Princeton Precision Health: An interdisciplinary, AI-driven approach to tackling big questions about health and disease
Scientists and engineers in the Princeton Precision Health initiative apply cutting-edge AI and computational models to massive datasets to develop a deep understanding of the factors that shape health and illness. Here, epidemiologist Jessica Metcalf looks on while colleagues shake hands at the inaugural team meeting of the 10-member PPH executive committee.
Human health is influenced by an extraordinarily complicated range of factors, from genetics and socioeconomics to air quality and lifestyle factors like exercise. Princeton Precision Health (PPH) is taking aim at this whole complex picture, bringing together a unique combination of experts, giant datasets and advanced computational methods to gain a comprehensive understanding of how to make and keep humans healthy.
The initiative’s 10 core faculty members include experts in sociology, psychology, computer science, engineering, genomics, environmental science, epidemiology and medicine, and PPH has awarded 22 endowment-funded seed grants to enable researchers from across the University to investigate health-related topics, such as the impact of technology on mental health and how computer vision tools can help diagnose autism.
Whereas faculty within a given academic department apply similar methods to different problems, “PPH aims to apply widely different methods and approaches to a common grand challenge,” said Matthew Salganik, a professor of sociology and core PPH faculty member.
The grand challenge: to achieve a deep understanding of human health — at the molecular, individual and societal level — using sophisticated computational methods that integrate all factors of health to reliably predict outcomes for individuals and groups.
The PPH approach centers on developing artificial intelligence (AI) and computational models to crunch massive datasets and understand the mechanisms behind how different risks and variables interact to shape health outcomes, Salganik said. With no immediate clinical goal and no patients, “We can understand interactions among various factors affecting health, addressing hard questions that take longer to study.
“This approach integrates rigorous models from various disciplines to uncover the why behind these interactions, making accurate and scientifically testable predictions,” he said. “The ultimate goal of precision health is to improve health for individuals and populations.”
Olga Troyanskaya, a professor of both computer science and the Lewis-Sigler Institute for Integrative Genomics, directs PPH. “The genome — essentially a code made up of four different letters — is about 3 billion letters long,” she said. “We have mapped the locations of most genes, but these account for less than 2% of the genome. The other 98% control when and how genes are activated. Until recently, our understanding of this 98% was limited. However, the emergence of deep learning models has transformed our ability to decode these mechanisms.
“We are now using AI models to uncover how genes are switched on and off and to predict when these activations will occur,” Troyanskaya continued. “At Princeton Precision Health, we take this further by integrating genomic data with other critical information — such as socioeconomic and environmental factors — to provide a more comprehensive understanding of health outcomes. This integrated approach is helping us dig deeply into a wide range of health challenges, including infectious diseases, autism, kidney disease and depression.”
The “perfect fit” for transformational change
While data has always driven health research, the sheer volume of health-related information and computational and AI advances make big leaps possible, said Mona Singh, a professor of computer science and the Lewis-Sigler Institute for Integrative Genomics and another core faculty member of PPH.
“The revolution that’s going on with neural networks, foundation models, language models — these AI techniques have shown remarkable performance in a way that was unimaginable even a decade ago,” she said. “It’s really exciting to think about all the different ways in which AI can help precision health reach its goals.”
And Princeton is ideally positioned to lead the charge, Singh said.
PPH Director Olga Troyanskaya (left) said the emergence of deep learning models has transformed the ability to decode "how genes are switched on and off and to predict when these activations will occur." An integrated, cross-disciplinary approach helps PPH dig deeply into health challenges including infectious diseases, autism, kidney disease and depression.
Princeton Precision Health (PPH) has distributed more than $2 million across 22 seed grants supporting 35 Princeton researchers, including these:
● Several researchers are partnering to determine if there is a causal link between widespread technology use and the rising mental health crisis in children and young adults — and, if so, what genetic, clinical, or environmental factors are contributing to the risk. Political scientist Jake Shapiro, neuroscientist Yael Niv, psychologists Molly Crockett and Kristina Olson, and PPH data scientist Manoj Kumar are working in this area.
● Ecology and evolutionary biologist Bryan Grenfell and computer scientist Adji Bousso Dieng are using new approaches and AI to understand individual variations in immunity and then extend this analysis more broadly to immune responses across entire populations.
● Sociologist Dalton Conley is investigating how a child’s genetic predisposition for educational attainment, ADHD and depression affects parental mood, marital quality and developmental investments in that child.
● Clinical psychologist Erik Nook is using AI’s large language models to quantify the most effective therapy for psychological disorders like anxiety and depression in teenagers, using thousands of hours of anonymized conversations between youths and therapists.
● Biologist Coleen Murphy is advancing research on what affects female reproductive aging by developing biomarkers for so-called “biological clocks” and exploring the social and environmental factors that influence them. She is collaborating with PPH’s Dr. Debra Baseman, an OB/GYN.
● Computer scientist Olga Russakovsky is using visual AI tools to better understand neurodivergence, from anonymized clinical videos. To date, more than 20 peer-reviewed journal articles have acknowledged support from PPH since its debut in May 2022, on topics ranging from COVID-19 to depression to kidney disease.
Health beyond medicine
Faculty from all four University divisions — Humanities, Social Sciences, Natural Sciences, and Engineering — already contribute to shaping the landscape of precision health research at Princeton. “Forty percent of faculty engaged in this research come from the social sciences and humanities, a key strength that sets Princeton apart,” Troyanskaya said.
Sociologists, psychologists, electrical engineers, and computer scientists might be unexpected collaborators for health research, but that’s part of PPH’s secret sauce, she said.
“By integrating expertise, methods and perspectives across disciplines, we can unlock deeper and otherwise inaccessible insights regarding health at the individual and population level,” said Erik Nook, an assistant professor of psychology and a PPH core faculty member.
It’s no coincidence that fundamental discoveries are happening at a university without a medical school, said Dan Notterman, M.D., M.A., a professor of the practice in molecular biology and another core PPH faculty member.
“PPH allows for interdisciplinary collaborations in a way which might not be possible in the tightly focused environment of the medical establishment, which is more disease-oriented, and may not have the latitude to explore the nuances of human genetics, behavior and the environment that affect health,” Notterman said. “At Princeton, we have the capacity to pursue research that requires a longer time frame and a broader, more diverse set of approaches.”
That paradigm explains why the research facility for PPH, located at 252 Nassau St., has no fume hoods or petri dishes but instead a sea of laptops interspersed with small gathering spaces and an open presentation space for their popular speaker series.
An infrastructure to tap vast datasets
To build their transformational understanding of human health, PPH researchers are drawing from some truly enormous datasets, including hundreds of thousands of complete human genomes and millions of disaggregated medical insurance claims — all with anonymized data and powerful safeguards around the records.
Datasets in health research present unique challenges for secure acquisition, storage and handling, said Salganik. No one researcher, however well resourced, could bring them all together and correlate the data from the disparate sources and disciplines. PPH brings together expertise — the data, clinical and research scientists — and resources to make this possible.
“Right now, an avalanche of diverse data — spanning genomic, molecular, social, behavioral and environmental factors, and clinical records — is being generated, but it remains siloed,” said Salganik. “PPH can provide the infrastructure, expertise and collaborative framework necessary to acquire, integrate and analyze these diverse datasets.”
PPH scientists have spent years getting these datasets to “talk to each other,” said Kara Dolinski, the executive director of PPH. Clinical data and environmental data, for example, provide information about very different aspects of health and are measured at very different scales. But with extraordinary effort and cutting-edge AI and computational techniques, the researchers can link, for example, which lifestyle factors interact with a particular genetic predisposition for kidney failure.
“Human beings are complex,” Nook said. “And PPH is transforming how we think about health research. It’s not just about isolated studies of genetics, behavior or environment anymore. It’s about integrating these factors at every level to gain a more comprehensive and predictive understanding of human health.”
Training the next generation
The PPH core faculty members are all professors, and they are looking forward to sharing this new way of thinking about health with Princeton students. The professors are developing innovative courses to equip students across majors with the skills to address complex problems in precision health using data-driven, computational and interdisciplinary approaches. The students and early-career scholars will carry PPH’s transformational approach into other academic, clinical and industry organizations — amplifying PPH’s impact far beyond Princeton, said Nook.
“Princeton students have a thirst for truly interdisciplinary training that will equip them to address both foundational and applied challenges in health,” said Troyanskaya, whose computer science class “AI in Precision Health” has enrolled students from majors including public affairs, anthropology, biology, chemistry and psychology — and has had a long waiting list each time she offered it.
“Our cross-cutting curricula and research, integrating data from across fields, will prepare students and early-career scholars to make transformative contributions in multiple health-related fields,” she said. “Future doctors, yes, but also students interested in biomedical research, engineering, health-related socioeconomic and policy fields, and public philanthropy will all have access to the same modeling tools and interdisciplinary insight — and will all learn to speak the same language.”
Guillermo Sapiro, a Distinguished Engineer with Apple, Inc. who is now Princeton’s Augustine Family Professor in Electrical and Computer Engineering, joined the Princeton faculty in large part because of PPH, he said.
“Before I joined Princeton, people would say to me, ‘But Princeton doesn’t have a medical school,’ and I would say, ‘Princeton has something better. Princeton gives its students interdisciplinary training to tackle modern challenges in human health, the knowledge and skills to grapple with data overload, disciplinary silos and the growing influence of AI,’” Sapiro said. “Through cross-cutting training, PPH will prepare students to make transformative contributions in medicine, biomedical research, engineering and health policy, empowering them to be leaders in diverse health-related fields.”
Sapiro’s research focuses on the relationship between health and human behaviors, including the use of tools like wearables.
PPH “is a super exciting initiative,” he said. “Princeton’s tradition of foundational, interdisciplinary research enables us to take a transformative quantitative and holistic view of human health — one that integrates cutting-edge science, data-driven insights and a deep understanding of the environmental, social and biological factors that shape health outcomes. In doing so, we are charting a path to transforming health in a way never done before.”
Poornima Apte contributed to this story.
Sociologist Dalton Conley (center) speaks while Dr. Daniel Notterman (left) and sociologist Matthew Salganik look on, at a meeting of the PPH executive committee.
Panelists speak at a conference on campus co-sponored by PPH.
From left: Clinical psychologist Erik Nook, psychologist and linguist Adele Goldberg, PPH Executive Director Kara Dolinski and sociologist Matthew Salganik at a meeting of the core faculty committee.
