experience to ensure synergy between scientific priorities and the priorities of those directly affected by dementia.

ADVANCING KEY SCIENTIFIC PRIORITIES RELATED TO PREVENTING AND TREATING AD/ADRD

The past few decades have brought significant advances in the understanding of AD/ADRD and in the development of tools and methods that can drive further progress. Notable milestones include the ability to detect specific AD-related pathologies (amyloid and tau) years before symptoms emerge, the discovery of many new genes linked to AD/ADRD that shed

light on pathogenic mechanisms, and the recognition that pathologies previously thought to distinguish different forms of dementia often co-occur (NIH, 2023a). While encouraged, the committee also found notable gaps in scientific knowledge and research capabilities during its review of the AD/ADRD research landscape. The following scientific gaps represent key bottlenecks that significantly impede progress toward preventing and treating AD/ADRD.

• There is a lack of rigorous evidence to support the identification of effective public health strategies for preventing AD/ADRD. Epidemiological research has yielded a large set of dementia risk factors (e.g., health behaviors, social isolation, socioeconomic disadvantage) based on statistical correlations. However, gaps in longitudinal data, data infrastructure, and study designs have resulted in limited understanding of causal relationships and the identification of exposures and system-level factors that, if mitigated, would have the greatest effect on the incidence of dementia. This impedes the development of effective public health approaches that could be implemented at a population scale to promote brain health across the life course and prevent AD/ADRD in diverse and especially disproportionately affected populations.
• There is an incomplete understanding of the biological basis and multiple etiologies underlying cognitive decline and dementia, as well as the mechanisms of resilience. There is still much that is not understood about AD/ADRD, conditions that may be characterized by decades of life-course health insults to brain and peripheral systems contributing to the development and detectable progression of neuropathology in the absence of clinical symptoms. Key knowledge gaps include how life-course exposures and other risk factors relate to pathobiology in diverse populations and the connections between molecular pathways that contribute to AD/ADRD and resilience (the maintenance of cognitive function despite the presence of brain pathology). This lack of understanding impedes drug discovery and the development of effective preventive and therapeutic intervention strategies. It also makes it difficult to predict the effect of risk factor reduction and drugs focused on single-pathology targets on dementia incidence.
• There is a lack of effective, validated, and accessible tools and methods (e.g., novel biomarker tests, digital assessment technologies) for detecting early changes in brain health and accurately diagnosing, subtyping, and monitoring AD/ADRD in diverse populations, This foundational capability is essential to trace how the natural history of AD/ADRD differs from “healthy” brain aging. Progress made

• in the development of tools for AD needs to be expanded beyond amyloid and tau and extended to the multiple etiologies leading to dementia. The current capability gap impedes efforts to measure disease incidence and prevalence, to intervene early when chances of preventing and mitigating disease are greatest, to detect when treatments modify the trajectory of AD/ADRD, and to target specific interventions to the right populations (precision medicine).

Within each of these major scientific gaps of knowledge, there is also promising research that suggests opportunities to break current bottlenecks as new discoveries emerge. The discovery of imaging and fluid biomarkers for AD has catalyzed a shift in phenotyping procedures used in research and needs to be tested now in clinical practice. With the identification of risk genes/loci and the discovery of additional biomarkers, particularly for related dementias, current barriers to early detection, diagnosis, prognosis (e.g., the likelihood of progression to clinical dementia), and longitudinal monitoring may be overcome, and it will be possible to quantify and better understand multiple etiology dementia. The combination of digital tools and computational methods such as artificial intelligence/machine learning (AI/ML) that may be able to identify changes in traits (e.g., speech, gait, sleep behavior) that may precede current measures of cognitive decline similarly shows promise for enabling early detection of changes in brain health, early diagnosis, and prognosis (Amini et al., 2024). Digital tools have also opened new opportunities for passive and remote data collection and are changing the way investigators engage with study participants, particularly those from underresourced and underrepresented populations, and the public (Kaye et al., 2021).

Investments in basic science and longitudinal cohort studies have led to a significant expansion of the therapeutic pipeline with novel promising interventions that are not specific to any single dementia type by uncovering shared molecular pathways contributing to AD/ADRD (e.g., autophagic and lysosomal, immune, metabolic, mylelination), as well as resilience factors (Cummings et al., 2024). Multiomics methods¹ are creating new opportunities to evaluate disease mechanisms in diverse populations (Reddy et al., 2024) and to identify molecular disease subtypes and endophenotypes (Fang et al., 2020), thereby creating the foundation for precision medicine approaches to prevention and treatment in the future. Increased understanding of the links between AD/ADRD and chronic diseases such as hypertension and diabetes (Nagar et al., 2022), along with encouraging

___________________

¹ Multiomics methods involve the integrative analysis of multiple “omics” datasets, such as those generated from genomic, proteomic, transcriptomic, epigenomic, and metabolomic methods.

evidence for multicomponent interventions focused on health behaviors, has highlighted the potential for public health strategies to reduce dementia risk.

Scientific Priorities for Advancing the Prevention and Treatment of AD/ADRD

Building on the aforementioned examples of momentum and lines of promising research, the committee identified 11 research priorities and associated near- and medium-term scientific questions that it believes should be a focus of NIH-funded AD/ADRD biomedical research for the next 3 to 10 years. These research priorities, which are summarized in Recommendation 1 and detailed in Table 5-1, fall into three broad areas:

1. Quantify brain health across the life course and accurately predict risk of, screen for, diagnose, and monitor AD/ADRD.
2. Build a more comprehensive and integrated understanding of the disease biology and mechanistic pathways that contribute to AD/ADRD development and resilience over the life course.
3. Catalyze advances in interventions for the prevention and treatment of AD/ADRD spanning from precision medicine to public health strategies.

NOTE: The numbering of research priorities in this table reflects the numbering in the report chapters.

The committee acknowledges that NIH has already made investments in each of these priority research areas to varying degrees. Given the breadth of the NIH AD/ADRD research portfolio, it is unsurprising that the committee did not identify any research priorities for which there had been no prior NIH investment. In some cases, research priorities identified by the committee, such as the development of biomarkers for monitoring brain health and the identification of factors driving risk in diverse populations, are already the focus of major NIH-funded research programs and initiatives, many of which are described in this report. Other identified research priorities, such as the characterization of the exposome and gene–environment interactions, the development of integrated molecular and cellular causal models, and the development of digital tools, represent scientific areas of more recent or limited NIH investment. Relatedly, efforts to achieve associated near-term research opportunities, which are detailed in the right-hand column of Table 5-1, may indeed be underway but have not yet been fully realized. Significant investment in the totality of the priority research areas is needed to address the knowledge gaps laid out in this report. Critically, beyond financial investment, success in tackling each of these research priorities will require an emphasis on the intentional expansion of research efforts beyond AD and the inclusion of diverse and understudied and/or disproportionately affected populations.

Importantly, Table 5-1 is not intended as a prescribed research agenda. Nor is the identification of these priority areas meant to imply that lines of scientific inquiry outside of these areas are not of value or that work in all other areas should be suspended. There is a great deal of uncertainty in the process for scientific investigation regarding which discoveries from current research will lead to transformational advances in the future and no guarantees can be offered regarding the ultimate fruitfulness of any specific line of inquiry. The committee’s intention, however, is that the priorities will be used as a guide in the rebalancing of NIH funding for AD/ADRD. Closing the scientific knowledge gaps raised by these priorities can occur by working to answer the committee’s proposed scientific questions and acting on opportunities to overcome barriers to progress, as detailed in the recommendations that follow.

• Aging, Demographics, and Memory Study as a supplement to the Health and Retirement Study); and
• collaborative efforts to harmonize outcomes (e.g., the Harmonized Cognitive Assessment Protocol) to facilitate cross-study comparisons and multicohort analyses.

Recognizing these prior investments, the committee encourages NIH to continue and expand support for longitudinal AD/ADRD research that can fill data gaps and address the scientific questions included in Table 5-1. In addition to establishing new cohorts that can meet the need for diverse and representative populations in dementia research, this should include a concerted effort to identify other existing cohorts—including those funded by agencies other than the National Institute on Aging (NIA) and the National Institute of Neurological Disorders and Stroke (NINDS), as well as international studies—to which a focus on AD/ADRD could be added through supplemental funding and other appropriate NIH funding mechanisms. Such efforts may provide an opportunity to pilot novel tools such as biomarker tests and digital technologies. Additionally, NIH should ensure funding opportunities are designed to maximize the insights from longitudinal research through attention to data accessibility and harmonization and the collection and storage of data (e.g., digital, exposure) and biosamples (see Recommendations 7, 8, and 9). Proactive coordination and planning through the convening of investigators from different cohort studies, before study initiation whenever possible, can help to identify approaches that would enable data access, interoperability, and harmonization.

Accelerating Translational and Clinical Intervention Research

Decades of research and hundreds of clinical trials have yielded only a limited number of treatments for Alzheimer’s disease (AD) that offer modest clinical benefits (Boxer and Sperling, 2023; Cummings et al., 2024; Kim et al., 2022), and no treatments have been approved by the U.S. Food and Drug Administration (FDA) for related dementias (Liu et al., 2019; MacDonald et al., 2022; Nag et al., 2020), beyond those for managing symptoms. Evidence for some nonpharmacological approaches to preventing cognitive decline and dementia (physical activity, cognitive training), while encouraging, has been inconclusive (NASEM, 2017), leaving the public with much uncertainty about steps they should take to protect their cognitive function as they age. This lack of progress toward effective strategies for preventing and treating AD/ADRD reflects the complex and multifactorial pathobiology of this group of diseases, but it also underscores the need to accelerate the translational research yielding novel targets for interventions and to expand and improve the efficiency of clinical trials.

Many entities (government, private, philanthropic, and academic) contribute to research for advancing interventions for AD/ADRD with complementary resources and expertise. NIH plays a critical role in this complex research ecosystem by funding research on interventions and certain trial designs that may be less appealing to industry owing to financial risk or the lack of financial incentives, incentivizing industry participation in collaborative efforts designed to develop and bring new and combination interventions to scale (see Recommendation 10), and supporting basic and translational research (e.g., target identification and validation) that feeds into the private-sector drug development pipeline. As discussed in Chapter 3, NIH has made significant infrastructure investments (e.g., Accelerating Medicines Partnership® Program for Alzheimer’s Disease [AMP-AD], Target Enablement to Accelerate Therapy Development for Alzheimer’s Disease [TREAT-AD], Model Organism Development and Evaluation for Late-onset Alzheimer’s Disease [MODEL-AD]) to generate a pipeline that can translate discoveries from basic research into candidates that can be evaluated in clinical trials. With the necessary infrastructure in place, there is now an opportunity to scale and diversify these efforts to a broader set of risk factors and pathways (e.g., neuroinflammation, cellular senescence, lysosomal dysfunction, mitochondrial dysfunction).

provides multidisciplinary training for current and future principal investigators on the design, conduct, and analysis of clinical trials. IMPACT-AD also works to strengthen the broader clinical trial workforce through dedicated training opportunities for clinician-researchers and clinical trial support staff (Berkness et al., 2021).

The expanded use of innovative clinical trial designs, described in Chapter 4, has the potential to significantly accelerate clinical research. Master protocols—trial protocols for use with multiple substudies (FDA, 2022)—and platform, combination, and adaptive trial designs create efficiencies and enable a shift away from a clinical research paradigm focused on testing individual interventions in single populations sequentially.

Additional opportunities to improve trial efficiency can be realized by ensuring a trial-ready pool of research participants and a ready clinical trial infrastructure. Recruitment and enrollment represent major bottlenecks in the clinical trial pipeline (Langbaum et al., 2023). As has been done in the cancer field, NIH-funded Alzheimer’s Disease Research Centers (ADRCs) and other AD/ADRD-focused centers should be evaluated and held accountable for enrolling participants in clinical trials.

A key strategy to improve trial efficiency at the recruitment phase is to reduce screen failure.² With master protocols, a single recruitment pool can be used to populate multiple substudies. Volunteers that may be ineligible for one substudy could potentially be enrolled in a different substudy under the same protocol. Highly phenotyped and increasingly diverse cohorts represent invaluable pools of prescreened participants (Gregory et al., 2022). Examples include ALLFTD (discussed in Chapter 2) and the North American Prodromal Synucleinopathy (NAPS) study, which enrolls people with rapid eye movement (REM) sleep behavior disorders who are at risk of developing LBD, Parkinson’s, or other neurological disorders (NAPS Consortium, 2024). Similarly, the Trial-Ready Cohort-Down Syndrome collects longitudinal data (e.g., blood and cognitive tests, imaging) on people with Down syndrome to fast-track the enrollment process as soon as they are eligible and matched with a qualifying clinical study (TRC-DS, 2024). While biases in the cohort populations need to be considered, drawing from such pools could accelerate startup.

Finally, there needs to be consideration of opportunities to use technology to move from screen fail to screen enroll. While registries³ are one mechanism to track volunteers who did not meet screening criteria but

___________________

² Screen failure occurs when a potential participant is screened for but is not able to enroll in the trial (Parekh et al., 2022).

³ NIA maintains a list of registries and matching services for AD/ADRD clinical trials at https://www.nia.nih.gov/health/clinical-trials-and-studies/registries-and-matching-services-clinical-trials (accessed October 19, 2024).

may be eligible for other studies (e.g., people who exhibit signs of cognitive impairment but fail the screen for AD pathology), other technologies, such as social engagement platforms, should be explored. There is also a need for studies on best practices for community screening and referral.⁴

A ready clinical trial infrastructure can help to speed up other aspects of the startup phase. These include centralized support functions, such as centralized institutional review boards (now required by NIH for most multisite trials) and contracting; systems for decentralized screening (online or at local community centers); and systems for electronic and staged consent processes. The development of clinical trial networks that use a hub-and-spoke model to centralize some core infrastructure can reduce the pragmatic challenges and burdens for investigators at trial sites embedded in communities (e.g., federally qualified health centers). To support the development of such networks, ADRCs could play a role in creating registries of regional clinical trial sites, and NIH-funded clinical trial consortia, if adequately supported, could provide training for clinical trial sites to disseminate knowledge, standards, and best practices.

___________________

⁴ While beyond the scope of this report, it should be noted that an important issue that arises with regards to screening of individuals for participation in clinical research is the return of screening results to the individuals (NASEM, 2018).

STRATEGIES FOR ADDRESSING CROSSCUTTING BARRIERS THAT IMPEDE PROGRESS ON AD/ADRD SCIENTIFIC PRIORITIES

The committee was asked to identify key barriers to advancing AD/ADRD prevention and treatment and to highlight opportunities to address these barriers to catalyze advances across the field. In its examination of the AD/ADRD research landscape, several impediments to progress were consistently identified across the continuum from basic to clinical research. These crosscutting barriers include

• siloing within the AD/ADRD field and across related domains of research (e.g., aging, neurodegenerative diseases more broadly, exposure science);
• insufficient population representativeness and generalizability of research;
• inadequate infrastructure and support for management and analysis of data, samples, and knowledge generated from AD/ADRD research; and
• inadequate support for innovative methods capable of realizing transformational progress.

The sections below discuss opportunities and strategies for addressing these barriers. As detailed in each of the sections, the committee recognizes the significant NIH investment to address each of these key barriers. Examples of NIH activities highlighted in the sections below are meant to be illustrative and do not represent a comprehensive cataloging of such efforts as the committee was not charged with a review of NIH’s programs. It should also be acknowledged that many barriers are not unique to dementia research, and other scientific fields are also working to overcome similar challenges. Accordingly, in considering the implementation of the recommendations below, NIH and other research funders should

continuously monitor the broader research landscape for examples of how such challenges have been successfully tackled in other fields and consider opportunities to apply those strategies in AD/ADRD research.

Breaking Down Silos Through Collaborative, Multidisciplinary Research

The heterogeneity of AD/ADRD, the prevalence of mixed pathologies, and the multifactorial and intersecting nature of the diverse pathways that lead to disease all suggest that the path to effective strategies for preventing and treating this group of neurodegenerative diseases lies in collaborative, multidisciplinary research. Yet, throughout its information-gathering process the committee encountered numerous silos, commonly reinforced by funding structures, that impede efforts to elucidate the biological basis of AD/ADRD and advance prevention and treatment. Current funding strategies that target individual diseases, which have historically favored AD, fail to address the reality of overlapping and mixed pathologies that contribute to neurodegenerative disease, and they have contributed to the current dearth of effective therapies for related dementias. There have been few efforts to develop a larger integrated model of aging and neurodegenerative disease despite clear overlap in research endeavors and shared mechanisms (see Chapter 3). As discussed in Chapter 4, research on pharmacological and nonpharmacological interventions are not well integrated, and as a result there have been few efforts to date to evaluate the effect of combination approaches despite a high likelihood that risk reduction and drug therapies will both be necessary elements of a strategy to reduce the incidence and impact of dementia. Moreover, the efforts of federal agencies supporting related areas of research are not adequately coordinated, resulting in missed opportunities to leverage synergies and effectively use existing investments in studies and infrastructure. Innovative funding strategies, such as multi-institute research consortia and public–private partnerships, and other incentives, as well as the application of collaborative research mechanisms, and greater coordination and integration of research and infrastructure are needed to address the current siloing of research and accelerate the development of interventions for preventing and treating AD/ADRD. Disincentives to team approaches, such as academic promotion structures, also need to be addressed.

Funding opportunities that encourage collaboration across disease areas by bringing together multidisciplinary teams have the potential to accelerate the development of not only interventions that target common underlying mechanisms but also disease-agnostic resilience mechanisms, as discussed in Chapter 3. Supporting collaboration across disciplines also facilitates the application of approaches and technologies to address research questions and technical challenges in novel ways. Multi-institute research consortia represent one mechanism for fostering collaborative research and bridging

the divide between basic and clinical research (Gladman et al., 2019). For example, the Biomarkers for Vascular Contributions to Cognitive Impairment and Dementia Consortium (MarkVCID) was established to advance the discovery and validation of biomarkers for small vessel diseases of the brain involved in cognitive impairment and dementia, as cerebrovascular small vessel disease is a commonly identified pathology in mixed dementia (Greenberg, 2017; MarkVCID, 2017). Ultimately, the consortium seeks to deliver biomarker kits that can be used in intervention trials, thereby translating basic science findings into clinical research (Gladman et al., 2019). Diverse VCID is another example of a multi-institute research program focused on understanding the role that cerebrovascular disease plays in AD/ADRD for diverse populations with the goal of improving diagnosis and treatment (Diverse VCID, 2024). Multi-institute consortia often have the benefit of effectively using existing, mature research infrastructure to scale up research efforts. Coordinating centers play a key role in such consortia to centralize resources and facilitate harmonization, coordination, and data sharing across the multiple participating research institutions. When supported by multiple funders, research consortia also provide opportunities to align and better use existing resources and future investments across funding organizations.

Multi-institute research collaborations should not be limited to U.S. institutions. Dementia is a global challenge—the burden of which is borne in large part by people living in low- and middle-income countries (LMICs)—and it will not be overcome by the siloed efforts of individual countries (Nature Medicine, 2023). From a health equity perspective, international collaborations can help to address the problems of overrepresentation of some populations in AD/ADRD research and the underrepresentation of many global populations living outside North America and Western Europe. Underrepresented populations include those within LMICs but also in some high-income countries (e.g., within Asia and the Middle East). Such global collaborations can also help to uncover rare genetic variants and answer important scientific questions regarding the relative contributions of genetic ancestry and sociocultural factors (e.g., social determinants of health) to dementia risk and resilience. It should be acknowledged, however, that current international laws and regulations, including but not limited to the General Data Protection Regulation of the European Union, pose a formidable impediment to the reciprocal exchange of data and biological samples with researchers from other countries. While not insurmountable (and not limited to AD/ADRD), these issues are complex and require careful legal analysis. Investigator-level relationships can lead to some workarounds, but ultimately these barriers need to be addressed at the level of national governmental leaders. As a major global funder of biomedical (including AD/ADRD) research and intellectual leader, NIH can spearhead

efforts to overcome these barriers by engaging its counterparts in other countries and the broader research community to understand their needs and jointly develop practical mid- to long-term solutions. Such solutions will need to address mechanisms for data access (e.g., cloud-based mechanisms, federated learning platforms) and security (e.g., encryption, use of synthetic data), and the more challenging issues of biosample sharing.

Public–private partnerships create unique opportunities for cross-discipline collaboration (within the United States and globally) and are established specifically to more effectively leverage the respective talents of investigators in academia and industry toward a shared goal. A notable example in the AD/ADRD field is AMP-AD (see Box 3-4 in Chapter 3). Government and philanthropic funding for such partnerships can incentivize industry engagement in activities that might otherwise be considered too financially risky, as well as encouraging innovation (e.g., innovative trial designs) (Boxer and Sperling, 2023). While there are multiple mechanisms for establishing public–private partnerships, the Foundation for the National Institutes of Health (FNIH)—a nonprofit organization—specifically focuses on convening partners, including NIH, academic institutions, industry, philanthropy, and advocacy organizations (FNIH, 2023) and may be positioned to facilitate collaboration without creating financial conflicts of interest for academic researchers. Current gap areas that may benefit from public–private partnerships include efforts to develop combination interventions that include pharmacological and nonpharmacological components (see Chapter 4), and the development of a platform for real-world data collection (see Chapter 2).

Challenge programs are another powerful mechanism for facilitating interdisciplinary approaches while simultaneously pushing boundaries and driving further innovation in AD/ADRD research. Such programs often set ambitious goals and use prize money to incentivize investigators to tackle complex problems from novel perspectives. By encouraging rapid iteration and establishing feedback loops, challenge programs can push researchers to develop and refine ideas and strategies in relatively short periods of time. A recent example is the Pioneering Research for Early Prediction of Alzheimer’s and Related Dementias Eureka (PREPARE) Challenge, a multiphase competition that was launched by NIA in fiscal year 2023 with support from the NASA Tournament Lab (Driven Data, 2024).⁵ The objective of the challenge is to spur innovation in data science to advance solutions for accurate prediction of AD/ADRD in diverse and historically underrepresented populations. PREPARE is designed to bring teams together to explore how AI/ML and other computing approaches can be used to collect and analyze data

___________________

⁵ More information on the PREPARE challenge is available at https://www.drivendata.org/competitions/253/competition-nih-alzheimers-adrd-1/ (accessed July 2, 2024).

in ways that could advance the development of tools and technologies for clinical and research use in predicting disease. By encouraging team science approaches and risk taking, challenge programs such as PREPARE stimulate creativity within the scientific community and accelerate the translation of discoveries into tangible solutions. They can also serve as a means of bringing new talent into a field. Moreover, these initiatives often cultivate a culture of innovation within the research workforce by establishing a supportive ecosystem where participants can exchange ideas, bring to bear diverse expertise, and forge partnerships that transcend institutional and geographic boundaries. While challenge programs are not the answer to every knowledge gap, NIH should continue to employ this model where applicable.

Collaborative research mechanisms that engage people living with AD/ADRD and members of the community in which the study is being conducted are critical to ensuring that research on prevention and treatment strategies is conducted in alignment with what is important to those living with, or who are at risk for, the diseases. Such mechanisms come in many forms with the level of engagement varying across a continuum, from advisory bodies (e.g., patient and community advisory boards, focus groups) to community-based participatory research and coproduction models featuring shared decision making between researchers and participants at all phases (Reyes et al., 2023; UK Research and Innovation, 2024). The mechanism employed should be informed by the nature of the study. Advocacy and other organizations that serve people with AD/ADRD and their care partners are also key partners and can facilitate opportunities for the engagement of people with lived experience in research. The benefits are myriad and include enhancing understanding of the experiences, needs, and values of those being asked to participate in research; improving equity in AD/ADRD research; facilitating recruitment and retention of hard-to-reach populations; and informing researchers as to the acceptability and feasibility of both the intervention strategies and the research methods (Kowe et al., 2022; Reyes et al., 2023).

Coordination and collaboration at the program and project level is facilitated and may be incentivized by analogous efforts at the federal level. Collaboration among NIH institutes, centers, and offices and with other federal agencies, such as the Centers for Disease Control and Prevention (CDC), the Centers for Medicare & Medicaid Services (CMS), and the Department of Veteran’s Affairs, occurs through a variety of mechanisms with variable levels of formality (e.g., National Alzheimer’s Project Act Federal SubGroup, NIH-Wide Microphysiological Systems working group) (NIH, n.d, 2024a).

Greater coordination is needed to reduce siloing across the existing major NIH research investments to leverage the knowledge that has

been generated by these individual efforts to advance the AD/ADRD field more broadly. For example, AMP-AD functions in parallel to the AMP for Parkinson’s Disease and Related Disorders (AMP-PDRD) and the AMP for Amyotrophic Lateral Sclerosis (AMP-ALS). Similarly, other major AD/ADRD-related programs for genetics research, biobanking, and data infrastructure funded by NIA, NINDS, and other NIH institutes are not well integrated. Breaking down these silos to maximize return on these research investments will require action beyond that which can be driven at the investigator level and will instead require NIH and others leading these programs to actively incentivize opportunities for coordination and integration and the breakdown of any logistical and technical barriers.

Recognizing the existing mechanisms already in place and the challenges of establishing new interagency bodies (e.g., time for agency personnel), the committee encourages NIA, NINDS, and other NIH funders of AD/ADRD research to identify further opportunities to maximally leverage the strengths, resources, and unique capacities of other agencies to advance shared focus areas. For example, existing federal-level collaboration and cofunding among NIA, NINDS, and the National Institute of Environmental Health Sciences (NIEHS) related to exposome research and precision environmental health approaches to AD/ADRD risk reduction and disease prevention (Stetler, 2023) could create new opportunities and incentives for environmental scientists at NIEHS-funded centers to work with ADRCs to advance exposome research related to AD/ADRD. Expanded collaboration across NIA, NINDS, and the National Institute of Mental Health (e.g., building on the joint investments in the Psych-AD program⁶ [NIH, 2023b] and other cofunding opportunities [NIH, 2023c]) can accelerate efforts to elucidate interactions between mechanistic pathways contributing to the development of neuropsychiatric symptoms and those underlying the development of AD/ADRD neuropathology. Such efforts may uncover novel targets or strategies for intervention.

Examples of collaborations with other federal agencies might include (1) collaborating with the U.S. Census Bureau to expand access to federal statistical research data centers (FSRDCs) and facilitate the linkage of multiple data types relevant to AD/ADRD within the FSRDCs, (2) working with CMS or FDA to tie expedited review processes for industry to data sharing policies, and (3) working with the CDC to generate a more robust evidence base for public health-level interventions.

___________________

⁶ More information about Psych-AD is available on the AD-Knowledge Portal at https://adknowledgeportal.synapse.org/Explore/Programs/DetailsPage?Program=Psych-AD (accessed October 19, 2024).

compounding of existing health disparities. As noted in a recent National Academies report, “An equitable clinical research enterprise would include trials and studies that match the demographics of the disease burden under study” (NASEM, 2022, p. 1).

As discussed in Chapter 1, the impacts of dementia are not experienced uniformly across the U.S. population (Brewster et al., 2019) or globally. In the United States, racial and ethnic disparities have continued to persist despite overall decreases in clinical dementia prevalence. Black and Hispanic people are more likely to develop clinical dementia compared to non-Hispanic White people (Chen and Zissimopoulos, 2018), and the lifetime risk for women is about twice that of men (Mielke, 2024). The prevalence of AD/ADRD is also increased in rural and lower-income areas (Powell et al., 2020; Wing et al., 2020). And yet, the populations that are disproportionately affected by dementia are persistently underrepresented in AD/ADRD research, both in observational studies and clinical trials (Gilmore-Bykovskyi et al., 2019; Godbole et al., 2023; Lim et al., 2023). The cascading effects on dementia research are myriad and include

• the limited understanding of the contributing factors (e.g., rare genetic variants) and pathways that increase risk and causally contribute to AD/ADRD in diverse populations, which may result in a focus on risk-reduction strategies and intervention targets that are not likely to be effective for some groups who are at higher risk of dementia;
• effects on the establishment of biomarker cutoffs for diagnosis and normative comparisons; and
• the limited generalizability of research findings (i.e., external validity) (Gianattasio et al., 2021), including safety and efficacy data from clinical trials (Canevelli et al., 2019; Franzen et al., 2022), which raises concerns regarding the potential to implement interventions that are ineffective or even unsafe for certain population groups.

Representative studies that incorporate measures of social determinants of health along with multiomics, cardiometabolic, AD and other dementia biomarkers, and cognition are urgently needed. Intensifying investment in understanding factors that contribute to cognitive decline and impairment in underserved populations will enable the implementation of comprehensive, innovative, accessible, and affordable therapies that will mitigate multiple mechanisms driving AD/ADRD.

Increasing the participation of underrepresented populations in dementia research has been a focus of past recommendations to NIH (NASEM, 2017, 2021), and it is clear that NIA, NINDS, and other funders of dementia research are committed to and actively working on closing this gap (Hodes,

2023). Several of the research implementation milestones established by NIH to support the goals of the National Plan to Address Alzheimer’s Disease (see Chapter 1) specifically address increasing the inclusion of diverse and underrepresented populations in AD/ADRD research.⁷ Many other milestones include a focus on diverse and higher-risk special populations (e.g., individuals with Down Syndrome). Additionally, NIA has indicated that investigator requests to submit large grant applications (i.e., grant application with direct costs totaling $500,000 or more for a single year of support) will receive priority review if they

(1) include proposed planned enrollment tables representative of the population affected by the disease/condition, and (2) are appropriately inclusive of racial and ethnic minority groups; participants across the lifespan; as well as other populations experiencing health disparities (Santora, 2023).

Given the multiple, interrelated factors that are associated with chronic underrepresentation of certain populations (e.g., some ethnic/racial, people with low socioeconomic or educational attainment), achieving greater inclusivity and accessibility in AD/ADRD research will require a multipronged approach. While many of the barriers are well known and NIH appropriately continues to support research to further elucidate these factors (Ashford et al., 2022, 2023; Mindt et al., 2023), such an approach will benefit from the regular analysis of recruitment, enrollment, and retainment outcomes. For example, the identification of factors that result in studies falling short of recruitment goals can help to identify strategies (e.g., oversampling, use of sampling frames) that can be used to overcome persisting impediments while avoiding the introduction of bias from the use of different recruitment practices for different subpopulations (Raman et al., 2021). Such efforts will ultimately help to build the science of recruitment. The implementation of standardized common data elements for recruitment- and enrollment-based factors could enable better ascertainment of biases at specific stages in the recruitment and screening processes (Manly et al., 2021).

Selection processes—resulting from targeted outreach, selective enrollment, and highly patterned attrition—make it difficult to understand how observations in a selected sample (i.e., study population) relate to the general

___________________

⁷ Implementation milestones specifically focused on inclusion of diverse and underrepresented populations in AD/ADRD research include Milestone 1.C: Population Studies: Diverse cohorts; Milestone 12.A: Recruitment: Diverse community partnerships; Milestone 12.K: Health Equity: Inclusion and retention of underrepresented populations in clinical research; and Milestone 12.L: Health Equity: Inclusion of diverse communities in AD/ADRD research. The AD/ADRD research implementation milestone database is available at https://www.nia.nih.gov/research/milestones (accessed July 1, 2024).

population of older adults (Gibbons et al., 2024). Statistical methods can enable generalization from a highly selected sample to more general target populations based on shared variables measured in the selected sample and the target population. Key to these approaches is having measurements that are identical in the highly selected sample and the target population. Many key demographics (e.g., race, ethnicity, age, gender) are measured with standardized questions, but some important characteristics, such as education and measures of health or cognition, may be measured differently across data sources. There has been little attention to using the same measures as are available in a surveillance-type study of the target population (e.g., the U.S. Census or the National Health and Nutrition Examination Survey), which could help generalize estimates from the highly selected sample to a more representative group. Thus, there is a need for standardized benchmark measurements to be incorporated into new and ongoing studies to evaluate and correct for selection bias.

One novel mechanism that may enable broad reach to a diverse population and mitigate some biases introduced by current recruitment practices is the establishment of an opt-in option for newly age-eligible individuals at the time of Medicare or Medicaid enrollment to receive information on AD/ADRD studies and invitations to contribute to research. While mailings to Medicare beneficiaries are already used in clinical trial recruitment strategies (Grill and Galvin, 2014), such an initiative could better target trial information to enrollees based on areas of interest and baseline data. Enrollees who opt in could be asked for a blood sample, a baseline digital cognitive test, and to fill out a basic life history questionnaire, all of which could be included in a repository. Screening and other data could be provided back to those who opt in as an incentive. Importantly, Medicare and Medicaid already cover routine costs that accompany clinical trial participation, a step that improved equitable access to clinical trials (Takvorian et al., 2021). A collaborative effort by NIH and CMS to establish the infrastructure for such an initiative would further advance health research equity and help to scale up efforts to improve the representativeness of AD/ADRD research.

In 2018, NIA released a National Strategy for Recruitment and Participation in Alzheimer’s and Related Dementias Clinical Research⁸ that aims to meet the goal of engaging “broad segments of the public in the Alzheimer’s and related dementias research enterprise, with a particular focus on underrepresented communities, so studies with an aim to better understand and eventually cure these disorders can successfully and more quickly enroll and retain individuals” (NIA, 2024a). This strategy and its

___________________

⁸ The National Strategy, along with the associated planning guide and online toolbox, is available at https://www.nia.nih.gov/research/recruitment-strategy (accessed July 1, 2024).

related planning guide outline best practices and potential resources for study sites and researchers in overcoming barriers to engaging with and retaining diverse and underrepresented populations in clinical research (NIA, 2024a). Practices highlighted in the strategy align well with those that have been emphasized in the published literature (Brewster et al., 2019; Davis and Bekker, 2022; Mindt et al., 2023) and were discussed with the committee during its public workshop (NASEM, 2024). Examples include the following:

• Establish relationships with communities, employing culturally and linguistically appropriate modalities and content during outreach efforts, and build trust.
• Incorporate community voices into all phases of the study (e.g., through such mechanisms as community advisory boards, community-science partnership boards, and community-based participatory methods).
• Embed trials in communities (particularly those with populations at increased risk), for example by working with federally qualified health centers.
• Take trial activities to the participants by expanding opportunities for virtual engagement, use of in-home testing kits (e.g., for blood sample collection), and digital/remote data collection to increase accessibility.
• Give back to communities, and compensate participants for their time.

To support such efforts, NIA also created an online toolbox of resources—the Alzheimer’s and Dementia Outreach, Recruitment, and Engagement Resources (ADORE)—that is based on the recommendations and practices outlined in the planning guide. Resources from this searchable repository include videos specific to diverse and underrepresented communities, outreach materials that can be adapted for use in local communities, and resources from other centers and organizations that can be used to improve recruitment and retention. Knowledge regarding the engagement of diverse populations in research can continue to be evaluated and practices optimized through the use of community-based participatory research projects and other approaches.

While these NIH efforts connect researchers to resources that can support more inclusive research, costs associated with employing these best practices may remain a barrier to adoption. The committee found limited examples of supplemental funding from NIH to expand the participation of underrepresented groups (NINDS, 2021, 2024). Ensuring adequate resources are budgeted for community engagement, recruitment, and the development

of culturally appropriate research tools is critical to furthering inclusive AD/ADRD research. Additional funding, including through supplemental funding, could be provided by NIH for recruitment of diverse populations and specific clinical trial financial needs not well covered by project budgets (e.g., transportation for participants, additional medical supplies needed by participants). Strategies for using technological advances to bring these best practices to scale in a cost-effective manner also need to be considered.

Building a diverse research workforce at all levels is another critical factor that was emphasized in the national strategy and planning guide (NIA, 2018, 2019a). Effectively engaging with communities requires insights into, and sensitivity to, the different perspectives and cultures represented therein. This cannot be achieved without a diverse and multidisciplinary study team (NASEM, 2021). Recognizing that inadequate diversity of the scientific workforce is a broader issue that transcends individual studies, NIH established and recently expanded the Resource Centers for Minority Aging Research program, which aims to provide career development (e.g., training, mentorship) to early-career scientists from diverse backgrounds who conduct research related to aging, AD/ADRD, and health disparities in older adults in the areas of social, behavioral, psychological, and economic research (NIA, 2024b). Acknowledging the work NIH is already doing to foster a diverse research workforce, continued efforts are needed to understand and mitigate the barriers to entry and ongoing career advancement into leadership positions. This will include system-level research on barriers, metrics for evaluation, and the expansion of such existing support mechanisms as paid internships and mentorships, collaborative research partnerships, competitive stipends, and salary support to retain researchers in academia).

The development of a diverse research workforce requires early-career professionals to choose to enter and advance their careers in the field. Inadequate compensation for trainees and postdoctoral researchers under NIH-funded awards, however, represents a significant financial barrier to entry into the biomedical research field that may deter pursuit of academic research careers and limit diversity within the research workforce (NIH ACD Working Group, 2023; Sainburg, 2023). Ph.D. training represents a financial burden, even when participating in a fully funded program or prestigious institutional or individual NIH-funded awards (e.g., T32, F31 or F99). Financial stressors extend into postdoctoral training with the rising costs of housing, childcare, and the desire to maintain a reasonable quality of life frequently cited as major career concerns. Between 2020 and 2022, there was a nearly 10 percent decrease in the number of NIH-funded postdoctoral researchers in the health and science fields in the United States (Gewin, 2023). The current NIH structure for postdoctoral stipends results in wages that are below the cost of living in major metropolitan areas and current rules for supplementing stipends from NIH training awards to bring

compensation up to a living wage requires the use of non-federal funds, which represents an additional barrier (NIH Office of Extramural Research, 2024; Sainburg, 2023). Relatedly, compensation following training years, in comparison to compensation that could be received in private industry with similar expertise, may not entice early-career researchers to initiate or continue to pursue a career in academic research (Sainburg, 2023). These early financial pressures, in combination with other factors (e.g., uncertain career prospects), disproportionately affect marginalized groups and act as impediments at all career stages to the development of a strong and diverse biomedical research workforce. These factors need to be better understood and swiftly addressed in the context of the AD/ADRD field.

Importantly, endeavors to foster a diverse research workforce need to be focused on overcoming systemic barriers, such as ethnic and racial disparities in the awarding of federal research funding (Ginther et al., 2011; Nguyen et al., 2023) and the lack of research infrastructure and protected faculty time for research at such institutions as historically Black colleges and universities and tribal colleges and universities. Such efforts not only help to address challenges related to underrepresentation in research but ensure that the scientific workforce benefits from the nation’s rich diversity of people and their broad range of perspectives and experiences.

Efforts to increase inclusive AD/ADRD research have largely focused on recruitment practices, with less attention given to other factors such as eligibility (Franzen et al., 2022) and retention (Gilmore-Bykovskyi et al., 2019). Exclusion based on eligibility requirements can have the undesirable effect of disengaging people who otherwise would have been interested and willing to participate in research and potentially limiting generalizability to the broader target population, as might be the case when comorbidities such as active depression and diabetes are used as exclusion criteria (Mitchell et al., 2024; Ritchie et al., 2015). Understanding that some eligibility criteria may be instituted to protect the safety of research participants, particularly in the context of clinical trials, consideration should be given to ways to overcome or work around common factors that contribute to attrition at the screening stage (e.g., waiving caregiver/care partner requirements if adequate cognitive function can be demonstrated through ongoing evaluation), particularly for members of underrepresented groups. For example, recent FDA guidance on enhancing the diversity of clinical trial populations suggests considering whether criteria used in earlier-phase trials can be eliminated or modified in later-stage studies (FDA, 2020). There also needs to be consideration as to how gaps in knowledge regarding AD/ADRD in diverse populations may exacerbate underrepresentation caused by eligibility requirements. In developing inclusion and exclusion criteria, study investigators also need to be cognizant of sex and gender differences and how those might influence recruitment and retention.

In recognition of the gap between the requirements for data sharing and real-world practices, NIH is also investing in research to better understand the factors that pose barriers to or facilitate data sharing (NIH RePORTER, 2023a). Such barriers may include concerns regarding the inability to control how data will be used, potential effects on the ability of the original investigators to publish their analyses of the data, the need to protect participant privacy and confidentiality, and the substantial resources and effort required to share data (Alzheimer Europe, n.d.; Arias et al., 2024). Such barriers point to the importance of data infrastructure and adequate financial support to cover costs, as well as culture, for aligning data sharing practices with current policy.

Cultural barriers to data sharing are a broader issue within the scientific community and beyond the scope of this report. Still, it is worth emphasizing the need for incentives to drive meaningful compliance with existing policies and ensuring consistency across data sharing policies (e.g., across different NIH institutes and centers). Academic reward systems in particular need to place greater value on data sharing. For instance, reuse of datasets could be given similar status as publication citations, but this necessitates a system to ensure original data generators are credited when their data are reused (Pierce et al., 2019).

Reflecting the substantial breadth of NIH-funded AD/ADRD research initiatives, AD/ADRD data infrastructure investments by NIH have included a number of different platforms and repositories to support data storage and accessibility (see Box 5-1). For example,

• All ADRC data (e.g., fluid biomarker, neuroimaging, neuropathology, and clinical data) are stored in the National Alzheimer’s Coordinating Center (NACC). Such data, including those derived from NIH-supported longitudinal cohort studies, have been critical to fluid and imaging biomarker discovery efforts.
• Data generated by the Alzheimer’s Disease Sequencing Project (ADSP) consortia (the AI/ML Consortium, the Phenotype Harmonization Consortium, and the Functional Genomics Consortium) and projects are deposited in NIA data repositories, including the NIA Genetics of Alzheimer’s Disease Data Storage Site (NIAGADS).
• Multiomics data collected through such programs as AMP-AD and MODEL-AD are stored in the AD Knowledge Portal and are informing drug development, beginning with target identification and validation and continuing through all phases of clinical trials (Hodes and Buckholtz, 2016; NIA, 2023c).

A key challenge before NIH is to link its major data hubs (e.g., NIAGADS, NACC, Laboratory of Neuro Imaging [LONI]/ADNI, AD Knowledge Portal)

into an agile, integrated data ecosystem while preserving the autonomy of the individual platforms and their respective strengths and networks (NIA, 2023b), with an emphasis on the inclusion of repositories with data on related dementias. This integrated system should enable researchers without deep data analytics expertise to locate, access, and query existing data from NIH-funded research and, when available, data submitted by other investigators. This is a formidable undertaking but critical to maximizing insights from AD/ADRD research and returns from NIH’s various investments. Insights captured during recent NIH-hosted workshops on the topics of data infrastructure and interoperability (NIA, 2023b, 2024c) can guide these efforts.

Given the diversity of data types, data sources, and constraints such as privacy protection needs, there is no single solution to data management and accessibility. For example, although AI/ML models may facilitate the sharing and processing of digital data, commercially available platforms lack transparency and have associated privacy and security concerns. It will be important for NIH to work with investigators to identify solutions to data access challenges. While many datasets can and should be made publicly available, for others, such as clinical datasets with protected health information generated by private health systems, data accessibility may

need to be achieved through other means (e.g., federated data analyses). One solution to data that cannot be shared is to enable code generated by external investigators (generated, for example, using a synthetic dataset) to be run using a secure system that outputs the results from the executed program without allowing the investigator to see the data being analyzed. Consideration could also be given to categorizing data into access levels. At the lowest level would be those data that can be made freely available without any permission process to access them. Higher levels would involve increasingly more stringent access permission processes based on the sensitivity of the data and privacy/confidentiality concerns. For datasets that include both sensitive and nonsensitive data, such an approach may improve access to those data that are not sensitive by reducing hurdles associated with permission processes.

Accessibility is necessary but not sufficient to ensure that data from past AD/ADRD research are usable to the fullest extent possible. A key challenge has been the lack of interoperability and comparability of data from different studies, which impedes data integration and cross-study analyses. Investments in basic and preclinical research and longitudinal cohort studies have resulted in the generation of large and diverse (e.g., genetic, exposome, functional omics) datasets that can guide the development of effective interventions for AD/ADRD. A notable challenge, however, is the harmonization and integration of data to make them available and ready for analysis so that the many potential targets can be comparatively evaluated and prioritized for intervention development, with consideration of the needs of diverse populations.

One way NIH has sought to make data more comparable is through the standardization and harmonization of data collection tools and outcome measures (e.g., common data elements [CDEs]). The NINDS Common Data Elements project, for instance, had resulted in the creation of data standards for neuroscientific clinical research, including both general and disease-specific CDEs (NINDS, n.d.). Another example of such efforts is the Harmonized Cognitive Assessment Protocol (HCAP) (see Box 2-11 in Chapter 2), the design of which was led by the investigators of the Health and Retirement Study with the goal of creating a flexible instrument for detailed cognitive assessment of older adults that would be comparable across populations in diverse contexts (e.g., cultural, educational, social, economic, political) (Gross et al., 2023). In the digital data space, NIH is seeking to develop standards for wearable/sensor data and automated scripts that can convert raw data into a standardized format to facilitate integration with other patient data types (e.g., clinical, molecular, digital) (NIA, 2019b, 2024d). The status of these efforts is unclear, however, and expansion of data standardization and harmonization remains a priority. In February 2023, NIA convened an expert meeting to explore the use of

CDEs to harmonize data from real-world-data (RWD) sources, such as health care claims and electronic health records, with the goal of accelerating AD/ADRD research (Lieberthal et al., 2023). Looking to the future, new harmonization methods will be required to meet the needs of a broad range of analytic approaches (e.g., biostatistics/epidemiology, data science/AI, and other emerging methods such as quantum computing) used for both structured and unstructured data (e.g., digital images and raw digital data streams such as voice recordings).

With the growing volume and diversity of data being generated through AD/ADRD research, data integration and analysis are requiring increasingly complex tools and analytic methods. Such capabilities will be needed to advance precision medicine approaches, for example, by enabling the extraction of patterns that could guide population stratification, matching of subgroups to interventions, and life-course timing for those interventions. AI/ML and other computational methods (e.g., network analysis) hold great promise for enabling the linkage and subsequent extraction of insights from large and complex datasets to inform intervention strategies. These methods need to be biologically informed and relevant to clinical phenotypes, which will require close collaborations between AI/data scientists and biomedical researchers (Recommendation 4).

Academia is well suited to the experimentation and small-scale, proof-of-concept development that can lead to novel analytic tools but may lack access to the computational resources available in private industry (NASEM, 2024). Thus, there is a need for national-level resources that can support the development of novel, open-source data acquisition, ingestion, cleaning, processing, and harmonization tools to deliver data in a sufficient form for different analytic methods. Some initiatives that could meet this need are already underway. The National Artificial Intelligence Research Resource (NAIRR) pilot, an NSF-led, multipartner effort—including federal agencies and private-sector, nonprofit, and philanthropic partners—to develop a proof of concept for shared national AI research resources, represents a first step in in this direction. As part of the NAIRR pilot, NIH will co-lead with the DOE an operational focus area on data privacy and security and “provide open and controlled-access to NIH computing and data platforms and biomedical datasets to support health care-focused AI research (NSF, 2024).

Additionally, in 2021, NIA awarded funds for the creation of Artificial Intelligence and Technology Collaboratories (AITCs), which serve as national resources to “promote the development and implementation of artificial intelligence approaches and technology through demonstration projects to improve care and health outcomes for older Americans, including persons living with dementia and their caregivers” (NIA, 2023e). The AITCs are developing approaches for the analysis of raw digital data; creating policies

and best practices for the incorporation of AI in partnership with industry, venture capitalists, and health care systems; and piloting new tools and technologies in collaboration with these partners and the NIA Small Business Programs (NIA, 2023e). The AITCs operate through pilot awards and are intended to create an ecosystem that fosters innovation and transdisciplinary collaboration spanning academia and industry (Abadir et al., 2024; NIH RePORTER, 2023b). In the longer term, NIH will need to evaluate the success of such innovative models and their potential role in broader efforts to use AI/ML and other technologies to accelerate data integration and knowledge generation.

Biobanking

The continued evolution of technology, tools, and analytic methods will create new opportunities to analyze previously stored biological samples in ways that may not be possible to predict at present. Such future analyses may lead to the development of novel therapeutics or biomarkers. Consideration of this future value of biological samples collected from research participants or donated by other members of the public is important when investing in infrastructure and plans for collection and storage. For example, attention needs to be given to the harmonized collection and storage of pre- and postmortem samples from across cohorts so that variation in collection and storage procedures does not introduce undesired variability that compromises the comparability of data from the later analyses of those samples. Furthermore, stored samples have little value if they are not accessible to the scientific community. Ensuring accessibility entails the

development of transparent inventories of samples that are available to external investigators and clear processes for sample requests and decision making on sample sharing. As described earlier, access to biological samples from international collaborations remains a priority barrier to be addressed.

NIH has made substantial investments in infrastructure for biobanking, as described in Chapter 2. However, critical gaps remain, and initiatives and resources appear fragmented. Research institutions outside of the ADRCs may lack the pathology laboratory infrastructure to collect certain valuable samples (e.g., well-characterized brain tissue) from decedents enrolled in longitudinal studies. With adequate support, ADRCs and other large NIH-funded programs such as the Alzheimer’s Clinical Trials Consortium could expand autopsy and brain collection services to participants enrolled in studies at other institutions. Such a system may help to address the current dearth of samples from diverse participants but would need to ensure that costs for autopsy and transportation to and from the donation site do not fall on the research participants or their families. Additionally, capacity to store biosamples, also a cost-intensive undertaking, is limited. As a result, precious samples may be discarded at the completion of studies. Repositories may be sent sample collections with little advance notice and through processes that are ad hoc.

Maximizing the use and value of biological samples will require a more structured and standardized system for collection, archiving, and access. Biosample sharing systems should also require the return of results from analyses of shared samples back to the parent study so that these new data are made available along with any previously generated phenotypic and other data related to the shared sample (see Recommendation 2). This will ensure future requests for biosamples are made in the context of knowing what has already been measured and will avoid duplicate measurements on precious samples.

Technologies that have enabled the digitization and computational analysis of histopathological samples are transforming the practice of neuropathology. The portability and rapid transferability of digital image files in comparison to traditional slides enhances the accessibility of these valuable data sources (Scalco et al., 2023). The capability to use AI/ML for analysis of whole slide images opens opportunities to shift to quantitative measurement of the neuroanatomic distribution of pathologies (Shakir and Dugger, 2022). This kind of scalable quantitative method has the potential to allow deeper phenotyping of AD/ADRD and the correlation of specific clinical phenotypes to patterns of neuropathology. Incorporation of digital pathology methods into cohort studies is just getting underway, and digitization of existing slide inventories at ADRCs has been slow. ADRC personnel have reported a lack of resource support for digital pathology and ML (Scalco et al., 2023). NIH can advance AD/ADRD

disease phenotyping and the current understanding of the neuropathologic landscape by ensuring adequate support for the implementation and refinement of digital pathology methods. Importantly, infrastructure support needs to go beyond equipment (e.g., slide scanners, data storage/servers) and include the full range of expertise needed for implementation (e.g., pathologists, ML engineers, information technology personnel, statisticians, and database managers) (Shakir and Dugger, 2022). These efforts should be guided by the existing ADRC Digital Pathology Working Group (Scalco et al., 2023).

Catalyze Transformational Change Through Innovation in AD/ADRD Research

Accelerating progress in AD/ADRD prevention and treatment will require transformational change that can only be achieved through greater support for innovation in NIH-supported research. Such efforts may look different at different stages in the research continuum and would include the following:

Cancer Moonshot initiative, such a capability also has clear relevance to neurodegenerative disease research and practice (see Recommendation 9).

Of note, because ARPA-H-funded research is time delimited by design, transition of the resulting scientific and technological breakthroughs into real-world application is an integral part of the life cycle of ARPA-H projects and programs. The Project Accelerator Transition Innovation Office (PATIO) helps ARPA-H awardees identify investors and customers but also works with CMS and FDA to develop strategies for regulatory approval (e.g., for technologies that may not fit into traditional FDA pathways) and reimbursement (ARPA-H, 2024b). This approach creates the necessary demand–pull conditions to ensure innovations reach the market (Pannu et al., 2023). The absence of a similar transition model for innovative technologies (e.g., digital technologies for AD/ADRD diagnosis) arising from NIH-funded research serves as a barrier to the financial sustainability of the innovative product and future product development.

The NIH Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) funding mechanisms and the NIH Blueprint for Neuroscience Research program provide some support for converting research ideas into commercially viable products. For example, in addition to providing funding for drug discovery and development, the Blueprint Neurotherapeutics Network offers access to consultants with extensive industry drug development experience (NIH, 2024b). Such mechanisms, however, fall short of addressing the needs around nontraditional regulatory pathways and reimbursement, gaps that warrant NIH attention.

Funding mechanisms such as SBIR/STTR and NIH Blueprint function to derisk investment in innovation. The phased funding in SBIR, for example, enables awardees to establish the scientific merit, feasibility, and commercialization of the proposed research in phase 1 (approximately 0.5–2 years) prior to receipt of a larger and longer-term (1–3 years) funding award for phase 2 (NIH, 2024c). A fast-track path allows a single application to cover both phases. Some philanthropic funders also seek to derisk novel approaches by funding proof-of-concept research and pilot-scale projects to generate the initial data needed to secure longer-term funding, including follow-on awards from NIH (FNIH, 2023).

These kinds of phased funding approaches, which can involve a review of a priori-specified milestones prior to transition to phase 2 (NASEM, 2015), could be adapted to further promote innovative and high-risk AD/ADRD research. For example, partnerships between NIH and philanthropic groups could create a seamless transition from philanthropic proof-of-concept funding to NIH grant funding. Additionally, NIH and the FNIH could collaborate to build private–public partnerships for fast-tracked phase 1–2 high-risk research opportunities. In such a model, NIH could fund phase 1 with the goal of lowering the risk for the phase 2 component, which would

be focused on scalability and could largely be funded through industry. Public–private partnerships enable the leveraging of the respective talents of industry and academia and best-in-class practices (e.g., data infrastructure and knowledge management, social engagement) from other sectors.

Other funding mechanisms to support innovation could include supplements focused on critical scientific areas or technologies and R56 awards (or similar) focused on high-risk, high-reward research. The R56 mechanism would allow discrete pieces of an R01 application that include novel ideas to be funded in cases where the full application did not meet criteria for an R01 award.