Adapting AI Prescreening Approaches to Advance 340B Operations

Artificial intelligence (AI) has gained momentum in recent years, with new systems emerging across nearly every industry. AI is slowly becoming part of daily operations in healthcare, showing promise in improving efficiency, accuracy, and workflow. One of the most notable areas where AI has proven effective is in clinical trial prescreening, and the strategies used there can be applied to 340B program management.

The 340B Drug Pricing Program was created by Congress in 1992 via the Public Health Services Act to help eligible health care organizations, known as covered entities, access outpatient drugs at significantly reduced prices. According to the Health Resources and Services Administration (HRSA), the program's intent is to allow these entities to "stretch scarce federal resources as far as possible, reaching more eligible patients and providing more comprehensive services."¹ In practice, this means safety-net hospitals, clinics, and specialized treatment centers can reinvest the savings from discounted drug costs into expanding patient access to care, funding essential clinical services, and supporting underserved communities.

Much like 340B’s goal of doing more with less, AI offers a modern way to stretch resources further, especially in areas like clinical trial recruitment, where efficiency and accuracy are crucial. A compelling example comes from studies using the RECTIFIER tool, a retrieval-augmented generation-enabled GPT-4 system designed to prescreen patients and evaluate the accuracy of eligibility assessments.

In 1 study, RECTIFIER achieved accuracy rates between 97.9% and 100%, compared to study staff who ranged from 91.7% to 100% when prescreening for heart failure clinical trials (COPILOT-HF).² The program analyzes unstructured data within electronic health records (EHRs). While traditional EHR searches can easily process structured information such as coded diagnoses, lab values, and medication lists, they often miss key details found in free-text physician notes, scanned documents, and other unstructured formats. This gap forces staff to perform time-consuming manual reviews, delaying recruitment and driving up costs.

In the same COPILOT-HF clinical trial, a separate RECTIFIER study showed that the program outperformed human reviewers in determining patient eligibility and significantly reduced prescreening time. It proved especially effective at identifying symptomatic heart failure and demonstrated measurable improvements in enrollment outcomes, with an AI eligibility rate of 20.4% compared to 12.7% manually, and a total of 35 patients enrolled versus 19 through manual review.³ These results highlight how AI can process structured and unstructured information at scale with accuracy and efficiency, enhancing trial recruitment.

Considering the results of the studies, the 340B program management can benefit from the integration of an AI tool. On the backend, 340B operations involve everything from audits and claim reviews to confirming prescription and site eligibility. Generally, covered entities rely on third-party administrators (TPAs) that use specialized software (i.e., matching logic) to link structured EHR data with prescription claims. While this works well, these systems cannot interpret unstructured data such as free-text provider notes, scanned documents, or narrative fields. Manual review often slows these processes and introduces a risk of human error. Much like in clinical trial screenings, unstructured EHR data contain eligibility verification data, which would require labor-intensive searching and interpretation. AI could transform these workflows by rapidly scanning records, identifying eligible prescriptions, and flagging potential problems for human verification. Integrated into EHR systems, it can tag prescriptions, providers, and sites as 340B-eligible in real-time, reducing the likelihood of missed opportunities or compliance errors. It could also detect patterns or anomalies that indicate potential noncompliance, enabling early intervention before issues become audit findings.

The key advantage of AI lies in its ability to handle the initial workload. By taking on the first-pass review of vast amounts of data, AI allows pharmacists and compliance staff to focus on high-value oversight rather than repetitive manual checks. This can increase efficiency, improve accuracy, and help capture all eligible claims, ultimately strengthening program integrity and ensuring that more resources remain available for patient care. The success of AI in clinical trial prescreening shows that similar tools could help 340B programs operate more quickly and precisely. While human expertise will always be essential for final decision-making, AI can serve as a dependable and unwavering assistant, turning a complex, time-intensive process into one that is faster, more accurate, and better aligned with the program's mission.

REFERENCES:

1. Health Resources and Services Administration. 340B Drug Pricing Program. US

Department of Health and Human Services. Updated October 2023. Accessed August 26,

2025. https://www.hrsa.gov/opa

1. Unlu O, Shin J, Mailly C, et al. Retrieval-augmented generation-enabled GPT-4 for clinical trial screening. New Engl J Med AI. 2024;1(7). doi:10.1056/AIoa2400181.
2. Unlu O, Varugheese M, Shin J, et al. Manual vs AI-Assisted Prescreening for Trial Eligibility Using Large Language Models — A Randomized Clinical Trial. JAMA.

2025;333(12):1084–1087. Published online February 17, 2025.

doi:10.1001/jama.2024.28047