning from real-world outcomes and provider overrides improves model accuracy.
  • Proactive coverage intelligence anticipates new treatments and policies instead of reacting late.
  • Regulatory readiness is enhanced through transparent, audit-ready decision logic.
  • Provider collaboration improves as denials become rare, predictable, and explainable.

This evolution shifts PI from a retrospective, adversarial process to a proactive, collaborative system. It supports both cost containment and clinical excellence, a balance the healthcare system desperately needs.

A Future Built on Collaboration

Dierks concluded her session with an urgent call to action: Payment integrity leaders must think beyond traditional denial workflows and embrace AI tools that learn and explain as fast as medicine evolves. With a focus on transparency, adaptability, and collaboration, payment integrity can become not just a safeguard for financial stewardship but a catalyst for healthcare innovation.

Ready to transform your payment integrity program and prevent improper payments with earlier, smarter intervention? Visit https://info.codoxo.com/pointzeropiblog to learn more.