A Moment with Eric J. Daza is part of our interview series featuring thought leaders in research and healthcare. Each interview includes 7 short and stimulating questions.

Dr. Eric J. Daza is a biostatistician and health data scientist at Evidation, a digital health company. He has worked for 20 years in both industry and academia, in pharma clinical trials, survey sampling, nutrition, maternal/child health, global/international health, health promotion & disease prevention, healthtech, digital health, and behavioral medicine. For more from Dr. Daza, find him on Twitter.


1. Tell us something we don’t know. (Anything!)

The American English word “boondocks” (as in “the boonies”) comes from the Filipino Tagalog word “bundok” (boon-DOCK). I’d wondered for years about how weird it was that “the boondocks” simultaneously meant “a remote rural area” and strangely sounded almost exactly like the word for “mountain” or “mountains” that I grew up saying in Metro Manila. Apparently, the American cognate was adopted by the U.S. military during the Philippine-American War—the first American war in Asia.

2. Which fiction book would you recommend to researchers and innovators in healthcare, and why?

I recommend reading the Foundation Trilogy (if not Series) by Isaac Asimov.

I had no idea I’d go into statistics as a career! I was fascinated by neurobiology in high school, and in college I majored in neurobio and cognitive studies. I can’t remember when, but somewhere in my young and early adulthood, I enjoyed my travels through these sweeping and iconic sci-fi stories.

The science-fictional field of “psychohistory” (not to be confused with the academic sub-field) plays a central part in their milieu. Asimov’s psychohistory is something of a mash-up of statistics, sociology, economics, and history, with a dose of Strathern/Goodhart’s Law for good measure.

Healthcare researchers and innovators might glean some inspirational nuggets of techno-social insight from Asimov’s pages. More importantly, these stories can help self-reflect on how your innovative idea today might set the course for entire systems and societies for centuries (and beyond)—for better and worse.

They’re also just great fun, originally having been written as pulp sci-fi stories set to the same galactic backdrop. In fact, a new TV series based on these novels came out in Fall 2021.

3. What are you working on right now that you’re excited about?

Stats-of-1! It’s a blog I created about personalized health—specifically, on how to collect and analyze recurring, quantifiable patterns in your own health or fitness data. Together with my co-editors Clair Robbins and Julio Vega, we are building a community of statistics-minded researchers and citizen scientists who have methodological ideas to share and co-develop.

Our mission is “to facilitate cross-disciplinary collaboration that will enhance idiographic data collection and analysis procedures across health disciplines. We call this statistical field esametry.”

Subscribers get updates whenever we post articles, which tend to be 3-5 minute non-technical reads. We try to post at least one article a month—and definitely welcome guest contributors aligned with our mission.

4. Who’s doing something that you admire in healthcare today, and why is it so cool?

Evidation is amazing. I currently work there as a health data science biostatistician. After 18+ years of working in pharma, clinical trials, public health, behavioral health, health tech, and digital health, I can say it’s one of my absolutely favorite places to work.

Why? For one thing, Evidation’s mission “to create new ways to measure and improve health in everyday life,” along with its technical infrastructure, capabilities, and interests, significantly overlap with and complement my own digital-health ambitions and skills (see Stats-of-1).

But more importantly, we also practice truly ethical innovation.

We prioritize collaboration, teamwork, and transparency alongside efficiency. We actively self-reflect to identify where we can improve in leadership, management, operations, research, and hiring—and then actually do the things we say we’ll do. It’s no surprise that I love working with my team, and with others across the wider org.

5. What’s the biggest barrier to getting things done in your line of work?

Aligning on common definitions and specifications for outcomes and analyses that matter for monitoring, maintaining, and improving health is a critical barrier to progress.

Current ambiguity adds noise to the evaluation of the safety and efficacy of new digital interventions and therapies. Lack of agreement on digital health metrics means we’re still largely dealing with moving targets. Lack of consensus on featurization techniques and analysis tools is like trying to do chemistry with hand-crafted glassware: Each piece yields results that cater to the alchemist’s own tastes—each an intricately biased tool that feigns true scientific consensus and objectivity. (My thanks to Michael Hudgens, my dissertation advisor, for the core glassware analogy.)

I heard a Digital Medicine Society talk once that put it well: A heartbeat is a heartbeat; it shouldn’t matter who made the sensor or device. In line with this, our scientific and regulatory systems hinge on having valid and reliable measurements of bio-behavioral phenomena. For any given observation, these metrics should vary little (if at all) across sensor/device manufacturers and digital health organizations.

We haven’t yet standardized outcomes and analyses, which makes it hard to definitively test possibly innovative solutions. But organizations like the Food and Drug Administration (FDA) Digital Health Center of Excellence, Digital Medicine Society, and Evidation (working with the American College of Cardiology) are working to change that.

6. Imagine you win an award for impacting healthcare. What did you do?

I helped bring digitally enabled, individual-focused “n-of-1” real-time health and fitness monitoring, diagnosis, and treatment to standard health promotion and clinical practice—for everyone, not just those privileged like me. Think of behavior-change recommendations, treatment plans, etc., that are truly scientifically rigorous in assessing causation, not just correlation, for each person based on their own unique recorded health history. On a grander scale, I helped shift the focus of healthcare research from our traditional population-based “nomothetic” statistical study designs and methods to individual-based “idiographic” ones (where sensible and feasible).

Some of my colleagues might call for an end to the dominance of population-based designs like randomized controlled trials (RCTs) as gold standards for determining an intervention’s efficacy or pragmatic effectiveness. But RCTs make plenty of sense if the intervention’s effects are likely to be almost identical among those it’s intended to help. And interventions targeting endpoints like death can’t be tested with individual-based studies that require endpoints to be reversible. The latter include symptoms of pain, migraines, and irritable bowel syndrome, each with idiosyncratic triggers.

That said, if we can ethically record valid and reliable metrics for each person over time, we’d be in great shape to assess truly individualized treatment effects. We could meaningfully improve the quality of daily life for folks suffering from distinctive recurring conditions.

How? Through quantitative idiographic (QI) approaches like n-of-1 trials and single-case designs, and their cousins, the individually adaptive approaches of JITAIs, MRTs, and SMARTs. QI approaches beat traditional population-based ones when intervention effects are so varied and heterogeneous that there is no clinically meaningful “average” effect that most people will experience. In perhaps the worst cases, the traditional mean endpoint would show “no average effect of the intervention”, when in fact there was a very real effect for a lot of participants—just in opposite directions. Your pain was my gain, but in the (analytical) end, we canceled each other out!

QI approaches could also improve “precision medicine” by helping discover groups of patients that respond in particular ways. Alongside their genetics, each patient’s own measurable health habits and behaviors can be used to define their own baselines for tailoring diagnoses and treatments; the “nature and nurture of you”. QI studies could also better disaggregate group-based health disparities from the bottom up: Start with each individual’s own quantifiable experiences, then aggregate up to find general patterns useful for making more nuanced community-wide recommendations, practices, and policies.

Todd Rose in The End of Average calls this the “analyze-then-aggregate” approach—in contrast to the traditional “aggregate-then-analyze” approach of statistics in biomedicine and clinical trials. Still, the idiographic methods I and others espouse do actually “aggregate-then-analyze”. But they do so for one person first, over time, before subsequently aggregating over an entire group. They privilege the individual’s own average within the context of all others.

7. What advice would you give innovators in healthcare?

Really grok the fundamental concepts of statistics—especially if you’re mesmerized by the allure of machine learning and artificial intelligence (AI). It’s super easy to “lose sight of the (random) forest for the trees.”

I’m not talking p-values, confidence intervals, and math in general, though these are definitely important. Rather, take time to habitually meditate on key statistical science concepts. Try to think about them when you read an article or report, or do an analysis. Pay attention to where your mind wants to go, and if your intuition agrees.

Two concepts in particular are highly counterintuitive.

Correlation does not imply causation” is such a popular aphorism precisely because we naturally think and feel otherwise. It’s a constant reminder to change a deep-seated natural habit: One that helps us individually navigate day-to-day situations, but makes us misinterpret analysis results.

I also like to remind myself and others that “significance does not imply importance.” Specifically, statistical significance, a technical type of significance that is important for doing science right, has nothing to do with practical, scientific, clinical, or business significance on its own. But we naturally—wrongly—equate the two all the time! It’s such a strong instinct that even deeply trained statisticians like me get it wrong from time to time.

Why do these two concepts matter? Because flawed statistical takeaways generate misguided decisions that help set the course for entire societies and histories, well beyond just science and academia.

Think about dietary and fitness advice, the quality of the food you eat, and how you move, exercise, and sleep. Think of the safety or danger of where you live or work, of how you get there and back. Do you walk or bike? Drive, or take public transit? Think of how easy or hard it is to see health providers like therapists, nurse practitioners, physicians, and dentists, and of how they decide what drugs to prescribe or procedures to do. Think of your family and friends—even your ancestors—and their own related experiences.

All of these are constantly shaped by “societal habits” like social policies, regulations, and laws, institutional and institutionalized norms, and business practices. (These businesses include pharma/biotech and healthcare, sure, but also social media companies.) Societal habits form your health context and milieu. They guide your behavior and affect your health on a daily basis—whether you like it or not.

And these societal habits are built, maintained, changed, and ended by your fellow flawed and biased humans, the decision-makers. They (working with analysts like me) must constantly repeat these two aphorisms as reminders whenever we use data both big and small to do statistical analyses—which include machine learning and AI. Our interpretations of results and the decisions they support codify and set in motion laws, guidelines, programs, and products with societal impacts spanning years, if not decades.

That’s how these two pithy, seemingly trite aphorisms help shape the health of entire communities and societies year after year, decade after decade. Asimov would approve.

For these reasons, you should internalize these aphorisms. At any given moment, it can be all too easy to be deft at code, models, and math, and also daft at statistics—even, on occasion, for statisticians like me. By correcting your natural tendencies to misinterpret statistical results, you’ll improve how you understand and value progress on your own innovative ideas that can and do shape our world.


About Dr. Eric J. Daza:

Dr. Eric J. Daza is a biostatistician and health data scientist at Evidation, a digital health company. He has worked for 20 years in both industry and academia, in pharma clinical trials, survey sampling, nutrition, maternal/child health, global/international health, health promotion & disease prevention, healthtech, digital health, and behavioral medicine.

Dr. Daza also created and edits Stats-of-1, a health statistics blog focused on digital within-individual statistical designs or methods (WISDOM). He investigates how to discover causal relationships from an individual’s own wearable device, sensor, and app data. He is also a member of the International Collaborative Network for N-of-1 Clinical Trials and Single-Case Experimental Designs.

As a privileged middle-class Brown Asian immigrant, Eric Jay earned both his BA in Neurobiology / Cognitive Studies and MPS in Applied Statistics at Cornell University, followed by his DrPH in Biostatistics at the University of North Carolina at Chapel Hill. He then trained as a postdoc at the Stanford Prevention Research Center. He is also Jesuit-trained.


Written by: Aline Holzwarth