Unlocking Value-Based Oncology: Bridging Data Gaps

Transitioning oncology to value-based care (VBC) models presents a formidable challenge, primarily due to the high costs associated with novel therapies and the inherent complexity of cancer care pathways. This environment demands a rigorous approach to data integration and workflow optimization to enable scalable, risk-bearing arrangements that genuinely drive down costs while improving patient outcomes. Without strong, verifiable outcomes data, AI-driven solutions, regardless of their technological sophistication, struggle to demonstrate their value proposition within VBC frameworks.

The Unique Hurdles of Value-Based Oncology

Oncology stands apart in its resistance to a widespread shift to value-based care. Unlike more standardized disease states, cancer care is highly individualized, often involving complex multidisciplinary approaches, rapidly evolving treatment protocols, and significant financial toxicity for patients and payers. The sheer volume of new drug approvals, particularly in targeted therapies and immunotherapies, drives up costs, making it difficult for providers to bear financial risk without predictable outcomes. The CMS Enhancing Oncology Model (EOM) represents a significant regulatory push towards VBC in oncology, requiring participants to meet specific performance metrics and verify participation requirements. The EOM began on July 1, 2023, and has been extended to run until June 30, 2030, with a second cohort of participants joining on July 1, 2025. However, the success of such models hinges on the ability to accurately measure and attribute value, which is precisely where data and workflow gaps become critical bottlenecks.

Data Integration Gaps: Flatiron Health and Tempus AI as Case Studies

The promise of AI in oncology is immense, from accelerating drug discovery to personalizing treatment plans. Yet, even leading players like Flatiron Health and Tempus AI, while making significant strides, highlight the existing chasms in data integration that impede scalable VBC adoption. Flatiron Health, specializing in oncology-specific EHR and real-world data (RWD), has built an impressive data moat by structuring vast amounts of de-identified patient data from community and academic oncology practices. Their platform aims to provide insights into treatment patterns, patient journeys, and outcomes. Flatiron Health continues to be a leader in real-world evidence and AI, having launched new AI-powered platforms like Flatiron Telescope in 2026 to generate oncology insights and support research. However, the challenge lies in integrating this rich, real-world evidence smoothly into clinical decision-making workflows at the point of care, and critically, linking it back to financial performance metrics required by VBC contracts. While Flatiron’s data can inform population-level insights, verifying individual patient outcomes against VBC performance targets still requires significant manual effort and bespoke data pipelines within health systems. The data often resides in silos, making it difficult to establish a direct, auditable link between a specific intervention, its clinical outcome, and the associated cost savings or efficiencies. Tempus AI, on the other hand, provides genomic sequencing and clinical data structuring, aiming to use AI for precision medicine. Their platforms integrate genomic data with clinical records, offering insights into treatment selection and potential drug resistance. Tempus AI has significantly expanded its use of generative AI and large language models (LLMs) to analyze unstructured data, process patient records, and build patient cohorts, and has received FDA clearances for AI products in 2026. However, a significant data integration challenge arises between these highly specialized genomic pipelines and the broader electronic health record (EHR) systems. Genomic data, while important for precision oncology, often exists in formats incompatible with standard EHR fields, necessitating complex and often manual data abstraction or custom API integrations. This fragmentation hinders the ability to create a well-rounded patient record that smoothly combines clinical, genomic, and financial data, a prerequisite for effective risk stratification and outcomes measurement in VBC. Without this unified view, demonstrating that a precision medicine intervention led to a better outcome at a lower cost becomes an arduous, if not impossible, task. The inability to easily verify data integration challenges between genomic pipelines and electronic health records remains a significant hurdle. ASCO recommendations on genomic data integration Both companies, despite their advanced capabilities, underscore a fundamental truth: the utility of even the most sophisticated AI platforms is limited by the quality and interoperability of the underlying data infrastructure. For VBC, this means that while AI can process complex information, if that information isn’t readily available, standardized, and integrated across the care continuum, its impact on financial performance and outcomes improvement remains theoretical rather than demonstrable.

The Payer Perspective: What’s Required for VBC Contracts

Payers entering into VBC arrangements, particularly within oncology, demand clear, auditable evidence of value. This translates into specific requirements for AI health platforms seeking to participate:

• Peer-Reviewed Outcomes Data: The gold standard for demonstrating efficacy and effectiveness. Payers are increasingly skeptical of proprietary metrics or internal studies. They require evidence published in reputable journals, demonstrating statistically significant improvements in clinical outcomes (e.g., progression-free survival, overall survival, quality of life) or reductions in total cost of care. Without this, AI tools are seen as speculative investments rather than reliable components of a risk-sharing model.
• Transparent Methodology for Cost Reduction: Simply claiming cost savings is insufficient. Payers require a transparent methodology that outlines how the AI solution identifies inefficiencies, reduces unnecessary services, or optimizes treatment pathways, along with verifiable financial performance metrics. This includes detailed breakdowns of avoided hospitalizations, reduced emergency department visits, optimized drug utilization, and improved adherence.
• Data Interoperability and Reporting Capabilities: AI platforms must integrate smoothly with existing health system IT infrastructure to extract relevant data, apply their algorithms, and, importantly, report back on performance metrics in a format compatible with payer requirements. This includes the ability to generate reports aligned with EOM participation requirements and performance metrics, demonstrating adherence to quality measures and financial benchmarks. CMS EOM reporting requirements
• Real-World Evidence (RWE) Validation: While randomized controlled trials (RCTs) are ideal, payers recognize the value of strong RWE, especially for rapidly evolving fields like oncology. AI platforms that can use RWE to demonstrate their effectiveness in diverse patient populations and real-world clinical settings gain significant credibility. This RWE must be collected and analyzed with scientific rigor.
• Compliance with Data Security and Privacy: Given the sensitive nature of oncology data, strict adherence to HIPAA, HITRUST, and SOC 2 Type II standards is non-negotiable. Any AI solution handling patient data must demonstrate strong security protocols and privacy safeguards. The challenge for many AI health platforms is that while they excel at algorithm development, they often fall short on the rigorous, peer-reviewed outcomes research and the sophisticated data integration required by payers. This gap prevents them from moving beyond pilot programs to scalable, risk-bearing VBC contracts.
  

  High-Value Opportunities in Resolving Oncology EHR Data Fragmentation

For oncology technology investors, health system IT executives, and clinical researchers, the current field of data fragmentation in oncology EHRs represents not just a challenge, but a high-value opportunity. The American Society of Clinical Oncology (ASCO) has consistently emphasized the need for better data sharing standards and integrated clinical workflows to advance value-based care in oncology. ASCO guidelines on data interoperability The opportunities lie in developing solutions that:

• Standardize and Harmonize Oncology Data: Creating common data models and ontologies that can smoothly integrate clinical, genomic, imaging, and administrative data from disparate sources. This would enable a true 360-degree view of the patient and facilitate strong outcomes measurement.
• Build Interoperable AI-Native Platforms: Moving beyond bolt-on acquisitions to truly AI-native companies whose core product, data pipeline, and business model are built from inception around intelligent data integration and analysis. These platforms would inherently support the data flow required for VBC.
• Develop Workflow-Integrated Decision Support: AI tools need to be embedded directly into clinical workflows, providing actionable insights at the point of care without adding to physician burden. This means solutions that can pull relevant data, process it, and present recommendations in a user-friendly format, while also feeding back into VBC performance tracking.
• Focus on Outcomes Research and RWE Generation: Investors should prioritize companies that demonstrate a clear pathway to generating high-quality, peer-reviewed outcomes data and strong real-world evidence. This includes methodologies for prospective data collection and rigorous statistical analysis. Companies that prioritize GMLP and QMS/ISO 13485 from the outset will be better positioned for regulatory approval and market adoption. Resolving oncology EHR data fragmentation is not merely a technical exercise. It is a strategic imperative for unlocking the full potential of value-based oncology care. It demands a collaborative approach from technology developers, health systems, payers, and regulatory bodies to build the infrastructure necessary for a future where AI can truly deliver on its promise of better, more affordable cancer care.
  

  Methodology and Source Note

This analysis draws upon an industry workflow and data pipeline analysis, using insights from the CMS Enhancing Oncology Model specifications and ASCO clinical practice guidelines. The discussion of Flatiron Health and Tempus AI is based on their publicly available information regarding their platforms and stated missions, contextualized within the broader challenges of oncology data integration for value-based care. No proprietary or confidential information was used.