Lee Hood’s Persistent Plan to Reinvent Medicine From the Ground Up

By Allison Proffitt 

February 26, 2026 | When Leroy Hood stepped up to the podium at last week’s Integrative Healthcare Symposium to accept the Visionary Award, he told the audience he had done 200 push-ups before coming downstairs that morning. He is 87 years old. 

It was a fitting opening for a talk that was equal parts intellectual autobiography, scientific tour de force, and provocation—a challenge to an entire healthcare system that Hood believes is looking at the problem of human health almost entirely backwards. 

“This is N-of-one medicine,” Hood said, describing the vision that has consumed the last two decades of his life. “And we are going to practice that in the future.” 

A Career Built on Paradigm Shifts 

Hood, founder of Phenome Health and one of the most decorated figures in modern biology, trace a career that has repeatedly forced the scientific establishment to catch up with him. As a young molecular immunologist at Caltech in 1970, he said, he became convinced that the complexity of human biology was the central problem worth solving—and equally convinced that solving it would require tools and frameworks that simply did not yet exist. 

“It was absolutely untouchable,” he told the audience. “One, we needed technologies that could extract from the individual their unique data. And two, we needed a new way of thinking about how you analyze data.” 

What followed was roughly thirty years of instrument-building. Hood’s lab developed six instruments capable of reading and writing DNA, including the automated protein sequencer—which enabled the cloning of previously un-sequenceable rare proteins and opened up five new fields of endeavor in human medicine—and the automated DNA sequencer that made the Human Genome Project technically possible. He was one of twelve scientists who met in Santa Cruz in the spring of 1985 to debate whether sequencing the human genome was achievable—and advisable. Ninety percent of biologists opposed it, Hood recalled; it took another five years to get the initiative off the ground. 

From there came the founding of one of the first genuinely cross-disciplinary biology departments at the University of Washington — made possible by Bill Gates — and then, after concluding that large institutional bureaucracies were structurally incapable of pursuing genuinely new ideas, the Institute for Systems Biology. 

The Institute for Systems Biology became the incubator for systems biology as a discipline: mapping the dynamic networks that underlie biological function, following the cascade of network changes that precede and accompany disease, and building the analytical frameworks needed to make sense of data at a scale biology had never before encountered. 

The 4 P’s — and the One That Remains Unsolved 

In the early 2000s, Hood said, he began articulating what he called the four P’s of medicine: predictive, preventive, personalized, and participatory. The first three, he argued, are now achievable — the science exists, the tools exist, and the data is beginning to exist. The fourth — participatory — remains the central unsolved problem in healthcare. 

“It is how you persuade patients, physicians, healthcare leaders, healthcare technology people, and ultimately politicians … to change our well-established disease-focused systems in line with wellness and prevention,” he said. The key: demonstrate the economics. Show that it works, prove the cost savings, and persuade a few leading-edge institutions to adopt it first. 

The scale of the opportunity, in Hood’s framing, is almost difficult to absorb. The U.S. spends $5 trillion annually on healthcare, with 86% of that consumed by chronic disease. “If in five years we can eliminate a quarter of chronic diseases,” he said, “we are saving trillions of dollars.” He drew a comparison to the Human Genome Project, whose return on investment, calculated a decade after completion, was estimated at $800 billion on a $3 billion investment. “I think phenomics will dwarf the ROI of the genome project,” he said. 

What the Data Already Shows 

The evidentiary foundation for Hood’s optimism comes substantially from the Arivale 5,000-person scientific wellness program his team has run over the past decade—the kind of longitudinal, high-dimensional data collection that he believes represents the future of clinical research (DOI: 10.1038/nbt.3870 and many more)  

Each participant was analyzed across six major data dimensions: genomic variants, blood proteins (initially around 500, now scaled to 5,000), metabolites, clinical chemistries, gut microbiome, and physiological data from wearables. The result was not just a collection of health snapshots but a dynamic picture of individual health trajectories — and a rich source of what Hood calls “actionable possibilities.” 

From the initial cohort of 100 people, Hood’s team identified 3,500 statistically significant correlations among analytes. Organizing these with community detection algorithms revealed roughly 70 distinct biological communities — clusters of analytes that co-vary in ways that map onto either normal physiology or specific pathological processes. One such community, centered on LDL cholesterol, included thyroxin as a correlated variable; the team hypothesized it as a potential drug target and later discovered that Eli Lilly was already in Phase 3 trials on a thyroxin-based cholesterol drug. The point, Hood said, is that drug targets are discoverable from this kind of data — without anyone having to guess. 

Among the most clinically striking findings from the program: 

Biological age diverges meaningfully from chronological age. The team developed methods to calculate biological age from blood proteins, metabolites, and other analytes. Participants in the wellness program lost, on average, roughly one year of biological age per year enrolled. More striking, individual organs could age at markedly different rates from each other and from the global biological age estimate — creating the possibility of organ-specific recommendations. Every disease examined in the study was associated with a significant increase in biological age. 

Microbiome health predicts survival in the elderly. In a study of approximately 9,000 individuals aged 60 and above, Hood’s team found a striking divergence in microbiome trajectories between healthy and unhealthy agers. Healthy individuals progressively lost their baseline microbiome from middle age and developed a unique, individually differentiated microbiome composition. Unhealthy individuals showed no such differentiation. Over a four-year observation window, those with unhealthy microbiome profiles were four times as likely to die. (DOI: 10.1038/s42255-021-00348-0) 

Cancer signals appear in blood years before diagnosis. Perhaps the most consequential finding: examining blood samples collected one to four years before participants received a cancer diagnosis, Hood’s team identified proteins that were “strikingly elevated”, sometimes mapping onto known disease networks characteristic of specific cancer types. They documented 167 wellness-to-disease transitions in the dataset, including 35 cancer transitions. “The only answer to most diseases is going to be prevention or early diagnosis and response,” Hood said. “We are never going to cure disease with just a single drug.” 

He described disease progression as a long pre-clinical period during which disease-associated biological networks expand exponentially, quietly, invisibly, before any symptom appears. By the time a clinical diagnosis is made, he said, “you have a really complicated set of disease networks,” and targeted therapies are playing catch-up against a problem that has already become complex. The intervention window, he argued, is during that earlier phase, when networks are simpler and potentially reversible. 

The Blood as a Window Into Every Organ 

A central organizing concept in Hood’s framework is that blood, flowing through all organs continuously, carries organ-specific molecular signatures, proteins secreted at detectable levels by each of the body’s approximately 25 major organs. By tracking these proteins longitudinally, Hood argued, clinicians can monitor the functional state of individual organs in ways that today’s medicine cannot. 

The brain, he noted, is a particularly interesting case: under normal conditions it produces almost 600 organ-specific proteins, but fewer than 10 of the corresponding proteins appear in blood. When the blood-brain barrier begins to lose integrity, more proteins escape into circulation, and the specific proteins that appear indicate where the barrier has been breached. In Alzheimer’s disease, he said, metabolic signals can appear in blood as many as 15 years before clinical diagnosis. The placenta, by contrast, produces 37 organ-specific proteins, offering an unprecedented window into the dynamics of fetal development.  

The Genomic Revolution, Continued 

Hood also outlined the next frontier of genomic analysis, enabled by long-read sequencing technologies. While most genomic research to date has relied on short-read sequencing, which struggles with high-GC content regions and cannot fully resolve structural variants, long-read sequencing allows complete assembly of individual genomes, comprehensive methylation analysis across every cytosine, phasing of maternal and paternal chromosomes, and characterization of the full landscape of insertions, deletions, and rearrangements that correlate with disease. 

“Epigenetics isn’t just methylation,” Hood said. “It’s going to be many other things.” Long-read sequencing, he suggested, will open up new dimensions of epigenetic analysis that current methods cannot access. His team is also incorporating viral history reconstruction, brain health biomarkers, voice-based disease stratification (drawing on work from Luxembourg collaborators who have done large-scale studies in diabetic patients), and multispectral saliva analysis capable of distinguishing healthy from diseased individuals in three seconds. 

The synthesis of all these data streams, Hood argued, will enable a level of biological integration and a density of actionable health insight that is qualitatively different from anything medicine has attempted before. 

Toward a Phenome Project 

Hood closed by outlining an ambitious programmatic agenda under the banner of a “billion-person genome and phenome project,” with current pilots underway in South Korea, South Carolina (a diabetes-focused study of 1,000 individuals), Tampa General (60 heart failure patients), and very early stage rural health initiatives in Alabama and Montana tied to federal funding announced by the Trump administration. 

The clinical trial model he envisions differs fundamentally from the conventional randomized controlled trial. Rather than enrolling a homogeneous population, he described a “broad patient spectrum” design that includes individuals at high risk with normal biochemistry, those with pre-disease states, and patients at early, middle, and late stages of a given disease and following all of them longitudinally for five years.  

“This is how all clinical trials should be done in the future,” Hood said. “You’ll get the life history of all the different subtypes of the condition being studied.” He argued that with the density of data now achievable, compelling results are possible with as few as 100 to 200 participants. 

The message was consistent throughout: medicine’s future lies not in better treatments for established disease, but in understanding each individual deeply enough—and early enough—to prevent disease from establishing itself in the first place.