How to take advantage of the oncology infusion revenue opportunity
Cancer treatment is often expensive, and denials can cause havoc for providers. Agentic AI can help close the margin gap.
Oncology infusion services sit at the intersection of the highest-cost drugs in medicine, the most complex billing workflows in the revenue cycle and some of the most aggressive payer scrutiny in healthcare. For health systems, cancer centers and community oncology practices, the infusion suite is a clinical lifeline needing a strong financial engine. When that engine leaks, the losses are outsized.
Oncology denial rates can average about 15 percent, but because specialty drug claims often have five- and six-figure prices, a single denied infusion can wipe out the profit on dozens of routine visits. Reworking a denied claim can cost up to about $180 per attempt. When you multiply that by hundreds of weekly infusion encounters, the cost to collect reimbursement can threaten margins. Meanwhile, the CMS GLOBE model, scheduled to launch in October, will further narrow buy-and-bill margins, and payers are continually increasing prior authorization requirements for biologics, immunotherapies and biosimilars.
Organizations that protect and increase infusion margins in this environment will not achieve this by simply adding staff or working harder behind the scenes. Instead, they'll do it by deploying agentic AI throughout the revenue cycle to sense, decide and act on revenue-code-level data in real time. This playbook is tailored for clinical and financial leaders in oncology who want to shift from reactive denial management to proactive revenue integrity.
Where oncology infusion revenue leaks
Before designing an AI strategy, executives need a shared diagnosis of where revenue leaves the system. In oncology infusion, leakage generally concentrates in five areas.
Drug reimbursement errors. Drug reimbursement is the financial backbone of most oncology operations and the area most exposed to error. Charge capture issues, incorrect units, mismatched National Drug Codes or delayed posting lead to underpayments or outright denials. Medicare Part B reimburses most physician-administered drugs at the average sales price (ASP) plus 6 percent, but after sequestration adjustments, the effective margin is closer to 4.3 percent. When acquisition costs rise faster than quarterly ASP updates, practices absorb the gap. Even a 3 percent systematic underpayment on high-cost agents can translate into six-figure annual losses for a mid-sized program.
Authorization and benefit verification failures. Oncology treatments require extensive payer approval, and requirements vary widely by plan and drug. Missing, expired or misaligned authorizations are a leading cause of denials. Payers initiate reauthorization requirements mid-treatment for infusion therapies, particularly within specialty biologics. A single lapsed authorization on a $30,000 immunotherapy infusion creates a write-off that no appeal process can reliably recover.
Coding and modifier complexity. Each oncology infusion encounter may involve multiple CPT and HCPCS codes for drug administration, hydration, injections, supportive care and the drugs themselves (J-codes and Q-codes). Frequent payer updates and National Correct Coding Initiative edits make staying compliant a moving target. CMS has identified oncology as a high-risk specialty because of its expensive drug portfolio and complex billing patterns. Incorrect or missing modifiers are among the most frequent causes of denials under current CMS structures.
Drug waste and unused portions. Single-dose vials often contain more drug than a patient requires. The difference between the amount billed and the amount administered is a compliance and revenue issue. Payers are increasingly auditing drug waste claims, and documentation requirements for JW and JZ modifiers have tightened. Failure to capture waste accurately results in either lost revenue or audit exposure.
Site-of-service and 340B dynamics. The reimbursement gap between hospital outpatient departments and physician office settings for the same infusion service remains significant. CMS finalized a policy paying for certain drug administration services at 40 percent of the outpatient department rate to reduce unnecessary site-of-service shifts. Meanwhile, 340B-eligible hospitals retain a substantially higher share of insurer spending on infused drugs compared with independent practices. Understanding these dynamics is critical for any margin strategy.
Why traditional RCM approaches fall short
Oncology infusion billing demands a different level of precision. A variety of factors make this challenging.
Drug-specific payer rules change quarterly with ASP updates and formulary shifts.
A single infusion visit can generate 10 or more distinct charge lines across administration, drugs, hydration and supportive services.
Furthering the complication is the fact that prior authorization requirements differ by payer, plan, drug and even indication for the same molecule.
After administration, retrospective audits and recoupments target oncology disproportionately because of high claim values.
Spreadsheets, retrospective audits and manual work queues cannot keep pace with this complexity. The gap between what the revenue cycle can detect and what payers are enforcing widens every quarter.
Agentic AI’s implications for the revenue cycle
Agentic AI orchestrates autonomous agents that perform tasks, collaborate and learn from outcomes. Unlike generative AI, which primarily provides advisory support, agentic AI can watch queues, pull data from the EHR, billing system and payer portals, decide what to work on first, draft or submit actions, and learn from results over time.
For oncology executives, the question has shifted from whether to use AI in the revenue cycle to how to design AI agents around infusion revenue data to achieve measurable, sustainable financial results.
Think in terms of jobs to be done, not generic analytics. Each agent operates on revenue-code-level data tied to the specific economics of infusion care.
Pre-bill revenue integrity agent
This agent scans claims for missing charges, mismatched revenue code and HCPCS combinations, missing modifiers (including JW/JZ waste modifiers) and incorrect drug units before submission. It suggests corrections with confidence scores and a clear rationale tied to payer-specific rules.
The impact of this agent is that it catches the charge-capture errors that account for a disproportionate share of oncology revenue leakage.
Authorization intelligence agent
This agent monitors authorization status across every scheduled infusion encounter and flags expirations, missing authorizations and plan-specific reauthorization triggers. It embeds real-time authorization likelihood scores at scheduling so high-risk cases are routed to senior staff or peer-to-peer specialists before the patient arrives.
The impact of this agent is that it prevents the single most costly denial category in oncology infusion – the high-dollar drug claim denied because of a lapsed or missing authorization.
Denial prevention agent
This agent learns payer-specific denial patterns associated with specific revenue codes, diagnoses and documentation gaps for infusion services. It flags at-risk claims in real time and prompts preemptive action such as documentation addenda, code adjustments, or alternate billing strategies.
The impact of this agent is that it shifts denial management from a reactive cost center to a predictive margin-protection function.
Underpayment detection agent
This agent compares expected reimbursement by contract, revenue code and ASP schedule against actual payments and adjustments for every infusion claim. It surfaces systematic underpayments and under-allowed patterns for contract management and payer negotiation.
This agent makes an impact because it recovers revenue that most organizations currently leave on the table because they lack the granularity to detect it.
Drug cost and reimbursement spread agent
This agent monitors drug acquisition cost vs. reimbursement in real time, factoring in ASP updates, group purchase organization (GPO) contract pricing and 340B eligibility. It alerts finance and pharmacy leadership when specific drugs shift from margin-positive to margin-negative, enabling proactive formulary and site-of-service decisions.
The impact of this agent is that it protects against the GLOBE-era compression of buy-and-bill margins by giving leaders real-time visibility into per-drug economics.
Appeals and recovery agent
This agent assembles draft appeal letters, cites applicable rules and contract clauses, attaches clinical evidence and routes to staff for final review. It works by prioritizing accounts with the highest likelihood of overturn and expected yield, reducing low-value appeal work.
The impact of this agent is that it accelerates cash recovery with lower cost-to-collect and frees staff capacity for complex cases.
Building the data backbone
Agentic AI is only as strong as the data it consumes. The first executable step is to establish a unified, governed revenue data layer for infusion services.
Step 1: Consolidate and clean. Aggregate infusion revenue code data across hospital outpatient, physician office and ambulatory infusion settings into one logical source of truth. Resolve duplicates, conflicting code sets and inconsistent facility practices so that similar encounters look alike to AI agents.
Step 2: Normalize mappings. Standardize mappings between revenue codes, CPT/HCPCS, J-codes, NDCs, APCs and payer-specific groupers. Capture relationships among revenue codes, clinical protocols, regimen order sets and treatment pathways so agents can learn across encounters and locations.
Step 3: Attach context. For each infusion claim line, attach payer and product (commercial, Medicare, Medicare Advantage, Medicaid); site of service and treating physician; drug, dose, waste and indication; and outcome fields, such as the paid amount, denial reason, adjustment codes and appeal results. This transforms a billing code into a rich, machine-readable signal that AI agents can reason about and act on.
The human operating model pairs expertise with AI
Technology needs the right human architecture. The most successful programs co-own agentic AI between clinical, financial and operational leaders. Here are some important components to achieve this.
Create collaborative pods. Gather revenue integrity and coding professionals who understand oncology-specific charge capture, as well as clinical documentation improvement and HIM specialists with oncology experience.
To them, add infusion pharmacy and nursing leadership who control drug ordering, preparation and waste documentation. Another key expertise comes from data scientists and AI engineers who build and refine agent logic.
Establish feedback loops. This is accomplished by running weekly reviews of agent recommendations and outcomes; tracking false positives, false negatives and user overrides, and translating frontline feedback into updated training data, business rules and workflows.
Agentic AI must be treated as a learning system that evolves with payer changes, clinical practice updates and new drug approvals.
An executive scorecard for oncology infusion AI
Executives need a concise scorecard that monitors financial and operational impact at the infusion-service-line level. Information on this scorecard should include:
Denial rate for infusion-related revenue codes, by payer and site.
Drug margin per encounter – acquisition cost vs. reimbursement, net of waste.
Incremental revenue captured – missed charges prevented and underpayments recovered.
Net collection rate and days in accounts receivable for infusion claims.
Automation rate – percentage of infusion claims or tasks fully handled by agents vs. humans.
Authorization success rate, which is the percentage of scheduled infusions with confirmed, valid authorization at time of service.
Review this scorecard monthly at an executive forum and quarterly at the board or finance committee. Use it to decide whether to scale, tune or sunset specific agents.
Guardrails: Compliance, governance and trust
This is enabled by the following factors.
Transparent logic and audit trails. Every agent action should be explainable and traceable, with before-and-after revenue codes, rationale, data used and user overrides.
Formal governance council. Include compliance, legal, HIM, clinical oncology, revenue integrity, pharmacy and IT. Charge the council with approving new use cases, monitoring performance and reviewing audits.
Release gates for autonomy. Start agents in advisory mode. Promote to partial automation, then full automation, only after they consistently meet predefined accuracy and financial impact thresholds over a sustained period.
I recommend these sample phased guardrails, which give boards, clinicians and regulators confidence that AI supports professional judgment rather than replaces it.
Phase 1 (0 to 3 months). Baseline denial and cash KPIs for infusion services. Clean and consolidate revenue code data and dashboards. Pilot AI for pre-bill claim scrubbing and authorization monitoring.
Phase 2 (3 to 9 months). Expand AI to denial risk scoring, underpayment detection, drug margin monitoring and worklist prioritization for infusion claims.
Phase 3 (9 to 12 months). Scale across all infusion sites. Embed AI insights into payer strategy, contract negotiation, and budgeting. Report margin and revenue effects to the board quarterly.
Revenue integrity as the strategic imperative
CMS is restructuring drug reimbursement through models like GLOBE, and oncology organizations cannot afford to treat revenue integrity as a secondary strategy.
Revenue integrity in oncology infusion is the discipline that connects clinical documentation, charge capture, coding accuracy, payer contract enforcement and post-payment audit defense into a single, continuous process. When any link in that chain breaks, the financial impact is immediate and outsized.
Agentic AI enables clinical and financial leaders to operationalize revenue integrity at scale. Rather than relying on retrospective audits or manual reviews to catch errors after revenue has already leaked, intelligent agents embed integrity checks at every stage of the infusion revenue cycle – before the claim is submitted, at the point of payment and during post-payment reconciliation. The result is an organization that collects revenue more efficiently, from the moment a treatment is ordered to the moment cash is posted.
The organizations that succeed in the next phase of oncology transformation will treat revenue integrity as a strategic asset, not just a departmental task. They will combine deep clinical and billing expertise with agentic AI, using infusion data as a common language among oncologists, pharmacists, coders and finance leaders.
For healthcare executives overseeing oncology services, the clear directive is to elevate revenue integrity from a reactive function to a board-level priority, enabling agentic AI to provide the intelligence, speed and scale needed to make it sustainable.
Moses Landon, MBA, EHRC, SA, FACHDM, is a senior executive advisor and works in advisory services for financial transformation for Premier Inc.