Gastroenterologist | Medical Director, Oshi Health | LinkedIn Top Healthcare Management Voice | Forbes Health Contributor

I’m a board-certified gastroenterologist and the Founding Medical Director at Oshi Health, where I’ve had the privilege of helping shape and expand a virtual-first GI care model now serving patients in all 50 states. Working alongside an incredible team, I’ve contributed to building scalable clinical pathways for IBS, IBD, and other gut conditions—anchored in team-based care, empathy, and measurable outcomes.

With 28,000 + followers, I’m among the most followed gastroenterologists on LinkedIn, where I regularly share insights on topics including:

• IBS & the gut-brain connection — emotional triggers, stress, and personalized care
• Digital health innovation — virtual-first models, AI in GI, and wearable integration
• Colorectal cancer screening — rethinking pathways and improving early detection
• Reducing unnecessary testing — CTs, labs, and low-yield procedures in IBS
• Value-based gastroenterology — lower cost, better outcomes, patient experience
• Health plan integration — improving patient access to covered virtual GI care
• Microbiome & nutrition science — evidence-based takes on fiber, probiotics, and gut myths
• Healthcare system reform — tackling disparities, burnout, and reimbursement models
• Behavioral science in GI — the role of habit loops, psychology, and coaching in digestive health
• Virtual care leadership — insights from helping scale a 50-state GI operation

As a LinkedIn Top Voice, I collaborate across digital health, clinical care, and health plans to drive forward-thinking solutions in gastroenterology—without waste, and with humanity.

Let’s connect if you’re working at the intersection of GI, tech-enabled care, or value-based delivery models.

Michael Bass, M.D.

Most of us have taken antibiotics and moved on. This new Sweedish study says your gut hasn't.

A new Nature Medicine paper linked 8 years of prescription data to stool metagenomes from nearly 15,000 adults in Sweden.

The headline finding: a single course of clindamycin, fluoroquinolones, or flucloxacillin was still associated with changes in gut bacteria 4 to 8 years later. Not months. Years.

Clindamycin use within a year of sampling was tied to 47 fewer detected species on average. That's a massive hit to an ecosystem most patients assume bounces right back.

The species that increased after antibiotics included bacteria previously linked to higher BMI, elevated triglycerides, and type 2 diabetes risk. The ones that decreased were tied to lower inflammation and better metabolic health.

This matters because the gut microbiome isn't a bystander in human health. It regulates metabolism, immune function, and inflammation. When you wipe out key species with a broad-spectrum antibiotic, you're not just treating an infection. You're reshaping an organ system.

And the recovery isn't automatic. The fastest rebound happened in the first two years. After that, diversity plateaued well below baseline. Some species may never fully return.

We prescribe a 10-day course for a dental abscess and tell the patient they're done. But the disruption we triggered may still be measurable half a decade later.

If you've had a patient who "just hasn't been the same" since antibiotics a few years ago, this paper gives you a biological reason to take that seriously.
We can't undo the antibiotics we've already prescribed. But we can stop pretending the consequences end when the pill bottle runs out.

undefined undefined

gastroenterology-specialists-of-delaware-llc

The Gut doesn't Forget 

LinkedIn Top Voice | Gastroenterologist I Global Medical Director @ ViomeDx

Dr Erika Tajra

The most dangerous phrase in cancer screening right now might be “no cancer signal detected.”

Multi-cancer blood tests are having a moment. A single blood draw that claims to detect 50+ cancers sounds revolutionary. And maybe someday it will be. But the evidence is still catching up to the excitement.

There are now roughly 40 multi-cancer early detection tests in development or already on the market. Yet very few have been tested in randomized trials. And the first large trial results? They didn’t show a clear reduction in late-stage cancers. Which is the whole point. Finding cancer only matters if you’re finding it early enough to change outcomes.

Why is this so hard? Because early cancers are biologically quiet. Most current tests search for fragments of tumor DNA circulating in the bloodstream. The problem is that very small tumors often shed almost no DNA at all. The signal can be incredibly faint. You’re essentially trying to hear a whisper in a stadium.

That’s why researchers are exploring another layer of biology: RNA. 

RNA reflects what cells are actively doing, not just what mutations exist. Signals can appear earlier, even when tumor DNA is scarce. It can capture the body’s response to cancer, not just the tumor itself. And it provides insight into active biological pathways rather than static genetic changes.

None of this means DNA-based tests are wrong. But it highlights something important: detecting cancer early may require listening to more than one biological signal. DNA is probably part of the answer. It’s not the whole story.

Patients deserve innovation. But they also deserve evidence that it actually improves outcomes. Because “no cancer signal detected” only means something if the test was capable of detecting the signal in the first place.

The AI prescribing debate feels very different after you’ve cleared 40 refill requests in your inbox at the end of a clinic day.

Most people think prescribing is the clinical decision. Clinicians know the reality is everything wrapped around it. Prior auths, quantity limits, step therapy rules, plans that only cover 90 pills, missing labs, drug interaction checks, and the quiet liability of deciding what can safely go through and what needs a visit.

That’s why a recent The Washington Post  opinion criticizing Utah’s AI refill pilot with Doctronic caught my attention. Joseph Sakran , Johns Hopkins trauma surgeon and a medical researcher, argues that autonomous AI prescribing is moving forward on dangerously thin evidence.

His case is straightforward. The safety data rests largely on a single preprint on medRxiv that hasn’t been peer reviewed. The study authors hold equity in the company. The underlying data hasn’t been publicly shared.
They also point out that the widely cited 99.2% accuracy figure comes from urgent care encounters, not chronic medication renewals, which is what the Utah pilot is actually doing. Different context, different risk profile.

And they raise a fair concern about scale. Even a small error rate applied across thousands of prescriptions could affect vulnerable groups. Elderly patients, people on complex medication regimens, patients whose conditions have quietly changed since their last visit.

Those are legitimate concerns. But the authors also acknowledge something that gets lost in the headline debate. Medication renewals for stable chronic conditions are among the lowest-risk prescribing tasks in medicine.

If AI could safely handle routine renewals, it could free physicians to focus on the stuff that actually requires a doctor.

That tension is where the real conversation should be. Because the system clinicians are currently working in is already broken. 

Practices average nearly 40 prior authorizations per physician per week. Physicians and staff spend more than 13 hours weekly on them. Nearly 90% say prior auth contributes to burnout.

So I find myself in the middle on this one. I don’t want AI replacing clinical judgment. But I’m very open to AI reducing the administrative friction wrapped around routine prescribing. Stable medication renewals with guardrails and escalation paths? That’s not AI practicing medicine. That’s AI doing paperwork.

Those are two very different things. And until we can talk about them separately, this debate is going to keep going in circles.

Byron Crowe

graphical user interface, text, application

Everyone says AI will replace doctors. But the first thing physicians are actually using AI for? Reading the medical literature.

A new AMA survey shows more than 80% of physicians now use AI professionally. And the most common use isn’t diagnosis. It’s summarizing medical research and standards of care. That tells you something important about where medicine actually is right now.

The biggest bottleneck in healthcare isn’t intelligence. It’s bandwidth. Thousands of new papers publish every week. Guidelines constantly evolve. Biology keeps getting more complex. No physician can realistically read everything. 

So knowledge diffuses slowly through the system. Conferences, guidelines, word of mouth. Sometimes it takes years for a new insight to reach everyday clinical practice. I’ve seen it firsthand. A study changes how we should think about a condition, and five years later half the clinicians managing it haven’t read it yet.

AI changes that dynamic. Instead of reading 20 papers, a physician can read a two-minute synthesis. Instead of discovering a new mechanism years later at a conference, it can surface in clinical workflow the same week. That doesn’t replace physicians. It amplifies them.

Which is why most real-world physician AI use today looks nothing like the headlines,despite what the Silicon Valley tech- bros might be tweeting. 

It’s summarizing research, generating chart summaries, creating discharge instructions, reducing documentation burden. Not autonomous diagnosis. Not robot doctors.

Caitlin Owens highlighted this in a recent Axios Healthcare newsletter. AI’s earliest clinical adoption is largely about helping physicians manage information and administrative load, not replacing clinical judgment. That distinction matters more than most people realize.

Medicine isn’t moving toward autonomous doctors. It’s moving toward augmented cognition. Doctors still make the decisions. But AI helps them see patterns faster in an ocean of information that no human brain can process alone. And as biology gets more complex, that ability won’t just be nice to have. It’ll be essential.

No alternative text description for this image

Novo Nordisk  just agreed to let hims & hers  distribute Wegovy and Ozempic through its platform and dropped the lawsuit it filed last month over compounded semaglutide. This isn’t a GLP-1 headline. It’s a preview of where telehealth is going next.

The first era of telehealth was about convenience. Fast visits, easy prescribing, consumer-friendly access. The next era will be judged on something much harder: can telehealth become a trusted distribution and care layer for complex chronic disease?

For years, telehealth has been underestimated as a lightweight front-end. Good for access, good for demand capture, good for meeting patients where they are. But this deal signals something bigger. Large pharma now sees digital care platforms as legitimate channels for branded therapy distribution. Not lead generators. 

Because once telehealth becomes a true distribution channel, the business model gets serious fast. It’s no longer about customer acquisition and conversion. It’s about clinical credibility, regulatory discipline, patient safety, adherence, supply continuity, and what happens after the prescription is written. 

Especially in obesity care, where anyone can build demand but very few can build durable care models. The companies writing scripts the fastest aren’t necessarily the ones that will survive this next phase.

There’s a second signal buried in this story too. Just days after U.S. regulators warned telehealth firms over misleading compounded GLP-1 advertising, one of the most visible players struck a deal around approved branded drugs instead. 

Telehealth is no longer disrupting healthcare from the outside. It’s being absorbed into the system itself. That creates more legitimacy and more accountability in the same move.

Penn Medicine, University of Pennsylvania Health System

Gastroenterology Physician Assistant

Tower Health

Trinity

Sports Medicine at Chelsea

Provided primary care services for an active, health-oriented population under a holistic diagnostic and treatment model with emphasis on alternative/eastern medicine.

taplio

The Battle

Michael Bass, M.D.

Katherine Alexander, MHS, PA-C

Followers

Posts

Engagements

Likes

Performance ScoreMichael Bass, M.D.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Michael Bass, M.D.

Katherine Alexander, MHS, PA-C

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Michael Bass, M.D.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreMichael Bass, M.D.'s LinkedIn Performance