ONC Announces AI Behavioral Health Care Grants

Health_IT

2026 executive update · ONC AI behavioral health grants · Leadership action

ONC Announces AI Behavioral Health Care Grants

The 2024 Leading Edge Acceleration Projects in Health Information Technology Special Emphasis Notice offered a focused lesson in federal grant readiness. ONC sought projects in two areas: improving healthcare data…

Greg Wahlstrom, MBA, HCMBlog

At a Glance

Competitive proposals are rarely built in the final weeks. They depend on a defined problem, an eligible lead applicant, credible partners, governance, usable data, community participation, measurable aims, a defensible budget, and a path beyond grant funding. Healthcare executives can create that foundation without speculating about…

Executive perspective

The 2024 Leading Edge Acceleration Projects in Health Information Technology Special Emphasis Notice offered a focused lesson in federal grant readiness. ONC sought projects in two areas: improving healthcare-data quality for responsible artificial intelligence and accelerating health IT adoption in behavioral health settings. The application period has closed, and HHS later announced two awards totaling $2 million. The strategic value now is not a stale reminder to apply. It is a practical model for how healthcare organizations can prepare before the next relevant funding window opens.

Competitive proposals are rarely built in the final weeks. They depend on a defined problem, an eligible lead applicant, credible partners, governance, usable data, community participation, measurable aims, a defensible budget, and a path beyond grant funding. Healthcare executives can create that foundation without speculating about future notices. Doing so also improves internally funded innovation because the same disciplines expose weak assumptions before capital is committed.

Grant readiness should therefore be managed as an enterprise capability. The goal is not to chase every opportunity. It is to recognize strong strategic fit early and assemble an application that can withstand programmatic, technical, financial, privacy, and implementation review.

Leadership priorities

Build an integrated leadership response

Translate the Funding Notice Into a Fit Decision

Build a structured review as soon as an official notice appears. Extract the purpose, areas of interest, eligible applicants, required partners, period of performance, award range, cost-sharing rules, deadlines, submission method, review criteria, reporting obligations, and restrictions. Assign an owner to each requirement and preserve the exact source. A summary slide is useful, but it cannot replace the controlling notice and official amendments.

Use a bid or no-bid scorecard. Test whether the opportunity advances an approved enterprise priority, whether the organization can lead or should join a consortium, whether the project can produce the requested public value, and whether required resources exist. Include leadership capacity, patient or community access, data availability, technical feasibility, compliance complexity, partner readiness, indirect costs, and sustainability. Require an explicit decision rather than allowing enthusiasm to become a commitment by default.

Define the problem in the funder's terms without distorting local need. For an AI data-quality project, identify the clinical decision, source data, known gaps, affected populations, and how improved quality will be tested. For behavioral health IT, describe the exchange or workflow barrier, the care setting, privacy considerations, user needs, and measurable benefit. A technology purchase alone is not a compelling problem statement.

Document why grant funding is necessary and what the organization will contribute. Reviewers should see a bounded acceleration opportunity, not an attempt to transfer ordinary operating costs to the government.

Assemble the Right Consortium Before Writing

Start with capabilities, then choose names. A credible team may need a public or nonprofit eligible lead, behavioral health providers, a health system, patients or peers, community organizations, academic evaluators, standards expertise, data science, privacy counsel, and technology partners. Map which work package each participant owns and which decision rights cannot be delegated.

Conduct partner diligence early. Confirm eligibility, registration status, financial controls, past performance, conflicts, cybersecurity posture, data access, human-subjects research implications, and willingness to meet federal reporting requirements. For-profit organizations may have a legitimate consortium or subrecipient role when permitted, but the arrangement must follow the notice and protect independence, competition, data rights, and public purpose.

Write a one-page consortium charter before the narrative. Include the problem, aims, governance, principal investigator or project director, work-package leads, patient and community roles, publication expectations, intellectual-property principles, data stewardship, dispute resolution, and decision cadence. Convert intentions into letters, scopes of work, and budget assumptions with enough time for institutional review.

Avoid symbolic participation. Behavioral health clinicians, frontline staff, patients, and community representatives should shape use cases, consent, workflow, outcomes, and dissemination. Define how they are supported and compensated where appropriate. A proposal built around real operating partners will be stronger and more implementable than one assembled for logos.

Design Testable Aims, Evidence, and Evaluation

Limit the project to a small number of specific aims that connect activities to outcomes. Each aim should name the population or setting, intervention, comparison or baseline, expected change, timetable, and evidence source. Distinguish implementation milestones from results. Building an interface, training users, or cleaning a dataset is an output; improved completeness, exchange reliability, decision quality, care coordination, or user experience is an outcome.

Create an evaluation plan alongside the technical design. Establish baseline measures, data provenance, analytic methods, missing-data treatment, subgroup analysis, qualitative methods, and the threshold for success. For AI work, document intended use, excluded use, model or tool boundaries, representativeness, performance across relevant groups, human oversight, drift monitoring, and failure response. The NIST AI Risk Management Framework can provide a shared structure without replacing healthcare-specific clinical and legal review.

For behavioral health IT, include usability, workflow burden, interoperability, patient experience, continuity, and equity. Account for HIPAA, 42 CFR Part 2, consent, state law, segmentation decisions, and information-sharing obligations applicable to the use case. Privacy should shape architecture from the beginning, not appear as a closing paragraph.

Predefine learning questions and dissemination products. A federal innovation award generally seeks knowledge that can travel. Explain what another organization could reuse, what must remain local, and how artifacts, standards contributions, implementation guides, code, or findings will be governed.

Invite an independent reviewer to challenge causal assumptions, feasibility, and measurement before submission. A short red-team review can expose an outcome that the available data cannot support, a missing stakeholder, or a work package that has no realistic dependency path. Resolve the finding or document why the design remains sound.

Build a Defensible Workplan and Budget

Convert aims into work packages with deliverables, dependencies, owners, dates, acceptance criteria, and risks. Show how discovery, design, procurement, security review, data preparation, development, testing, deployment, evaluation, and dissemination connect. Include realistic time for data-use agreements, institutional review, contracting, interface work, and frontline recruitment. Optimistic sequencing signals weak execution discipline.

Create the budget from the workplan rather than dividing the award ceiling among partners. Document personnel effort, fringe, travel, equipment, supplies, contractual work, participant support where allowed, indirect costs, and organizational contribution. Reconcile every narrative commitment to a funded activity. Finance and grants management should test allowability, allocability, reasonableness, documentation, and subrecipient monitoring before submission.

Build a resource-loaded risk register. Common risks include delayed agreements, data-quality failure, vendor dependency, low site participation, interoperability constraints, privacy redesign, staff turnover, and inability to recruit the intended population. Give each risk a trigger, mitigation, contingency, and owner. Reserve management attention even when a formal contingency budget is not allowed.

Plan reporting as part of delivery. Establish timekeeping, expense documentation, milestone evidence, subrecipient invoices, data-quality controls, and approval routes. A project that cannot produce reliable federal reports is not ready, regardless of its clinical promise.

Prove Sustainability and Public Value

Federal funds are time limited. Define what must continue after the period of performance, who will own it, and how it will be financed. Separate durable capability from temporary project activity. An open-source tool may still require hosting, security updates, user support, governance, and integration. A new data workflow may require permanent stewardship and quality monitoring.

Build a total-cost and adoption model. Estimate post-award technology, staff, licensing, interface, training, evaluation, and change-management costs. Identify the operational benefit, reimbursement or contract value where applicable, avoided cost, research value, and mission outcome. Do not force every project into a narrow financial return, but do show who will make the continuation decision and what evidence they will use.

Address transferability honestly. Document implementation prerequisites, standards, local dependencies, population limitations, and failure conditions. Publish negative findings when appropriate and permitted. Public value increases when others can understand what did not work and why.

Maintain a reusable readiness library containing organizational narratives, audited financial information, registrations, biosketches, facilities descriptions, privacy and security controls, data inventories, partner templates, budget assumptions, and prior performance. Review it quarterly. When the next aligned notice opens, the team can focus on the scientific and operating case rather than reconstructing basic evidence.

Leadership cadence

Start, strengthen, and measure the system in 90 days.

Start

Phase 1, days 1 to 30

Form a grant-readiness team, review the archived 2024 LEAP requirements as a case study, inventory eligible leads and partners, and identify two enterprise problems that align with responsible AI data quality or behavioral health interoperability.

Strengthen

Phase 2, days 31 to 60

Develop one-page concepts, test bid or no-bid criteria, validate data and workflow access, draft consortium charters, and build preliminary aims, evaluation measures, risk registers, and total-cost models.

Measure

Phase 3, days 61 to 90

Run an internal mock review against programmatic, technical, financial, equity, privacy, and implementation criteria. Correct gaps, approve reusable materials, and place the strongest concept in a monitored pipeline for future official opportunities.

Decision-grade measurement

Decision-Grade Metrics

  • Days from official notice publication to a documented bid or no-bid decision
  • Required capabilities filled, partner commitments secured, and institutional approvals outstanding
  • Aims with validated baselines, accessible data, named owners, and measurable success thresholds
  • Budget traceability to work packages, unresolved assumptions, and post-award total cost
  • Privacy, security, research, procurement, and data-agreement milestones completed on time
  • Community and frontline participation in design, governance, testing, and dissemination
  • Mock-review score, material weaknesses corrected, and continuation decision criteria approved

SEO

SEO title: ONC AI Behavioral Health Grants: Readiness Guide
Meta description: Use the 2024 ONC LEAP opportunity to build grant readiness across fit, partners, evidence, budgets, privacy, execution, and sustainability.
Focus keyphrase: ONC AI behavioral health grants

Conclusion

Turn strategy into an accountable operating system.

The 2024 ONC opportunity showed that responsible AI data and behavioral health interoperability can be framed as focused, testable public investments. The application window is closed, but the readiness disciplines remain valuable: strategic fit, capable partners, measurable aims, realistic execution, compliant stewardship, and a credible life after the award.

Executives should build those capabilities before the next notice arrives. An organization that can make a disciplined no-bid decision is as mature as one that can submit a strong application. The goal is not grant volume. It is funded work that the organization can execute, evaluate, sustain, and share.

Executive questions

Frequently Asked Questions

1. Is the 2024 ONC LEAP opportunity still open?

No. The 2024 Special Emphasis Notice application period closed on July 12, 2024, and HHS later announced the awardees. Organizations should use current official funding pages for new opportunities and treat the 2024 notice as a readiness case study.

2. Should a hospital lead every proposal it supports?

No. The lead should be eligible and best positioned to manage the program, finances, partners, and reporting. A health system may create more value as a clinical site, data partner, subrecipient, evaluator, or adoption partner.

3. What makes an AI data-quality aim fundable?

It needs a defined healthcare use, documented data problem, rigorous evaluation, relevant subgroup analysis, human oversight, and a reusable contribution. Simply applying a new model to available data is not enough.

4. How early should privacy and 42 CFR Part 2 review begin?

Begin during concept design. Privacy, consent, segmentation, access, redisclosure, security, and state-law constraints can change the architecture, partners, timetable, and budget. Late review can make an otherwise strong proposal infeasible.

5. What should executives require before approving a submission?

Require confirmed eligibility, strategic fit, committed partners, testable aims, accessible data, a resource-loaded workplan, a compliant budget, risk and governance plans, community participation, and explicit sustainability criteria.

Related Blogs