Pilot Testing, Implementation, and Evaluation of the Cervical, Breast and Colorectal Cancer Clinical Decision Support Tools

Academic PIs and health-services researchers just got a five-month head start: RFA-DP-27-036, a $7.25 million CDC cooperative agreement to pilot test and evaluate cervical, breast, and colorectal cancer clinical decision support tools, was forecast on Grants.gov July 24 with applications estimated due March 23, 2027.

A forecast notice is not an open solicitation, and that gap is the whole opportunity

The listing that went up on Grants.gov on July 24, 2026 is a forecast record, not a live synopsis. CDC's National Center for Chronic Disease Prevention and Health Promotion has published its intent, its money, and its structure — but the actual notice of funding opportunity is not estimated to post until January 21, 2027.

The numbers CDC committed to in the forecast are specific enough to plan against. Estimated total program funding is $7,250,000. Expected number of awards: two. The funding instrument is a cooperative agreement, not a standard grant. Cost sharing is not required. Award ceiling and floor are both listed as zero, which in forecast records means "not yet specified" rather than "unlimited" — do not build a budget on that field. The fiscal year is 2027, estimated award date is August 30, 2027, and estimated project start is September 30, 2027. The forecast record archives April 22, 2027. The program contact is Celeste Sanders, PhD, at researchnofo@cdc.gov.

Read the estimated dates as a timeline, not a promise. Forecast dates slip routinely. But the shape of the timeline is the actionable part: roughly nine weeks between the NOFO posting and the deadline. Nine weeks is enough time to write a proposal. It is nowhere near enough time to acquire the thing this NOFO actually requires.

The tools already exist, which changes what you are proposing

This is the detail that will sink applications from investigators who skim the title and pattern-match to a typical R01.

CDC is not asking anyone to build a clinical decision support tool. The agency, working with the MITRE Corporation's health federally funded research and development center and scientists from the National Cancer Institute, has spent years translating narrative screening guidelines into computable form. The cervical cancer product of that work — the CCSM CDS tools, built on FHIR, Clinical Quality Language, and a SMART on FHIR interface — is already public, with source code, documentation, test cases, and a working demo. The methodology was published in the Journal of Women's Health in 2022 by Saraiya and colleagues.

The forecast language is explicit about this. Component A covers "system-wide implementation and evaluation of a publicly available cervical cancer CDS tool within a large health system," with the objective of improving guideline-concordant screening and ensuring timely follow-up after abnormal results. Component B covers pilot testing "publicly available, interoperable CDS tools" for breast and colorectal screening, including technical integration into real-world EHR systems, documentation of the implementation process, and empirical evaluation of accuracy and impact.

Every operative verb is an implementation verb: implement, integrate, document, evaluate, assess. None of them is develop.

That means the competitive proposal here is an implementation science proposal. Hybrid type 2 or type 3 effectiveness-implementation designs. RE-AIM or CFIR framing. Fidelity measurement, alert-burden and override-rate outcomes, stepped-wedge or interrupted time-series analysis across clinics. A PI whose strongest asset is novel algorithm development is applying to the wrong half of the translational pipeline. A PI who can credibly measure whether a functioning tool changes clinician behavior at scale is exactly the target.

The "cooperative agreement" instrument reinforces the point. Cooperative agreements carry substantial federal programmatic involvement — CDC staff and, given the provenance of these tools, likely the MITRE technical team will be actively in the project. Applicants who present a sealed-box research plan with no collaboration surface tend to score poorly against that instrument.

Two components, two awards, and arithmetic worth doing early

Two awards against $7.25 million averages roughly $3.6 million per award. The structure strongly implies one award per component rather than two awards drawn from a common pool.

If that reading holds, this is not a single competition with two winners. It is two separate competitions with one winner each. That materially changes the go/no-go calculus, and it changes it differently depending on which component you target. Component A is a single health system deploying a mature, already-piloted cervical tool system-wide. Component B is a pilot integration of two less-mature toolsets across a fundamentally messier evidence base. Applicant teams should pick deliberately and early, and should ask the program contact directly whether components are separately competed once the NOFO posts.

Eligibility is unusually wide. The forecast lists sixteen applicant types, including public and state-controlled institutions of higher education, private institutions of higher education, nonprofits with and without 501(c)(3) status, tribal governments and organizations, small businesses, for-profit entities, and an explicit "unrestricted" category. Academic medical centers, integrated delivery networks, and health IT vendors are all eligible to lead. Expect an academic-plus-health-system consortium to be the winning shape, but do not assume the field is limited to universities.

The health system partner is the long pole, not the science

Here is why the five-month runway matters more than the nine-week writing window.

Both components require a large health system willing to let you modify its production EHR. That is not a letter-of-support relationship. It requires clinical informatics governance approval, security and privacy review, a data use agreement, IRB determination, and a slot in an EHR build queue that is typically scheduled quarters ahead. At most academic medical centers, the sequence from first conversation to signed authorization runs three to nine months. Starting that conversation when the NOFO drops in January means missing the March deadline with an unsigned partner.

The technical due diligence is equally unforgiving. The published cervical cancer work identified a specific, hard barrier: applying risk-based management guidelines requires cervical cytology, histology, and related pathology data, and pathology reports are overwhelmingly unstructured narrative text that automated systems cannot parse. The proposed workarounds — natural language processing over pathology reports, structured data capture forms for manual entry, SMART Health Cards for standardized result transmission — are each a real engineering subproject with a real failure mode.

An applicant who has already run a data-availability audit against their partner's pathology feed, and can state in the proposal what percentage of results arrive structured, is going to be far more convincing than one promising to find out in year one.

Why breast and colorectal is the harder half

Component B looks like the softer target because "pilot test" sounds lighter than "system-wide implementation." It probably is not.

The cervical tool has years of development, a published methodology, a public code repository, and a 2024 MIPS improvement activity (IA_PM_23) that explicitly rewards clinicians for using computable cervical screening guidelines. The breast and colorectal tools are earlier in that lifecycle.

They also sit on top of guidance that has moved recently. USPSTF now recommends biennial screening mammography beginning at age 40 rather than 50, and recommends offering colorectal cancer screening starting at age 45 with a B recommendation. Encoding a guideline that shifted within the last few years — and where clinician awareness and uptake are still uneven — means the CDS logic and the clinician behavior it targets are both moving. That is scientifically interesting and operationally risky. Say so in the proposal rather than hoping reviewers do not notice.

What to build between now and January 21

Investigators who follow federal cancer research funding will recognize the rhythm here from NCI's NCORP 2027 renewal cycle, where the teams that placed well were the ones with network infrastructure standing before the RFAs posted. The same dynamic applies with more force here, because the required asset is a signed health system partner rather than an existing consortium.

Between now and the estimated January 21 posting, the work is partnership and diligence, not prose:

• Identify and formally approach a health system partner with a single-instance EHR and enough screening volume to power your endpoints. Get informatics governance engaged, not just a physician champion.
• Audit pathology and screening data structure at that partner. Quantify what is structured versus narrative.
• Pull and read the public CCSM CDS repository and run the demo against realistic test cases before committing to a component.
• Email researchnofo@cdc.gov to be notified when the NOFO posts, and to ask whether Components A and B are competed separately.
• Draft the evaluation design now. It is the part of the application least likely to change when the actual NOFO language lands.

To see what else is currently open in this space while you wait for the January posting, search active cancer screening and clinical decision support solicitations on Granted.

Forecast notices exist precisely so that applicants who need long-lead resources can go get them. For RFA-DP-27-036, the long-lead resource is a health system that has already agreed to let you touch its EHR. Nobody assembles that in nine weeks.