Jacob Kantrowitz, MD, PhD
Jake co-founded River Records to build better software for physicians. He's a practicing internist, Assistant Professor at Tufts University School of Medicine, and a published researcher in clinical informatics and natural language processing.
His research on duplicate information in the electronic medical record — published in JAMA Network Open — showed that more than half of all text in a large academic health system's chart was copied from prior notes. That finding became the foundation for Stream's problem-oriented approach to clinical documentation.
When he's not working with River Records or seeing patients, you'll find him coaching his kids on the soccer field, baking bread, or riding his Peloton.
Posts by Jacob
68 articles
Chart Prep Shouldn't Take Your Staff an Hour Per Patient
Before every visit, someone reconstructs the patient's story from scratch — what's active, what happened last time, what's overdue. It's invisible labor, it's redone from zero every single visit, and it's pure reconstruction toil. Here's why it takes so long and what it should be instead.
Jul 12, 2026Where Referrals Go to Die
A referral is a handoff, and handoffs are where continuity of care quietly breaks. The letter, the fax, the follow-through, the note that comes back — or doesn't — and nobody tracking any of it. Here's why referrals vanish, and what it takes to make them something you can actually see.
Jul 12, 2026Risk Adjustment Is a Documentation-Structure Problem
Under Medicare Advantage and value-based contracts, a patient's risk score has to be rebuilt from documentation every single year. Conditions that aren't re-documented simply vanish — even though the patient still has them. That annual recapture problem is a persistence problem, and persistence is exactly what a time-organized note is worst at.
Jul 11, 2026The Undercoding Tax: What Invisible Complexity Costs an Independent Practice
Most independent practices bill below the level of care they actually deliver — not out of laziness, but because the note can't defend the visit. The undercoding tax is invisible, self-imposed, and large. Here's where it comes from and how documentation structure changes it.
Jul 11, 2026Your Denials Start in the Note
A denied claim feels like a billing-office problem — something that happens downstream, weeks after the visit. But most documentation-driven denials are decided at the moment of documentation, not the moment of submission. Here's why, and what changes when the note is built to support the claim.
Jul 11, 2026The Work Before the Work
The enormous, unmeasured labor that stands between a physician and the practice of medicine. On articulation work, extraneous load, moral injury, and why removing friction is different from resilience training.
Jul 10, 2026Normalization Is What Makes Personalization Possible
Consistency versus autonomy is a false tradeoff — it's a layer error. Customize the projection, never the artifact. Why structured data is the precondition for a chart that adapts to every reader.
Jul 9, 2026Medicine Doesn't Scale Because It Lives in Our Heads
Every good physician builds a private operating system — and it's completely non-transferable. Why care delivery doesn't scale, and what externalizing the work of care actually means.
Jul 8, 2026I Don't Read Charts. I Hunt Through Them.
Chart review is a re-orientation problem, not a reading problem. Why scatter — not overload — eats the clinical day, and what a normalized review surface would change.
Jul 7, 2026