Our Best Posts

I move so I can still my mind.People often ask how I stay energized while leading a startup, conducting research, teaching, and investing in my own growth.

The truth is, movement has always been my medicine.

As a child, I ran every chance I had. As an adult, yoga found me when running no longer served my knees. It became my anchor.

Movement shifted from being about fitness to being about alignment, both physical and emotional.

What I did not expect was this: the more I moved, the more I could pause.Exercise taught me to listen. Meditation taught me to trust.And somewhere along the way, I became a better leader.

Not because I worked harder but because I learned when to stop.

So today, my self-care is intentional:

Daily movement for my body.

Quiet mornings for my mind.

A mindset that reminds me I am not here to grind. I am here to grow.

If you feel stretched or exhausted, try this:

Do not chase resilience. Create rituals that sustain it.

We recently published a manuscript on AI and nutrition, led by Diana Thomas, in The American Journal of Clinical Nutrition.

Working in digital health makes one thing clear. Wearable and sensor data are powerful, but they are not clean. They are shaped by device limitations, user behavior, context, missingness, and population differences. The promise of AI in nutrition and digital health depends not just on better algorithms, but on how honestly we confront those realities.

As AI adoption accelerates across wearables and consumer health platforms, there is pressure to move quickly from data to insight. But AI is not a shortcut to decision-grade evidence. When applied late in the pipeline or treated as a black box, it can produce confident outputs that obscure measurement error, bias, and subgroup instability.

This paper argues for a higher standard. AI modeling should be treated as a scientific design activity. Statistical, nutrition, and AI expertise should be integrated from the start.

Measurement error and uncertainty in wearable data should be modeled explicitly.

Outputs should be framed probabilistically rather than as definitive signals.

For leaders building or deploying digital health products, this is ultimately about trust. If AI-driven insights from wearables are going to inform clinical decisions, personalized nutrition, or population health strategies, they must be built with rigor, transparency, and respect for the populations reflected in the data.

Proud of this collaborative work and the role it plays in shaping more responsible, decision-ready AI for digital health and nutrition.

A wrong MFD can cost you more than a bad investment.



Here’s how to choose wisely...more

Imagine this: your portfolio drops 20% overnight. Your instincts scream “sell.”


A good MFD says, “Stay put.” ...more