Welcome to the Blog

What a cross-country journey taught me about the distance between health advice and the lives people actually lead.

{Read the full post here: https://evrahealth.com/news/from-on-healing-to-a-view-between-villages-across-america/}

In January 2025, I began traveling across the country for the publication of my book, On Healing. The book grew from a question I had been asking for years as a physician, journalist, researcher, and patient: What actually helps us heal?

The answers were never confined to a clinic. Healing was shaped by sleep and food, movement and meaning, relationships and grief, the places where people lived, and whether they felt seen. On the book tour, from New York and Connecticut to California, Arizona, Florida, Massachusetts, Chicago, and Colorado, I heard people describe health in similarly expansive terms. They wanted good medical care, of course. But they also wanted time, trust, community, clear information, and some sense of agency.

The tour ended back in April 2025, and seeded the idea for Evra as an AI-enabled health platform designed to connect the dots between a person’s medical records, laboratory results, medications, wearable data, daily habits, and social circumstances and turn them into practical next steps, with human care when it is needed. As Evra took shape, our team began hearing directly from users across the world, including from all over the U.S. I had more time to learn from users in Maine and Texas as well. After spending most of the first part of Summer 2026 in San Francisco, engaging with partners and investors, it became clear that Silicon Valley is a bubble for wellness and longevity, and immensely disconnected from how most people navigate health day-to-day.

So I went back on the road – my first idea had been to rent a VW minibus and drive it from Northern California across the Mountain West, Midwest, and South with some people on my team (or perhaps my partner would volunteer – he hesitated, understandably). The next best thing was to simply fly to some of the places I hadn’t travelled to in 2025, especially the places that are currently debating healthcare and many other issues ahead of the midterms.

Idaho

The Evra Health ‘Listening Tour’ picked up where the On Healing tour left off. I traveled through the intermountain West, Midwest, and South, speaking with people in cities and small towns, at farmers’ markets, hotels, salons, festivals, parks, cafés, and in the back seats of rideshares. I asked what made it hard to stay healthy, what people wished the healthcare system understood, and what kind of support would genuinely help.

The route covered thousands of miles: from Reno, Mammoth Lakes, Bozeman, Jackson Hole, Salt Lake City, southern Utah, to Des Moines, Omaha, Ann Arbor, Denver, Indianapolis, Louisville, and Nashville. New Hampshire and Vermont would come next. Conferences will take me to New Orleans next week and Wisconsin in October, and I’ll surely learn more.

Continue reading at: https://evrahealth.com/news/from-on-healing-to-a-view-between-villages-across-america/

Americans are already using AI to fill gaps in care. What comes next will shape trust, access, and outcomes.

Artificial intelligence is already changing health care, in hospitals and where 99% of care and management happens: where we live, work, and play.

AI is already acting as informal health infrastructure. Physicians use tools such as OpenEvidence and Doximity to synthesize rapidly evolving medical literature. AI can also reduce administrative burden through chart review and ambient documentation tools and is thus already embedded across the health ecosystem.

AI is also stepping into the space ‘between healthcare visits’ by default. So, we must build systems that are accurate, ethical, and worthy of public trust. People are turning to chatbots for health advice because the traditional system has left a vacuum. For better or worse, AI is filling a long-standing gap around health information, continuity, and access, even when privacy guardrails are risky.

As both a physician and a startup founder working at the intersection of AI and health, I see five forces shaping where this goes next. Together, they will determine whether AI becomes a force for healing, or another accelerant of inequity and mistrust in healthcare.

First: AI health solutions must be capable of integrating a user’s different health needs.

Focusing narrowly on a single condition — menopause, diabetes, weight loss — ignores the reality that most are managing multiple, overlapping health issues, either at once or over their lifetime. Diabetes during menopause, for example, is not the same as diabetes alone. The single biggest women’s health issue is heart disease: it kills one woman a minute in the U.S., and many women are not aware of how cardiac disease symptoms differ between genders.

As well, solutions must meet people where there are now. Two moments crystalized this for me. The first came in 2019, in the back of a cab in Franklin County, New York, one of the poorest regions in the state. The driver, a 70-year-old man with diabetes, told me he was rationing insulin because he couldn’t afford it and couldn’t figure out how to stick to lifestyle changes his doctor recommended. What struck me wasn’t that he lacked a smartphone (he had one) but that he lacked clear, actionable guidance and sustained support. He wasn’t listening to biohacker podcasts. He simply wanted to live long enough to spend more time with his grandchildren.

The second moment came years later, while I was on book tour for On Healing. I spoke with more than 500 people across the country about health and what actually gets in the way of change. They shared that the wellness echo chamber served a narrow, privileged slice of the population and the problem wasn’t information. They wanted simplicity: clear steps, grounded in science and trust, tied to outcomes.

Second: science is under attack, and rebuilding trust must be part of the work.

Much of the anxiety around health AI stems from something deeper than technology: mistrust, often secondary to misaligned incentives for healthcare providers to provide treatment that is not always in the best interest of the patient. There are millions of Americans for whom the medical system has failed people who were rushed, overprescribed, or steered toward interventions they did not need. I experienced it personally myself in 2020 when a surgeon attempted to upsell me expensive surgery for my broken wrist when I only needed a cast (thankfully I could advocate for conservative management). Everyday Americans like Diane, whom I interviewed, based in Asheville North Carolina, spoke to the same mistrust that led her and millions of Americans to seek alternative treatments, and heard versions of the same story across the country.

We have to hold two truths at once. Mistrust in the health care system is real and, in many cases, earned. That mistrust has increasingly spilled over into skepticism toward medicine and science as a whole. This is not a partisan problem. And this is not only about vaccines.

Trust cannot be rebuilt through authority alone. It requires transparency, evidence, and accountability. That makes rigor in health AI non-negotiable. Foundational models, including those developed by companies like Anthropic and OpenAI, represent an extraordinary leap forward, but they are not designed to coach and guide people towards healthier behaviors; they can’t boil the ocean. Hallucinations remain a real risk in health contexts, where the cost of error is measured in lives.

To rebuilt trust, health AI solutions must be grounded in clinically validated data, transparent methods, and continuous evaluation. Safety in medicine has always required constraints, not scale alone.

Security and privacy are equally central to trust. The recent 23andMe data breach was a warning. Health data is among the most sensitive data we have. Without strong guardrails, trust collapses.

And the threat is not only domestic. China is already deploying doctor chatbots at scale. Health is wealth. Health is national security. Investing in building secure, science-based domestic health AI is strategic, and bipartisan support for innovation is crucial.

Third: health AI must be researched like a public-health intervention.

Most startups do not conduct independent research or test outcomes over time, and the ones that do rarely compare their offering to placebo or to a competitor.

Yet health AI demands that level of scrutiny. My colleagues and I recently released two research papers in preprint examining the state of AI in chronic disease: one analyzing current use and limitations, the other exploring how AI could support sustained behavior change.

The message is simple: effectiveness cannot be assumed. If a product claims to improve health, it should be able to show how, for whom, and under what conditions. The recent efforts like MAHA Elevate and CMS Access place outcomes as central, and this is a good way to support efforts that are hype-free. High impact medical journals, which institute rigorous peer review, are an asset to the public and to health AI companies seeking to make a sustainable impact in population health.

Fourth: global health cannot be an afterthought.

AI makes it possible to think differently about scale. Health challenges do not stop at borders, and with cuts to global health initiatives provides an opportunity for social enterprise to fill the gap. Our team has made it a mission to contribute to global health improvement through launching concurrently across countries across the developing and developed world. In places like India, where chronic disease burden is high and mobile access is widespread, AI-enabled guidance can reach populations long excluded from continuous care. Building for multiple populations from the outset forces better science and exposes hidden assumptions.

Finally: AI offers a once-in-a-generation opportunity for access and equity, if values are explicit.

The recent government shutdown placed healthcare access at the core. AI will not fix the health system on its own. But it can widen the doorway with bipartisan support. That promise depends on the values embedded in these systems from the start. Design choices about what data to include and prioritize matter. For instance, our team has decided to exclude race and ethnicity in our intake forms. Race is not the cause of inequitable outcomes; racism and structural discrimination are, and we’ve learned painful lessons around what happens when race is embedded in AI algorithms in healthcare. Increasingly, evidence suggests that income constraints, lived experience, family history, and geography are more predictive of chronic disease risk than race alone. We’ve made our values explicit, and hope others will do the same.

As we make our New Year’s resolutions and look toward 2026, I’m reminded of a moment just before the pandemic began. In early 2020 — only a month before COVID reshaped daily life — I wrote about the importance of coaching for New Year’s resolutions. I never would have imagined that within six years, it is conceivable that every person on the planet could have access, for the first time, to credible health guidance in their pocket: a science-backed, privacy-assured health companion to help them make better decisions, every day.

That future is not inevitable. It will depend on whether we choose rigor over hype, ethics over speed, and public good over short-term gain. But if we get it right, we won’t just be building better AI. We’ll be rebuilding trust in health itself, and making chronic disease less inevitable than it has ever been before, for every one of us.

Read this on Medium.