The All of Us milestone shows precision medicine’s next infrastructure challenge
By Lisa Conroy, MPH
The NIH’s All of Us Research Program has become the world’s largest integrated genomic and electronic health record database. Its June 30 data release includes information from more than 747,000 participants, including over 535,000 whole genome sequences and nearly 482,000 linked electronic health records.The scale is significant. But for clinical trial sponsors, the more important lesson is what the release reveals about infrastructure.
Precision medicine is becoming less limited by the ability to generate data at scale. The bigger challenge is whether that data can be connected, standardized, and made usable across fragmented health systems.
The latest All of Us release shows both sides of that shift: the promise of large-scale genomic and clinical datasets, and the persistent difficulty of turning participant consent into usable, linked health data.
What the data release contains
The latest release, CDRv9, includes over 1.3 billion genetic variants, more than 553,000 genotyping arrays, 600,000 physical measurements, and survey responses from 747,000 participants. EHR data grew by 22%, and the dataset now includes the program’s first multiomics data, including proteomics, RNA sequencing, and long-read whole genome sequencing from participant subsets.
The program is also notable for its diversity. More than 645,000 participants, or 86% of the total, come from communities historically underrepresented in biomedical research. The dataset spans all 50 states, covers more than 98% of U.S. three-digit ZIP codes, and has already supported more than 1,400 peer-reviewed publications.
NIH Director Jay Bhattacharya framed the release around a central paradox of precision medicine: tailoring treatments to individuals requires very large populations to uncover the patterns connecting genetics, lifestyle, environment, and health outcomes.
Consent does not equal connected data
The most revealing detail in the release may not be the data that grew, but the data still missing.
According to STAT News, 98% of All of Us participants have consented to share their electronic health records. Yet more than 300,000 participants reportedly still have no EHR data available in the database. The program has responded by tapping into health information exchange networks, clinical data-sharing infrastructure originally built for care coordination, to retrieve records that participants already agreed to share.
This is not a criticism of All of Us. It is a clear example of a problem that extends across precision medicine. The bottleneck in precision medicine is increasingly not sequencing. Sequencing costs have dropped to the point where generating genomic data at scale is operationally feasible. The harder problem is connecting that genomic data to the clinical records that give it meaning: diagnoses, treatments, lab results, outcomes.
Consent is necessary, but alone it does not move data from fragmented hospital systems, disparate EHR platforms, and siloed health networks into a unified research dataset. That requires integration infrastructure.
Why this matters for clinical trial sponsors
For sponsors running genetically stratified trials, the All of Us experience reflects a challenge many teams encounter at study level. Sponsors need genomic profiles connected to medical histories to understand how many eligible patients exist, where they are, what clinical features they have, and whether a protocol reflects the real-world population. When that connection breaks, the consequences show up downstream: overestimated feasibility, slower enrollment, avoidable screen failures, and protocols built around incomplete population data.
The data integration challenge is structural. Patient registries often suffer from fragmentation, with data scattered across institutions, databases, and formats. Even when participants are willing to share records, the operational work of retrieving, standardizing, and linking those records to genomic profiles requires dedicated infrastructure.
This is the infrastructure Sano helps sponsors build. Through HIPAA-compliant medical record retrieval and data structuring, Sano connects genomic profiles with clinical history, including physician notes, diagnoses, lab results, imaging, and longitudinal medical records. That linked view helps sponsors identify eligible participants for genetically defined studies and assess feasibility using a more complete picture of each patient’s clinical context. The point is simple: turning consent into usable, linked research data is an infrastructure problem. It requires systems that can retrieve, standardize, and connect information across fragmented data sources.
What to watch next
Three developments from this release are worth watching.
First, multiomics. Proteomics, RNA sequencing, and long-read whole genome sequencing are currently available for participant subsets, but these layers will grow. As they do, the integration challenge becomes more complex. Sponsors will need to connect more data types to the same longitudinal patient record.
Second, AI-driven drug discovery and trial design. Large datasets are only as useful as the quality and completeness of the records they contain. For AI models, missing or poorly standardized clinical context can limit the reliability of predictions.
Third, sponsor-owned patient data. All of Us has substantial federal investment, broad participant consent, and national-scale infrastructure. If a program of that scale still faces EHR linkage gaps, individual sponsors need to plan for integration early rather than treating it as an operational detail later.
Looking forward
The All of Us milestone is significant for what it achieves and for what it reveals. At more than 747,000 participants, the program gives researchers an unprecedented resource for studying the connections between genetics, environment, and health. Its EHR gap shows that precision medicine’s most pressing infrastructure challenge is shifting from generating data to connecting it.
For clinical trial sponsors, the lesson is clear. The same integration work All of Us is scaling nationally is required at study level: linking genomic profiles to medical records, standardizing fragmented data sources, and maintaining those connections over time.
Programs that build this infrastructure early will be better positioned to identify eligible patients, design stronger protocols, and turn participant consent into usable longitudinal evidence.
To learn how Sano helps sponsors connect genomic and clinical data for genetically defined studies, get in touch below.