The NIH’s All of Us Research Program is a landmark effort that has built one of the largest and most comprehensive health databases in history. With more than 870,000 participants from all 50 states across the United States, All of Us brings together electronic health records, surveys, biospecimens, wearable data, genomic information, and physical measurements. Designed as an open, secure resource for registered researchers, All of Us reflects real-world variability that has long been missing from biomedical research, including of the fullest possible range of factors like ancestry, geography, age, and lived experience. Because the data are broadly available and continually expanding, researchers can revisit the dataset as new methods, technologies, and hypotheses emerge, enabling discoveries that were not possible when the data were first collected.
This breadth makes All of Us particularly valuable for understanding substance use disorders and addiction, which are shaped by a complex interplay of biology, behavior, environment, and social context. By integrating genomic data with longitudinal health records, self-reported substance use, mental health measures, and aspects of lifestyle and social environment, the program enables researchers to examine addiction across the full life course. This breadth makes it possible to identify risk and resilience factors, patterns of comorbidity, and trajectories of recovery that would be difficult to detect in smaller or more narrowly defined studies. As substance use patterns, policies, and treatments evolve, the openness and scale of All of Us allows new questions to be asked—supporting insights that can inform diagnosis, prevention, personalized treatment, and effective public health responses to addiction. All of Us also helps us advance our understanding the overlap in the physiological processes underlying different diseases necessary for whole person medicine.
Last June, NIDA launched the “Unlocking Insights: Analyzing All of Us Data for Drug Addiction Research” Challenge, a data-science competition to spur innovative use of the All of Us Research Program’s vast dataset to help understand substance use, misuse, and addiction. The challenge invited registered researchers within multidisciplinary teams—from universities, companies, nonprofits, and beyond—to leverage the rich All of Us Researcher Workbench to uncover novel patterns and relationships relevant to drug addiction. Participants were tasked with identifying and characterizing correlations across genetic, environmental, developmental, and behavioral factors that may contribute to addiction risk, resilience, treatment outcomes, or other key aspects of substance use disorders. By catalyzing new analytic approaches and discoveries using this open and diverse data resource, the Challenge aimed to accelerate insights that could ultimately inform prevention, intervention, and policy.
Judges from all NIDA divisions evaluated the 62 entries, and from a list of nine finalists, six winners were selected and announced January 30 (see below).
Together, the winning projects show the power of thoughtfully incentivizing researchers to use large, secure data resources to tackle complex public health challenges. By pairing strong privacy protections with clear rules for responsible data access, the NIH Unlocking Insights Challenge demonstrated that it is possible to generate meaningful, real-world discoveries while safeguarding participant confidentiality. For example, these teams may go on to translate their project findings into tools for clinicians. Challenges like this not only accelerate innovation and collaboration but also maximize the public value of data that participants have generously contributed, helping ensure that scientific insights translate into better, more equitable health outcomes. More than 21,000 researchers are already using All of Us data as registered researchers in the Researcher Workbench. We encourage investigators to use this resource to make new scientific discoveries by registering for access at www.researchallofus.org.
Winning Challenge Projects
Predicting Imminent Opioid Overdose among People with Opioid Dependence Using Multi-Modal Data
Mohammad Rifat Haider, Samantha Clinton, Golam Sarwar, Md Hafizur Rahman, Smita Rawal
Using electronic health record data, this team of researchers from the University of Georgia tested whether it’s possible to predict who is at risk of an overdose during the first 30 days after diagnosis with an opioid-related condition. They compared traditional statistical models with newer machine-learning approaches and found that both could identify short-term overdose risk with meaningful accuracy using data already available in health records. Notably, the study also uncovered that lower body mass index (BMI) was linked to higher overdose risk—an unexpected finding that points to new questions for future research. It demonstrated how the All of Us dataset can support practical, near-term risk-prediction tools that help clinicians identify people who may benefit from early, targeted interventions.
Analysis of Single and Polydrug Profiles in the All of Us Cohort: Self-Reported Health, Wearable Sleep Metrics, and Family History
Akane Sano, Nidal Moukaddam
This pair of researchers at Rice University and Baylor College of Medicine used a rich mix of All of Us data including surveys, electronic health records, and Fitbit data to explore how different patterns of drug use relate to everyday health and behavior. They found that people who used multiple substances or stimulants tended to report poorer overall health, had more disrupted sleep, and were less physically active, while people who only used cannabis looked similar to non-users on many of these measures. The study also found higher rates of family history of substance use among people who used multiple drugs, pointing to possible intergenerational risk. By combining clinical records with real-world data from wearable devices, this project showed how continuous, passive health measures like sleep and activity can reveal substance-related health differences, pointing the way to more personalized digital prevention health approaches. Future research should clarify whether cannabis use was for medicinal or recreational purpose
Association of GLP-1 Receptor Agonists with Acute SUD-Related Events
Sugy Choi, Jae Man Park, Xiao Li
GLP-1 receptor agonist medications, which are commonly used to treat type 2 diabetes and support weight loss, have shown promise for substance use disorders. Using All of Us data, this team of researchers at New York University, UTHealth Houston School of Public Health, and the University of Nebraska at Kearney tracked substance-related emergency department visits, hospitalizations, and overdoses in people who began GLP-1 medications and those taking other diabetes or weight-loss drugs over a period of six months. They consistently found lower rates of substance-related emergency visits and hospital stays among people who started GLP-1 medications. While more research is needed to confirm these findings, the study suggests that these widely used medications could potentially offer added benefits for people managing both metabolic conditions and substance use disorders, opening the door to new clinical and policy discussions.
GWAS-driven Prioritization of Mouse Strains for Improved Preclinical Modeling of Substance Use Disorders
Alyssa Klein, Robyn L. Ball, Ashley Auth, Elissa Chesler
This team of researchers at the Jackson Laboratory in Bar Harbor, Maine used genetic results from the All of Us program to help choose mouse strains that better reflect risk for substance use disorders. The team fed rare variant genetic findings into a tool called Strain Recommender, which predicts which mouse strains are more likely to show addiction-related traits. They then checked these predictions against real behavioral data from addiction studies and found strong agreement: Strains predicted to be vulnerable showed higher impulsivity, anxiety, sensitivity to novelty, and stronger responses to cocaine, along with learning deficits, compared to strains predicted to be resistant. The project demonstrated how All of Us data could feed back and improve preclinical research by helping researchers select better animal models, making addiction research more biologically relevant and effective.
Using All of Us to Enhance the Generalizability of Evidence Derived from Randomized Controlled Trials for Substance Use Disorder Treatment Interventions
Ali Jalali, Catherine Rabin, Caroline Andy, Colby Lewis V, Rachel Heise
Using data from a large opioid treatment trial as a test case, this team at Weill Cornell Medicine combined trial data with the more diverse All of Us dataset to estimate how two medications for opioid use disorder perform in routine care. When the trial results were applied to the broader population, the estimated benefits of the treatments were smaller or even changed direction, suggesting the original trial did not fully capture real-world effectiveness. This project showed how data fusion methods can be used to adapt results from a clinical trial to better reflect the diverse patients seen in everyday healthcare settings.
Investigating Pharmacogenetic Predictors of Substance Abuse
Elizabeth Atkinson, Yi-Sian (Helen) Lin, Astrid Manuel, Erik Stricker
Another winning project explored whether genes that affect how people process medications might also be linked to risk for substance use disorders. These researchers at Baylor College of Medicine studied common pharmacogenomic genes in the All of Us dataset and found that variation in six of these genes was associated with substance use status, with most showing a protective effect and one increasing risk. They also examined electronic health records and found that some genetic profiles linked to substance use were associated with higher rates of injuries, pain, and other medical conditions, suggesting that side effects or drug responses may play a role. Although deeper genetic network analyses did not show these genes as direct drivers of addiction biology, the findings highlight the importance of drug metabolism and side effects in understanding and predicting substance use risk.
Dr. Nora Volkow, Director
Here I highlight important work being done at NIDA and other news related to the science of drug use and addiction.
