Remote patient monitoring finally pays. Here's what's worth building now.

For most of a decade, remote patient monitoring lived in pilot purgatory. A health system would buy two hundred blood pressure cuffs, enroll a few wards of patients, publish a press release, and quietly wind the program down eighteen months later when the grant money ran out. The technology was rarely the problem. The business model was.

That changed over roughly ninety days this winter. On January 1, Medicare's 2026 fee schedule introduced billing codes that pay for monitoring from as few as two days of readings a month, replacing a 16-day threshold that made partial patient adherence worth exactly nothing. In February, Congress extended the hospital-at-home waiver for five years, through September 2030. Hospitals that had been holding budget back because the program might vanish next quarter can now plan in years.

If you build software or connected devices for healthcare, this is the corner of the market where budgets are actually moving in 2026. It pays to understand the mechanics, because the way reimbursement works dictates what's worth building.

The reimbursement math finally works

The old rules had a cliff in them. CPT 99454, the device supply code, required at least 16 days of readings in a 30-day period. CPT 99457, the management code, required 20 minutes of clinical staff time. A patient who took readings for twelve days generated real clinical value and zero revenue. Programs either chased patients to hit thresholds or wrote off the month.

The 2026 Physician Fee Schedule fixed that. New code 99445 covers 2 to 15 days of readings. New code 99470 covers the first 10 to 20 minutes of management, as long as there's one real-time interaction with the patient that month. Neither stacks with its older sibling; for each patient, each 30-day window, you bill one or the other depending on what the data shows. CMS also opened RPM billing to federally qualified health centers and rural health clinics, which puts monitoring within reach of clinics serving patients who arguably need it most.

Then there's hospital-at-home. The Acute Hospital Care at Home waiver, born as a pandemic measure, kept getting extended in six-month increments that made it impossible to invest in seriously. The Consolidated Appropriations Act signed in February 2026 extended it to September 30, 2030. There are currently 366 approved programs across 139 health systems in 37 states, and every one of them just got permission to think long term. Continuous monitoring is the infrastructure layer that makes acute care in a living room safe, so that five-year runway flows directly into device and platform spending.

Analysts put the global RPM market at around $67 billion in 2026, heading toward $118 billion by 2033. I'd treat any seven-year healthcare forecast as decoration, but the direction matches what you can verify: about 46% of US hospitals already offer some form of RPM, and the policy tailwind just got stronger.

The evidence is better than you might expect, with caveats

Healthtech has a long history of products that demo well and change nothing clinically. RPM has cleared a higher bar than most. The largest recent study, a retrospective cohort of 6,595 hypertension patients, found that 66.3% had uncontrolled blood pressure at enrollment. After at least 90 days on monitoring, that fell to 40.2%. Systolic pressure dropped 7.3 mmHg on average, and 16.7 mmHg for stage 2 patients. Those are numbers a cardiologist will take seriously.

Heart failure programs report readmission reductions around 50%, and broader chronic care cohorts show cuts in the 20% range. Since a single heart failure readmission costs a hospital five figures and Medicare penalizes excess readmissions on top, the return on investment argument mostly writes itself.

Now the caveats. Most of this evidence is retrospective, some of it comes from vendors studying their own programs, and there's a selection problem baked in: patients who keep taking readings for 90 days are not a random sample of patients. Randomized trials have been more mixed. But here's the thing that matters commercially: payers have stopped arguing. When CMS expands billing codes two years in a row, the debate about whether this category is real is over.

Why programs still fail

Talk to anyone who has run an RPM program and they'll tell you the same thing: the dashboard was the easy part. Programs die on adherence. The cuff sits in a drawer. The scale's battery dies and nobody replaces it. The patient takes readings enthusiastically for three weeks and then stops, and every gap in the data erodes both the clinical picture and the billing.

Studies on dropout keep finding the same culprits: devices that need charging and pairing, patients who never got real onboarding, and clinics that don't have staff hours to chase missed readings. The 20-minute management codes quietly assume somebody is doing the managing. In a small practice, that somebody is often nobody.

It's telling that CMS designed the new codes around this exact failure mode. The 2-to-15-day codes are an admission that partial adherence is the normal case, not the exception. For builders, that's the brief: the product that wins is the one that makes twelve days of readings billable, useful, and a step toward sixteen, instead of treating them as a failed month.

If you're building in this space

A few opinions from the building side, having shipped healthcare systems that live under HIPAA (including work for US government healthcare clients we can't name):

• Don't start with hardware. FDA-cleared cuffs, scales, pulse oximeters, and glucometers with APIs already exist from multiple manufacturers. Your defensible layer is workflow, adherence, and billing logic. Building a new device adds 18 months and a regulatory pathway before your first user.
• Choose cellular over Bluetooth for the Medicare population. Every pairing screen, app install, and Wi-Fi password costs you patients. A cuff that transmits the moment you press its one button will beat a prettier app on adherence every time.
• Treat the billing engine as core product, not an afterthought. Tracking days of readings and minutes of management per patient per 30-day window, and picking 99445 versus 99454 when the codes can't be combined, is exactly the kind of logic that quietly leaks revenue when it lives in a spreadsheet.
• The data has to land inside the EHR, in the clinician's existing workflow. Certified EHRs expose FHIR APIs; use them. A standalone portal that nurses have to remember to open is how monitoring data goes unseen until the lawyers ask why.
• Budget for compliance as a feature, not a tax. HIPAA governs the US side, and if your device ships to the EU, the Cyber Resilience Act's reporting obligations start this September. Retrofitting either one is far more expensive than designing for them.

Notice what's not on that list: AI. There are real applications, mostly in triaging which of ten thousand daily readings deserves a nurse's attention. But teams that lead with the model and treat device logistics as a detail have it backwards. In this market, an unglamorous returns-and-refurbishment process beats a clever algorithm.

The window here is specific. Reimbursement just turned favorable, the evidence argument is settled enough, and the incumbent platforms are still built around the old 16-day assumptions. That combination doesn't stay open long. The teams that win it won't be the ones with the most impressive demo. They'll be the ones whose cuffs come back charged, whose data lands in the chart, and whose invoices survive an audit.

Care is moving home. The budgets finally followed. Build for the drawer the cuff ends up in, not the demo.