Target Audience
- Researchers, healthcare professionals, and research administrators interested in developing and operating large-scale biobanks and longitudinal cohort studies.
- Professionals and policymakers seeking to understand developments in genomic medicine, precision medicine, and healthcare big data.
- Readers interested in the differences between UK Biobank and the All of Us Research Program, as well as the lessons that can be learned from building research infrastructure at a national scale.
Summary
- The All of Us Research Program is a large-scale precision medicine research initiative led by the U.S. National Institutes of Health (NIH). Originating from the Precision Medicine Initiative announced in 2015, it aims to enroll more than one million participants.
- Unlike UK Biobank, All of Us places particular emphasis on participant diversity, collaboration with participants, returning individual genetic results, and providing access to data through a cloud-based research environment.
- By 2026, more than 888,000 individuals had enrolled, and a large-scale dataset containing whole-genome sequences from more than 535,000 participants had become available for research.
- Major challenges include declining long-term funding, population representativeness, missing electronic health records, longitudinal follow-up, and privacy protection.
- The history of All of Us offers important lessons not only about building large-scale research infrastructure but also about ensuring its long-term sustainability and translating research findings into healthcare benefits.
1. Introduction: What Is the All of Us Research Program?
The All of Us Research Program (hereafter referred to as All of Us) is a long-term research initiative led by the U.S. National Institutes of Health (NIH).
It originated from the Precision Medicine Initiative (PMI), announced by the Obama administration in 2015, and began nationwide participant enrollment in May 2018.[1][2]
Its goal is to continuously collect genomic, medical, lifestyle, environmental, and other health-related information from more than one million people in the United States to accelerate research into personalized prevention, diagnosis, and treatment.
Although All of Us shares many characteristics with the United Kingdom's UK Biobank, it has several distinctive features. These include actively recruiting populations historically underrepresented in biomedical research, establishing two-way relationships with participants, returning individual genomic results, and enabling data analysis through a cloud-based research environment.
1.1 Progress as of 2026
As of the research reference date of November 4, 2026, key figures published by the NIH were as follows.
| Indicator | Number | Reference period and notes |
|---|---|---|
| Total enrolled participants | More than 888,000 | NIH figures published in 2026 |
| Participants with data available for research | More than 747,000 | June 2026 data release |
| Whole-genome sequencing data | More than 535,000 | June 2026 data release |
| Research community | More than 25,000 | Number of researchers reported by the NIH |
Source: NIH data release dated June 30, 2026, and NIH funding information.[3][4]
These figures represent different populations: total enrolled participants, participants with research-ready data, and participants whose genomes have been sequenced. They should not be treated as interchangeable.
All of Us can currently be understood as being at a turning point, shifting its emphasis from constructing a large-scale data infrastructure to sustaining its operations and translating research findings into practical benefits for society.
Five issues are particularly important:
- Securing long-term funding: How can the research infrastructure be maintained and expanded as major sources of initial funding come to an end?
- Data representativeness: How should the program address the fact that, despite its emphasis on diversity, its participants do not necessarily form a statistically representative sample of the entire U.S. population?
- Data completeness and follow-up: How can missing electronic health records, fragmentation across healthcare institutions, and gaps in continuous data collection be addressed?
- Returning value to participants: How can the program establish mechanisms for returning genetic information and translating it into actual health improvements?
- Privacy and trust: How can the protection of highly sensitive genomic information be reconciled with expanded research access?
Among these issues, funding is particularly pressing.
The NIH has projected that the program's budget in fiscal year 2027 could be approximately 80% lower than its fiscal year 2023 level. It has also stated that funding reductions are already affecting new participant enrollment, data collection, and the expansion of the pediatric cohort.[4]
2. Why Was All of Us Conceived?
2.1 Background: Advances in Genomic Medicine and the Limitations of Traditional Research Cohorts
Understanding the origins of All of Us requires examining the advances in human genomics and large-scale epidemiological research that began in the 2000s.
The Human Genome Project was completed in 2003. Subsequently, declining genome sequencing costs, the increasing adoption of electronic health records, and advances in cloud computing made it increasingly feasible to integrate large volumes of genetic and health information.
However, traditional genomic research faced two major limitations.
- Research participants were disproportionately drawn from populations with European ancestry. As a result, genetic findings obtained from these studies could not necessarily be generalized to other populations.
- Genetic information alone was insufficient to explain disease. Researchers increasingly recognized the need to integrate lifestyle factors, socioeconomic conditions, environmental exposures, medical histories, and other information.
In 2015, an NIH working group developed a plan proposing a large-scale longitudinal cohort covering diverse populations across the United States to address these limitations.[1][6]
2.2 The Policy Origins: The Precision Medicine Initiative (2015)
On January 20, 2015, then-President Barack Obama announced the Precision Medicine Initiative during his State of the Union address.
The initiative aimed to advance healthcare tailored to patients' genetic characteristics, environments, and lifestyles.
Its initial priorities included genomic medicine in cancer research and the establishment of a research cohort involving approximately one million people across a broad range of diseases.
The initial fiscal year 2016 budget allocated $130 million to the NIH for building the large-scale cohort and $70 million to the National Cancer Institute (NCI) for cancer genomics research.
Thus, All of Us did not emerge simply as a research cohort spontaneously established by individual investigators. Instead, it was conceived as a national research infrastructure project launched as a strategic policy priority of the U.S. federal government.[1][8]
2.3 The Principles Behind the Initial Concept
From the beginning, the program was designed to treat participants not merely as providers of biological samples but as partners in research.
This approach aimed to enable participants not only to contribute information but also to express opinions about research governance, receive research findings, and obtain their own genetic results.
The program was also designed as a broad research infrastructure capable of supporting investigations into many different diseases rather than focusing on a specific medical condition.
In this sense, All of Us was conceived not only as a precision medicine research cohort but also as an institutional framework that promotes equity in research participation and treats research data as a shared resource for society.[1][6]
3. Timeline: From Conception to the Present
The following timeline summarizes the major developments in All of Us from its conception through 2026.
| Year | Development stage | Major events |
|---|---|---|
| 2015 | Policy conception and initial design | The Precision Medicine Initiative was announced in January. In September, an NIH working group recommended the basic design of a cohort involving one million participants. |
| 2016 | Development of organizational and research networks | The NIH began funding research partners. Preparations for participant recruitment, biobanking, and data infrastructure began. The program was renamed All of Us in October, and the 21st Century Cures Act was enacted in December. |
| 2017 | Pilot operations | Beta enrollment began on May 31. Procedures for informed consent, sample collection, and data processing were tested. |
| 2018 | Nationwide launch | The program officially launched nationwide on May 6. Approximately 28,000 core participants had already been enrolled during the beta phase. |
| 2019–2020 | Release of research data infrastructure | A publicly accessible Data Browser was launched in 2019. Pilot operation of the Researcher Workbench began in 2020. |
| 2021–2022 | Research use of large-scale genomic data | Whole-genome sequencing and genotyping activities expanded. By 2022, genomic data from approximately 98,000 participants had become available for research, and the return of health-related genetic results also progressed. |
| 2023–2024 | Expansion of genomics research and the cohort | Whole-genome sequencing data from approximately 245,000 participants were released. Other developments included pilot pediatric participation, expanded digital health information, and publication of major findings in Nature. |
| 2025 | Expansion of data scale and types | Genomic data from approximately 400,000 individuals became available for research, and the researcher community expanded. Additional safeguards and review mechanisms for research involving Indigenous participants' data were also established. |
| 2026 | Large-scale data release and operational transition | A major data release in June included whole-genome sequences from more than 535,000 participants. Responding to reduced funding became a major challenge. The CEO stepped down at the end of October, and the program transitioned to interim leadership. |
Sources: NIH official publications, the 2019 New England Journal of Medicine article, the 2024 Nature article, and NIH announcements in 2026.[1][2][3]
3.1 Changes in Leadership in 2026
On September 16, 2026, the NIH announced that Dr. Josh Denny, CEO of All of Us, would step down effective October 31.
Dr. Denny was a central figure in the program and had also contributed to the development of its initial plan in 2015.
Until a permanent successor is appointed, Dr. Stephanie Devaney, the program's Chief Operating Officer, is serving as interim CEO.
The NIH has stated that the program's commitments to participants and its core research infrastructure operations will continue.
Therefore, the latter half of 2026 represents a transition not only in financial terms but also from the leadership structure that supported the program's original vision to its next phase.[2]
4. Key Features of All of Us
4.1 Diversity as a Central Principle of Research Design
The defining characteristic of All of Us is not simply the size of its participant population.
Since its initial conception, the program has consistently emphasized the active inclusion of populations that have historically been underrepresented in biomedical research (UBR).
These populations include racial and ethnic minorities, rural residents, older adults, people with disabilities, and individuals facing socioeconomic disadvantages.
According to the June 2026 data release, 86% of participants represented in the data were classified as belonging to populations historically underrepresented in biomedical research, based on the NIH's broad definition of UBR.
Importantly, this 86% figure does not refer exclusively to racial or ethnic minorities.[3]
4.2 Integration of Multiple Data Types From the Same Participants
All of Us collects several types of information from its participants.[3][6]
| Data category | Main contents |
|---|---|
| Genomic information | Whole-genome sequences, genotypes, and structural variants |
| Clinical information | Diagnoses, medications, laboratory tests, and electronic health records |
| Surveys | Lifestyle, health status, and social background |
| Physical measurements and wearables | Blood pressure, body measurements, physical activity, sleep, and related data |
| Biospecimens and omics | Blood, urine, saliva, proteomics, and other data |
| Environmental and social factors | Residential environments, socioeconomic conditions, and environmental exposures |
The major June 2026 data release newly included proteomics data from approximately 10,000 participants, RNA sequencing data from approximately 9,000 participants, and long-read whole-genome sequencing data from more than 14,500 participants.[3]
An important consideration is that these data types are not available for every participant.
Depending on the disease or research question being investigated, researchers must identify participants for whom the required combinations of data are available.
4.3 Treating Participants as Research Partners
All of Us has established several mechanisms for building and maintaining relationships with participants.[1][6]
- Participation procedures based on electronic informed consent
- Continuous engagement through community organizations and healthcare institutions
- Mechanisms for incorporating participant perspectives into research governance, ethics, and data use
- Return of information about genetic ancestry, traits, and health-related genetic findings to participants
- Opportunities to participate in additional research activities
According to an NIH announcement in September 2026, health-related individual genetic results had been returned to more than 277,000 participants.
Among these participants, more than 5,620 had received information concerning medically actionable genes, while more than 158,000 had received information related to drug response.
This is an example of a system designed not only to accumulate research data but also to provide participants with information of potential personal value.
However, returning genetic test results must be distinguished from demonstrating that those results have actually improved health outcomes.
4.4 Providing Access to an Analysis Environment Rather Than Distributing Raw Data
Rather than allowing researchers to freely download sensitive individual-level data, All of Us employs a managed cloud-based environment called the Researcher Workbench.[6]
Access to research data is organized into the following tiers.
| Access tier | Available content |
|---|---|
| Public | Publicly available aggregated information |
| Registered | Lower-risk data made available to approved researchers |
| Controlled | More sensitive data, including genomic information |
The integrated data infrastructure contains genomic information, electronic health records (EHRs), survey responses, physical measurements, wearable device data, and other information.
Researchers use these data within the Researcher Workbench according to their access permissions.
The Researcher Workbench supports activities such as:
- Identifying and constructing research cohorts
- Statistical analysis
- Genomic analysis
- Collaboration among researchers
Researchers must complete identity verification, training, and agreement to the program's terms of use before obtaining access. Institutional agreements are also required.
Although registration itself is free, researchers incur costs for cloud computing resources and data storage.
This is an important practical consideration for international collaborative research and smaller research institutions.
5. Comparison With UK Biobank
All of Us and UK Biobank share the goal of combining genomic information with long-term health data.
However, they differ in participant recruitment, research timelines, and relationships with participants.[1][6][7]
| Comparison item | All of Us | UK Biobank |
|---|---|---|
| Main launch period | 2018 | Recruitment conducted from 2006 to 2010 |
| Participant population | Target of more than one million; currently more than 888,000 | Approximately 500,000 |
| Eligible participants | Diverse U.S. residents, initially focusing on adults and subsequently expanding to children | Participants aged 40–69 at initial enrollment |
| Recruitment status | Participant recruitment is ongoing | Initial recruitment ended in 2010 |
| Recruitment methods | Healthcare institutions, community organizations, online recruitment, and other channels | Assessment centers across the United Kingdom |
| Main emphasis | Diversity, precision medicine, and participant collaboration | Long-term follow-up and disease research using multidimensional data |
| Main data sources | Genomics, EHRs, surveys, physical measurements, wearables, and other sources | Genomics, medical records, imaging, surveys, biospecimens, and other sources |
| Data access | Managed cloud-based research environment | UK Biobank Research Analysis Platform and other mechanisms |
| Return of information to participants | Active return of individual genetic results | Primarily contributions to long-term research and provision of research information |
5.1 The Most Important Differences
UK Biobank is already a mature longitudinal cohort.
Initial participant recruitment ended in 2010, allowing researchers to analyze changes in health status over more than 15 subsequent years.
It is particularly well suited to studies requiring long-term follow-up of new disease onset, mortality, changes in imaging findings, and other outcomes.
In contrast, All of Us is a newer cohort that continues to expand.
It actively collects data from diverse populations and enables research that combines EHRs, genomic information, and wearable device data.
Although its follow-up period since recruitment is shorter than that of UK Biobank, its EHR data may include clinical information recorded before participants enrolled in the program.
Neither cohort represents a random sample of the entire population.
Therefore, a large number of participants does not, by itself, guarantee demographic representativeness.[5][7]
6. Achievements to Date
6.1 Establishing a Globally Important Genomic Research Infrastructure
In February 2024, a paper titled "Genomic data in the All of Us Research Program" was published in Nature, reporting the release and characteristics of whole-genome sequences from 245,388 participants.[2]
This was particularly significant because it enabled large-scale analysis of genetic variation in ancestry groups that had previously been insufficiently represented in genomic research.
By June 2026, the number of available whole-genome sequences had expanded to more than 535,000.[3]
The growth in whole-genome sequencing data available for research can be summarized as follows.
| Period | Whole-genome sequences available for research |
|---|---|
| 2022 | Approximately 98,000 |
| 2023–2024 | Approximately 245,000 |
| June 2026 | More than 535,000 |
Sources: NIH program publications, the 2024 Nature paper, and the NIH announcement in June 2026.[2][3]
These figures represent the number of genomes available for research, not the total number of enrolled participants.
6.2 Use in Scientific Research
According to the NIH, data from All of Us had been used in more than 1,400 peer-reviewed publications as of June 2026.[3]
Examples of research topics include:
- Prediction of genetic risk for cardiovascular disease
- Risk models for prostate cancer
- Research on genetic variants and existing medications related to Alzheimer's disease
However, the number of published papers is an indicator of research activity and utilization of the infrastructure.
It does not directly demonstrate clinical utility or reductions in healthcare expenditures.
6.3 Research Using More Diverse Data
The 2024 Nature paper identified the substantial inclusion of populations historically underrepresented in genomic research as a major achievement.[2]
In addition, in 2025, data from more than 30,000 participants identifying as American Indian or Alaska Native became available for research under additional protections developed through consultation with Tribal communities.
These developments demonstrate that All of Us has focused not only on increasing the scale of its dataset but also on adjusting the ways data can be used to respect the rights and maintain the trust of participating communities.
7. Current Challenges
7.1 Challenge 1: A Substantial Reduction in Funding
All of Us has been supported by the NIH's regular annual budget as well as special funding authorized under the 21st Century Cures Act of 2016.
However, with the latter funding scheduled to end in fiscal year 2026, the NIH has projected that the program's regular budget in fiscal year 2027 will be approximately $122 million, representing a reduction of about 80% compared with its fiscal year 2023 level.[4]
In fiscal year 2026, approximately $106 million in additional support was also provided through co-investment by the NIH and external partners. Nevertheless, this does not fully restore the previous level of funding.
The NIH has identified several specific consequences of the funding reductions, including reduced participant recruitment, reduced data collection, and delays in expanding the pediatric cohort.
Using the fiscal year 2023 budget level as a baseline of 100%, the fiscal year 2027 outlook can be represented as follows.
| Comparison period | Relative budget level |
|---|---|
| Fiscal year 2023 (baseline) | 100% |
| Fiscal year 2027 (NIH projection) | Approximately 20% |
| Reduction | Approximately 80% |
These are approximate figures based on the NIH's published budget outlook, not finalized budget amounts.
Assessment: This is the most immediate operational challenge facing the program.
The value of All of Us does not arise solely from genome sequencing conducted in a single year. It also depends on documenting changes in participants' health over several decades.
Consequently, the loss of long-term funding threatens not only the volume of data collected but also the fundamental value of the research infrastructure itself.
7.2 Challenge 2: Participant Representativeness and Selection Bias
Although the diversity of All of Us participants is a major strength, participants are not selected through random sampling of the general population.
A study published in the Journal of the American Medical Informatics Association (JAMIA) in 2024 compared approximately 287,000 All of Us participants with EHR data and approximately 502,000 UK Biobank participants.[5]
The study found that, among 314 diseases for which comparisons with the general U.S. population were possible, 254 diseases (80.9%) had higher recorded prevalence in All of Us.
In comparisons with UK Biobank, 2,152 of 2,515 diseases and phenotypes (85.6%) had higher prevalence in All of Us.
At the same time, the study also identified a degree of consistency in association analyses across the two cohorts for multiple diseases.
These findings have an important implication.
The inclusion of diverse participants in All of Us does not mean that disease prevalence in the cohort can be used directly to estimate prevalence in the entire U.S. population.
Studies of specific diseases, genomic associations, and differences between population groups must account for biases associated with recruitment methods, health status, and access to healthcare.
7.3 Challenge 3: Missing EHR Data and Fragmentation Across Healthcare Institutions
The U.S. healthcare system is divided among numerous healthcare providers and insurers.
When an individual receives care at multiple institutions, their medical records may not be fully integrated.
This can lead to several problems:
- Diagnostic information and laboratory results may be missing.
- Follow-up may become discontinuous when patients change healthcare providers.
- Existing diseases may not be recorded in the available EHR data.
- Differences in healthcare utilization may influence analytical results.
In the June 2026 data release, whole-genome sequencing data were available for more than 535,000 participants, while EHR data were available for approximately 482,000 participants.[3]
This illustrates that the number of participants with available data differs across data categories.
In 2026, the NIH explained that it had expanded EHR data collection through participant-mediated record sharing and health information exchange networks.
These efforts also highlight the importance of improving data linkage.
7.4 Challenge 4: Long-Term Participant Retention and Continuous Measurement
All of Us is designed for longitudinal follow-up rather than a single assessment at the time of enrollment.
However, continued data provision may become difficult as participants age, relocate, change their level of interest in research, or vary in their use of digital devices.
Furthermore, wearable device data are not provided uniformly by all participants.
The 2026 research dataset included survey responses from more than 747,000 participants, but Fitbit records were available for only approximately 68,000 participants.[3]
The research challenge is therefore not limited to preventing participant dropout.
It is also essential to understand which participants continue providing data, when those data are collected, and how consistently and accurately they are measured.
7.5 Challenge 5: Privacy Protection and the Ethics of Genomic Data
Whole-genome sequences are particularly sensitive compared with many other categories of health information.
All of Us has implemented measures such as removing directly identifying information, establishing tiered data access, and controlling research use within a cloud environment.
Nevertheless, datasets combining genomic information, medical histories, and social background information still carry risks of re-identification and potential harm to population groups.
Research involving race, ethnicity, geographic communities, and Indigenous populations must consider not only scientific usefulness but also the potential benefits and harms to the groups concerned.
Additional protections introduced for Indigenous data in 2025 included prohibitions against re-identifying individuals or Tribal communities, researcher training, and review mechanisms involving Indigenous representatives.[6]
7.6 Challenge 6: Translating Research Findings Into Healthcare
All of Us provides extensive research data and is increasingly returning genetic test results to participants.
However, the following four stages are distinct:
- Discovering associations through research.
- Developing clinically valid prediction models.
- Implementing those models in healthcare settings.
- Demonstrating actual improvements in patient health.
In particular, explaining genetic risk, providing genetic counseling after results are returned, coordinating with healthcare providers, and externally validating prediction models all require funding and specialized personnel.
Going forward, it will be important to assess not only the number of genomes analyzed or papers published but also clinical utility, healthcare access, and effects on health disparities.
8. Future Directions
Based on the NIH's published policies and the current state of the research infrastructure, several important future directions can be identified.[4][6]
| Area | Future direction | Main constraints |
|---|---|---|
| Cohort | Achieving the one-million-participant target and strengthening longitudinal follow-up | Recruitment and retention costs |
| Data collection | Expanding beyond genomics to multi-omics and environmental information | Measurement costs and missing data |
| Research | Analyses of disease risk, prevention, and drug response | Bias and external validation |
| Clinical application | Returning genetic information and supporting its clinical use | Clinical utility and cost-effectiveness |
| Data infrastructure | Expanding cloud-based environments and collaborative research | Privacy and computing costs |
| Program operations | Co-investment by the NIH and external organizations, and expanded research partnerships | Uncertainty surrounding long-term funding |
The NIH's future strategy includes enrolling one million participants, continuously collecting data from at least 500,000 individuals, and expanding collaborative research.
However, these represent goals and strategic directions.
Given the financial outlook for fiscal year 2027 and beyond, it cannot be assumed that all of these plans will proceed exactly as originally envisioned.
9. Overall Assessment: Lessons From All of Us
Looking back at the approximately 11-year history of All of Us, its development can be divided into three broad phases.
9.1 Phase 1: 2015–2018
Transforming a national policy vision into an operational research infrastructure.
This phase focused on establishing strategic policy objectives, allocating NIH funding, developing research networks, and building public trust.
Beginning with the national vision of the Precision Medicine Initiative, these efforts ultimately led to the nationwide launch of participant recruitment in 2018.
9.2 Phase 2: 2019–2025
Expanding participation, data resources, and research utilization.
Genomic sequencing, EHR integration, cloud-based research infrastructure, and the return of individual results were implemented, increasing the value of the program as a research platform.
During this period, progress extended beyond expanding the quantity of data.
The program also advanced participation by diverse populations and mechanisms for returning information to research participants.
9.3 Phase 3: 2026 and Beyond
Addressing long-term sustainability and demonstrating value for healthcare.
Rebuilding the financial foundation, maintaining longitudinal data collection, translating research into practice, and sustaining trust with participants are becoming increasingly important.
Having successfully established a large-scale data infrastructure, the program now faces the challenge of maintaining that infrastructure over the long term and ensuring that it continues generating value for society.
9.4 Lessons From the Comparison With UK Biobank
One particularly interesting insight from comparing All of Us and UK Biobank is that the value of a cohort is not determined by participant numbers alone.
UK Biobank has accumulated extensive measurement data and long-term follow-up information.
In contrast, All of Us has developed an approach emphasizing the participation of diverse populations, integration with EHRs, return of individual genetic results, and participant involvement in research governance.
Four important lessons can be drawn from these programs when designing large-scale research infrastructure.
-
Initial construction costs and long-term operating costs must be considered separately.
Even when a project successfully collects a large volume of data, its long-term value may not be fully realized unless those data can be maintained and updated.
-
Data quality, missingness, and follow-up duration are as important as dataset size.
In studies examining disease onset or prognosis, the consistency of longitudinal data is particularly important.
-
Relationships with participants should be designed as an integral part of the research infrastructure.
Informed consent, the return of information, equitable research participation, and trust within communities are essential.
-
The pathway from research utilization to clinical implementation should be clearly defined.
In addition to generating academic publications, research infrastructure should support the evaluation of improvements in prevention, diagnosis, and treatment.
10. Conclusion
All of Us is best understood not simply as a project to collect genomic data from one million people, but as a national platform for integrating diverse populations' health information over time and connecting biomedical research with personalized medicine.
As of 2026, the program has made substantial progress in establishing a large-scale data infrastructure.
At the same time, it has entered a phase in which its long-term operating model must be reconsidered.
Financial sustainability, participant representativeness, and the completeness of EHR data are particularly important challenges for understanding the program's current situation.
These issues are supported by NIH publications and peer-reviewed research.
However, caution remains necessary when drawing definitive conclusions about the program's long-term cost-effectiveness or its contribution to actual improvements in health outcomes.
The experience of All of Us offers valuable lessons for future efforts to design large-scale biobanks and precision medicine cohorts.
Ultimately, the long-term value of research infrastructure depends not merely on collecting large amounts of data, but also on building trust with participants, consistently collecting high-quality data over time, and translating research findings into meaningful benefits for healthcare and society.
References
[1] The All of Us Research Program Investigators, "The 'All of Us' Research Program," The New England Journal of Medicine, vol. 381, pp. 668–676, 2019. https://doi.org/10.1056/NEJMsr1809937
[2] All of Us Research Program, "Genomic data in the All of Us Research Program," Nature, 2024. https://www.nature.com/articles/s41586-023-06957-x
[3] National Institutes of Health, "NIH's All of Us Research Program now largest integrated genomics-health database in the world," NIH News Release, 2026. https://www.nih.gov/news-events/news-releases/nihs-all-us-research-program-now-largest-integrated-genomics-health-database-world
[4] National Institutes of Health, "Funding and Research Partnerships," NIH. https://www.nih.gov/allofus/funding
[5] Journal of the American Medical Informatics Association, "Comparative Study of All of Us and UK Biobank," 2024. https://pubmed.ncbi.nlm.nih.gov/38263490/
[6] National Institutes of Health, "All of Us Research Program Protocol," NIH. https://www.nih.gov/allofus/about/protocol
[7] UK Biobank, "Our history," UK Biobank. https://www.ukbiobank.ac.uk/about-us/our-history/
[8] Japan Agency for Medical Research and Development (AMED), "All of Us," AMED. https://www.amed.go.jp/aboutus/collaboration/ga4gh_dp001.html