Abstract
Community Health Needs Assessments (CHNAs), mandated by the Affordable Care Act for tax-exempt hospitals, represent an underutilized yet rich data source for disease-specific advocacy. This commentary proposes a novel framework in which disease advocacy organizations—such as Alzheimer’s Los Angeles, the American Heart Association, and the National Alliance on Mental Illness—deploy artificial intelligence (AI) agents to systematically analyze CHNAs, identify gaps in condition-specific care, generate personalized outreach to hospital leadership, and publicly score health systems on their responsiveness to identified needs. Using Alzheimer’s disease and dementia care in Los Angeles County as a primary case example, this article describes how AI-driven automation of data collection, natural language processing of CHNA documents, and coordinated advocacy campaigns can transform the current passive CHNA cycle into an active mechanism for population health improvement. The framework combines reputational accountability through public scorecards with constructive, evidence-based recommendations, creating a “carrot-and-stick” dynamic that existing literature on public performance reporting suggests can achieve engagement rates of 40%–70% and meaningful institutional change in 30%–60% of targeted systems. This approach is adaptable across chronic conditions and disease advocacy organizations, wherever publicly reported community needs data intersect with organized patient advocacy. Implications for population health management, health system quality improvement, and the responsible integration of AI in public health advocacy are discussed.
Introduction
The Affordable Care Act’s requirement that tax-exempt hospitals conduct Community Health Needs Assessments (CHNAs) every 3 years created an unprecedented public repository of information about the health priorities, resource gaps, and unmet needs of communities across the United States. These documents, typically filed with the Internal Revenue Service and published on hospital websites, are intended to guide community benefit spending and strategic planning. In practice, however, the CHNA process has remained largely a compliance exercise. Hospitals conduct the assessment, file the required documentation, develop implementation strategies of variable specificity, and often move forward with limited external accountability for whether identified needs are meaningfully addressed.
Disease advocacy organizations occupy a unique position in the health care landscape. Groups such as Alzheimer’s Los Angeles, the American Heart Association, and the National Alliance on Mental Illness possess deep expertise in their respective conditions, established public credibility, extensive networks of patients and caregivers, and a mission-driven imperative to improve care quality. Yet these organizations have largely not leveraged CHNAs as a systematic tool for holding health systems accountable to identified community needs.
Simultaneously, augmented intelligence technologies—systems designed to enhance human decision-making through automated data collection, natural language processing, document analysis, and structured communication, operating under human-defined rules and expert oversight—have matured to the point where they can perform at scale what would previously have required prohibitive manual labor.1,2 The convergence of these 3 elements—publicly available CHNA data, established disease advocacy infrastructure, and scalable AI tools guided by human expertise—creates an opportunity to fundamentally transform how community health needs translate into institutional action.
This commentary presents a framework for disease advocacy organizations to deploy AI agents in analyzing CHNAs, generating targeted recommendations to health systems, and establishing public accountability mechanisms. Using Alzheimer’s disease and dementia care as a primary case example, it describes the operational components, anticipated outcomes, adaptability to other conditions, and implications for population health management.
The CHNA Accountability Gap
Under IRS Section 501(r), tax-exempt hospitals must conduct CHNAs at least every 3 years, solicit community input, make the assessment widely available, and adopt implementation strategies to address identified needs. 3 The resulting documents typically contain demographic analyses, morbidity and mortality data, social determinants assessments, existing resource inventories, and prioritized health needs. While the IRS requires hospitals to describe how they will address prioritized needs, there is no enforcement mechanism tied to whether meaningful progress occurs.
This creates what might be termed the CHNA accountability gap: a structured process that reliably identifies community health problems but lacks external pressure to ensure that identification translates into action.4–6 Hospital community benefit officers are required to receive and consider stakeholder input, but the absence of systematic, condition-specific follow-up from credible external organizations means that many identified needs persist across successive CHNA cycles without substantive intervention.
The consequences are particularly stark for conditions affecting aging populations. Alzheimer’s disease and related dementias represent a growing crisis: ∼6.9 million Americans aged 65 and older are living with Alzheimer’s dementia, and prevalence is projected to rise to nearly 13 million by 2050. 7 In Los Angeles County alone, an estimated 190,300 residents aged 65 and older are living with the disease, with Hispanic Americans 1.5 times more likely and Black Americans twice as likely to be affected compared to White Americans. Despite these numbers, CHNAs frequently identify dementia as a community need without hospitals subsequently implementing specialized memory care programs, caregiver support services, or culturally responsive dementia interventions.
Augmented Intelligence as an Advocacy Force Multiplier
Augmented intelligence, in this context, refers to technology-enabled systems that extend and amplify human expertise—performing data collection, analysis, and communication tasks at scale under human-defined rules, parameters, and oversight. Rather than operating autonomously, these systems function as structured tools guided by clinical experts, advocacy professionals, and organizational leadership who establish analytical criteria, approve outputs, and make all consequential decisions. For the CHNA advocacy framework, this human-guided approach operates across 4 functional domains.
Data collection and analysis
The agent identifies hospitals within a defined geographic region using publicly available directories and IRS filings, then retrieves their most recent CHNAs—typically PDF documents posted on hospital websites or filed with the IRS. Using natural language processing, the agent parses each CHNA to extract condition-specific content: identified prevalence data, community needs related to the target condition, existing hospital programs, noted service gaps, and stated implementation priorities. For an Alzheimer’s-focused campaign, the agent would identify whether each hospital’s CHNA references dementia prevalence, memory care access, caregiver support programs, early detection protocols, or partnerships with dementia-focused organizations. This analysis is cross-referenced against evidence-based practice recommendations—such as the Alzheimer’s Association’s Dementia Care Practice Recommendations—to identify both acknowledged and unacknowledged gaps. 8
Personalized communication and outreach
Based on the gap analysis, the agent generates personalized letters addressed to hospital leadership—specifically community benefit officers, quality directors, or chief medical officers. Each letter references the specific CHNA findings for that hospital, acknowledges existing programs, identifies concrete opportunities for improvement, and proposes evidence-based interventions. Critically, these letters are issued under the branding and authority of the advocacy organization, not as unsolicited individual correspondence. The difference in institutional response between a letter from an unknown individual and a letter from a recognized national advocacy organization is substantial—the latter is treated as legitimate stakeholder input within the IRS 501(r) framework.9,10 Evidence from stakeholder engagement research consistently demonstrates that written outreach from credible, mission-aligned national organizations receives substantive institutional review, particularly within regulatory compliance contexts where hospitals are obligated to document their responsiveness to community stakeholders. Beyond letters alone, advocacy organizations may further amplify accountability pressure through coalition endorsements from multiple organizations representing the same condition, targeted media engagement, and formal public comment submissions during accreditation processes or state regulatory review cycles. It should be acknowledged, however, that the capacity and willingness of individual advocacy organizations to participate in this framework will vary. Resource constraints, legal risk tolerance, and organizational priorities may limit engagement across all potential participants, and not all organizations—even well-established national ones—will have the staff, funding, or institutional bandwidth to engage at the full level the framework envisions. The framework should therefore be designed to accommodate tiered participation, with meaningful impact achievable even with partial adoption.
Public accountability through scorecards
The framework’s most powerful component is the integration of public performance reporting. The agent compiles a preliminary rating for each health system based on CHNA content, publicly available quality metrics (such as Hospital Consumer Assessment of Healthcare Providers and Systems [HCAHPS] patient experience scores), and condition-specific indicators. The advocacy organization reviews and finalizes these ratings, then announces them publicly through press releases, website-hosted scorecards, and social media. This creates a “carrot-and-stick” dynamic: hospitals receive constructive recommendations first, accompanied by the understanding that their responsiveness will factor into a publicly visible rating. The existing literature on public performance reporting in health care demonstrates that transparent scoring drives behavioral change in 30%–60% of targeted providers, motivating quality improvement activities, resource reallocation, and program development.9,11–13
Response tracking and iterative engagement
The agent monitors responses—acknowledgments, meeting requests, public statements, policy changes—and generates aggregated reports showing regional patterns, response rates, and before-and-after comparisons of hospital actions. This iterative loop allows the advocacy organization to update scorecards, escalate engagement with nonresponsive systems, and document systemic patterns that inform policy advocacy at the state and federal levels.
Case application: Alzheimer’s and dementia care in Los Angeles County
Los Angeles County illustrates both the need and the opportunity. With 190,300 residents aged 65 and older living with Alzheimer’s dementia and a population that is 48.3% Hispanic, 25.2% White, and 14.8% Asian, the county faces compounding challenges of high prevalence, significant health disparities, and cultural and linguistic barriers to care access.
A comprehensive needs assessment reveals that while adequate medical diagnostic resources exist—including the UCLA Alzheimer’s and Dementia Care Program, Rancho Los Amigos Alzheimer’s Center, and multiple Program of All-Inclusive Care for the Elderly (PACE) programs—critical gaps persist in several domains. Respite care is severely limited, with long waiting lists and insufficient overnight and weekend options. Despite nearly half the population being Hispanic, Spanish-language programming remains inadequate. Early-stage dementia services—cognitive stimulation programs, peer mentoring, and social engagement activities for individuals newly diagnosed—are scarce. Services concentrate in affluent areas, leaving low-income and geographically remote communities underserved.
An AI agent deployed by an organization like Alzheimer’s Los Angeles could systematically review CHNAs across the county’s health systems, identifying which hospitals acknowledge dementia as a priority need, what specific services they offer or plan, and where gaps align with the needs assessment findings. Personalized letters would propose targeted interventions: establishing culturally responsive Memory Café programs, developing bilingual care coordinator positions, creating specialized respite services for individuals with challenging behaviors, and partnering with community health worker programs in underserved neighborhoods. The accompanying public scorecard would rate each system’s dementia readiness and responsiveness, providing both a tool for community members choosing health services and a motivator for institutional improvement.
Anticipated Outcomes and Evidence Base
The projected impact of this framework draws on evidence from 3 converging domains: nonprofit advocacy campaign effectiveness, public performance reporting in health care, and AI-augmented process automation. 14
Generic unsolicited letters to hospitals—even well-crafted ones—typically yield response rates of 10%–30%. However, letters bearing the imprimatur of a recognized advocacy organization, paired with the prospect of public scoring, fundamentally alter the engagement calculus. Based on evidence from public reporting and stakeholder pressure campaigns, the framework projects overall engagement rates of 40%–70% (encompassing acknowledgments, meeting requests, or substantive replies), with 30%–60% of targeted systems making concrete commitments or implementing changes. These projections align with systematic reviews showing that public performance scorecards drive quality improvement activities in a majority of targeted providers and that transparent reporting, while producing modest direct effects on patient choice and clinical outcomes, generates substantial internal motivation for institutional improvement.9,11,12,15,16
The AI agent’s contribution to these outcomes is primarily one of scale and consistency. What would require hundreds of person-hours of manual research, document analysis, and letter drafting for a single metropolitan area can be accomplished in hours with appropriate automation. This efficiency gain allows advocacy organizations to conduct comprehensive regional campaigns rather than piecemeal outreach, maintain consistent follow-up over the multiyear CHNA cycle, and generate the aggregated data necessary for both public reporting and policy advocacy.
Realistic expectations must acknowledge limitations. Not all hospitals will engage substantively; some will issue neutral acknowledgments or react defensively. Smaller health systems with limited community benefit infrastructure may lack the capacity to implement recommendations regardless of motivation. Ratings must be evidence-based and methodologically defensible to maintain credibility and avoid legal risk. Success ultimately depends on sustained follow-up, coalition building, and the advocacy organization’s established reputation within the target region.
Adaptability Across Conditions and Organizations
The framework’s core mechanism—AI-driven CHNA analysis paired with credible advocacy outreach and public accountability—is condition-agnostic. It is applicable wherever 3 conditions converge: publicly available community needs data, an organized advocacy infrastructure, and identifiable gaps between documented needs and institutional response.
Neurological and cognitive conditions beyond dementia present immediate opportunities. The Parkinson’s Foundation could deploy agents analyzing CHNAs for rehabilitation access, movement disorder specialist availability, and speech therapy services. The National Multiple Sclerosis Society could target infusion care access and care coordination gaps. The Brain Injury Association of America could evaluate post-acute rehabilitation and community reintegration support.
Mental health conditions represent perhaps the most impactful application. The National Alliance on Mental Illness and Mental Health America could analyze CHNAs for crisis intervention capacity, primary care integration with behavioral health, substance use disorder treatment access, and community mental health infrastructure—gaps that CHNAs almost universally identify, yet that frequently persist across assessment cycles.
Chronic metabolic and cardiovascular diseases offer another natural fit. The American Diabetes Association and American Heart Association could focus on preventive care programming, health education infrastructure, chronic disease management capacity, and equity in screening and treatment access. The American Cancer Society could evaluate screening program adequacy, survivorship support, and care navigation for underserved populations.
Aging-specific organizations such as AARP and the National Council on Aging could target geriatric care gaps, fall prevention programming, end-of-life care access, and integration with community-based services—particularly relevant given the aging of the population and the increasing recognition that social determinants profoundly shape health outcomes for older adults. Organizations focused on emerging conditions, including long COVID advocacy networks and rare disease coalitions operating through the National Organization for Rare Disorders, could similarly adapt the framework to their specific needs assessment priorities.
Ethical and Practical Considerations
The deployment of AI agents in health advocacy raises important considerations that must be addressed proactively. First, all data used in the framework is publicly available—CHNAs are required to be accessible to the public, and quality metrics referenced in scorecards are drawn from public reporting programs. No protected health information is involved. Nevertheless, compliance with anti-spam regulations (specifically the CAN-SPAM Act) requires that automated email communications include opt-out mechanisms and accurate sender identification.
Second, human oversight must remain central. AI agents can perform data collection, analysis, and draft generation with remarkable efficiency, but final letter approval, scorecard validation, and strategic decisions about public release must involve human judgment. This is particularly important for scorecard accuracy: ratings that are perceived as unfair, methodologically flawed, or defamatory risk undermining the advocacy organization’s credibility and exposing it to legal challenge. Specific legal risks merit deliberate attention: Defamation claims may arise if scorecard ratings are perceived as factually inaccurate or misleading regarding a hospital’s programs or services; false light claims could emerge from the juxtaposition of comparative ratings without adequate contextual disclosure; and advocacy organizations operating across multiple jurisdictions should evaluate whether rating and outreach activities implicate state unfair business practice statutes. Health care legal counsel—not only general nonprofit counsel—should review scorecard methodology and communication templates prior to any public release. Organizations are also advised to establish a formal pre-publication appeals process by which health systems may contest preliminary ratings and submit clarifying evidence, reducing both legal exposure and reputational risk to the advocacy organization. An internal review process that includes clinical experts and legal counsel is essential before any public rating is issued.
Third, the framework should be deployed by established nonprofit organizations with genuine advocacy missions, not by commercial entities seeking competitive intelligence or by individuals pursuing personal grievances. The credibility that drives institutional response depends on the advocacy organization’s reputation, mission alignment, and established relationships within the health care community.
Finally, data quality limitations must be acknowledged. CHNAs vary substantially in depth, methodology, and specificity across institutions.4,6,17 Automated Natural Language Processing (NLP) analysis may miss nuances, misclassify content, or generate inaccurate characterizations of hospital programs. Iterative testing, starting with pilot deployments targeting a small number of hospitals before scaling regionally, allows for calibration and error correction. To address the heterogeneity of CHNAs directly, the framework should incorporate a formal audit mechanism built into both pilot and scaled operations. During the pilot phase, a clinical and methodological review committee should evaluate a random sample of AI-generated CHNA analyses to assess accuracy, completeness, and appropriate classification of hospital programs relative to condition-specific benchmarks. As the framework scales, structured audits—conducted at each CHNA renewal cycle or annually for active programs—should be embedded in program governance, with findings reported publicly alongside scorecard data to maintain transparency and credibility. Oversight responsibility for this function should rest with the participating advocacy organization’s quality or program committee, supported by an independent clinical advisory panel with expertise in both the target condition and health care data analytics.
Implications for Population Health Management
This framework carries several implications for the broader field of population health management. It demonstrates how AI can be responsibly integrated into public health advocacy to amplify organizational capacity without replacing human judgment or stakeholder relationships. 14 It proposes a mechanism for converting the CHNA process from a compliance exercise into an active driver of community health improvement—aligning hospital community benefit activities more closely with the population health needs that assessments are designed to identify.
Perhaps most significantly, the framework positions disease advocacy organizations as critical intermediaries between community health data and institutional action. In the current landscape, CHNAs often exist in a vacuum—produced by hospitals, filed with regulators, and occasionally referenced by researchers but rarely leveraged by the condition-specific organizations best positioned to interpret their findings and advocate for targeted interventions. By equipping these organizations with AI tools that make systematic CHNA analysis feasible at scale, the framework creates a new accountability pathway that complements existing regulatory mechanisms.
The aggregated data generated through this process—regional patterns of unmet need, institutional response rates, longitudinal tracking of program implementation—also has substantial value for policy advocacy. State and federal policymakers, public health agencies, and accreditation bodies could use these data to evaluate whether the CHNA process is achieving its intended purpose and to target resources or regulatory attention where accountability gaps persist.
Conclusion
The convergence of publicly mandated community health needs data, established disease advocacy organizations, and scalable augmented intelligence technology creates an unprecedented opportunity to close the gap between identifying population health needs and acting on them. The framework described here—augmented intelligence–driven CHNA analysis, institutionally branded advocacy outreach, and public performance scoring—transforms what has been a largely passive compliance process into an active mechanism for health system accountability and community health improvement. While challenges of data quality, institutional resistance, and ethical deployment must be addressed, the potential to drive meaningful change in how health systems respond to community needs—particularly for conditions like Alzheimer’s disease that disproportionately affect underserved populations—merits serious attention from population health researchers, advocacy organizations, and health system leaders alike.
Authors’ Contributions
R.G.S.: Conceptualization, literature, review, revision and writing.
Footnotes
Acknowledgments
The author acknowledges the contributions of AI-assisted tools in the editing of this article. All content has been created, reviewed, and approved by the author.
Author Disclosure Statement
The author reports no conflicts of interest relevant to this work.
Funding Information
No funding was received for this article.