DEV Community: Daniel Salzberg

Healthcare's Next Decade Will Be Measured, Not Guessed

Daniel Salzberg — Mon, 09 Nov 2026 16:29:29 +0000

Originally published at danielsalzberg.com.

I've spent much of my career watching healthcare buy technology faster than it can absorb it. Health systems have spent billions on EMRs, portals and analytics platforms, and clinicians are still charting late into the night. The next ten years won't go to the organizations that adopt the most technology. They'll go to the ones that turn it into better decisions where care happens.

Here's where I think things are going.

From snapshots to continuous data

A lot of medicine still runs on snapshots. A patient comes in, we take a few measurements, ask how they've been feeling, and decide based on about fifteen minutes. What happened between visits mostly stays invisible.

That's starting to change. Remote monitoring, wearables and measurement tools in the clinic give clinicians a running picture instead of a single frame. The point isn't to collect more data. It's to know sooner whether a patient is getting better, leveling off or slipping, and to adjust care before a small problem turns into an expensive one.

Objective measurement instead of "how does it feel?"

Some of the most important questions in medicine still get answered by feel. How well is this patient moving? Is their balance better than last month? Is the therapy working?

What patients report matters, but it's hard to compare over time and between providers. Objective measures of movement, function and recovery give clinicians a common language and a real baseline. They also give patients something motivating, which is proof that the work they're putting in is paying off.

Workflow decides what survives

Every technology I've seen fail in a health system failed for the same reason: it didn't fit how people actually work.

A tool can look great in a demo and still add ten clicks to a nurse's shift. The organizations that do well will pick technology the way good engineers design systems. They'll start with the workflow, find where the friction is, and only then choose the tool. If it doesn't save time or improve a decision in the first few weeks, people quietly stop using it, however good the software is.

AI as an assistant

AI is already changing documentation, scheduling and image review, and it will go a lot further. The most useful AI in healthcare won't make decisions on its own. It will take on the tedious work, point out what a clinician might have missed, and stay out of the way.

Trust will limit how fast this goes. Clinicians will rely on AI when they can see how it works, when it's consistently accurate, and when someone is accountable for it. Health systems that build governance into the rollout, not after it, will move faster over time.

Value-based care needs the data

Population health and value-based contracts have been the stated direction for years. I've led population health and ACO technology work, and the contract was never the hard part. The data was: knowing which patients needed attention, proving outcomes, and connecting the care delivered to the results.

Continuous, objective data makes that connection possible. When outcomes are measured and not estimated, a value-based arrangement stops being a reporting exercise and becomes a better way to deliver care.

What I'd tell healthcare leaders

If you lead technology or operations in healthcare, here's what I'd do:

Spend a day with the people who will use a tool before you buy it. You'll learn more from the workflow than from the vendor demo.
Measure what matters and make it visible. Data that doesn't reach the clinician at the right moment doesn't change care.
Build trust into every rollout. Training, governance and honest feedback matter as much as the software.
Plan for operational resilience. The best systems make the right thing easy to do, even on the hardest days.

Technology isn't going to replace clinical judgment. What it can do is get better information to that judgment sooner, in a form people can use. That's the work I enjoy most, and I think it's where the next ten years of progress will come from.

Daniel Salzberg is a Richmond, Virginia healthcare technology executive and VP of Operations at BioMech Health. More at danielsalzberg.com.

Movement Is the Vital Sign We Aren't Measuring

Daniel Salzberg — Mon, 09 Nov 2026 16:28:46 +0000

Originally published at danielsalzberg.com.

Every clinic visit starts with the same few numbers: blood pressure, heart rate, temperature, breathing rate, oxygen level, and often a pain score. They take a couple of minutes, they're recorded the same way everywhere, and they go into the record where they can be trended, flagged and used by the systems that help with clinical decisions.

What's usually missing is how the patient moves. Can they stand up from a chair without using their arms? How fast do they walk? How steady are they when they turn? For a lot of patients, especially older adults, people recovering from surgery and people with chronic conditions, that tells you more about their health than almost anything else on the chart. We mostly capture it by eye, in a line of free text, or not at all.

This is the first of six posts in a series I'm calling Healthcare, Measured. The idea is to treat functional movement as a vital sign. Over the next few months I'll cover what it takes to measure movement reliably, how to do it at home, how to turn it into a data point people can trust, and how to give AI something better to work with than a note that says "ambulating well."

This isn't a new idea

Clinicians have been saying this for years. In 2009, Stacy Fritz and Michelle Lusardi published a white paper in the Journal of Geriatric Physical Therapy called "Walking Speed: the Sixth Vital Sign." In 2011, JAMA published a pooled analysis of nine cohort studies covering more than 34,000 older adults. It found that gait speed was associated with survival.

So we've known for a long time that function tells us something important. What has lagged is how we capture it.

Why it never became routine

Blood pressure became a vital sign because it's quick, it's standardized, and it ends up as a number. Functional measures mostly haven't had that. Timed walk tests and chair-stand tests exist, but they're usually done with a stopwatch and a clinician's eye, they're done differently at each site, and the result lands in a note or a scanned form. They also tend to live in therapy settings, not in primary care.

After years in health system IT, I've come to one simple rule. If a data point isn't captured in a structured, consistent way, it might as well not exist for anyone downstream. You can't trend it, nothing can alert on it, it won't show up in a population report, and no model can learn from it.

One reading misses the story

Even when movement is measured well, it's usually one reading taken in a clinic. Function changes from day to day. The walk in from the parking lot, how stressful the appointment is, even the time of day, all affect the result.

The changes that matter most, like a slow decline before a fall or a recovery that stalls after a knee replacement, happen between visits. A vital sign checked every few months will miss the trend. That's why measuring at home matters, and it's what the next post is about.

Why this matters for AI

Health systems are building and buying AI tools to predict readmissions, falls and deterioration. Those models run mostly on labs, vitals, diagnoses, medications and utilization history. Function is usually missing, or buried in notes where it's hard to use.

A model can't use what it can't see. Objective, repeatable, structured movement data would give those tools a new kind of signal, one tied closely to independence and to the outcomes patients care about. The data has to be good, though. A noisy measure doesn't improve a model. It makes the predictions worse and more confident at the same time.

What it would take

Turning movement into a vital sign doesn't need a scientific breakthrough. It needs four things. Standardized tests with the same protocol everywhere, so a result means the same thing in every setting. Objective capture instead of a stopwatch and a judgment call. Structured data in the record, next to the other vitals, where clinicians and systems can use it. And measurement at home as well as in the clinic, so you can see trends between visits.

That's the same discipline that made blood pressure a number everyone trusts.

What's coming in this series

Part 2: Bringing functional testing home
Part 3: What makes a movement measure a real biomarker
Part 4: Giving AI better inputs
Part 5: Getting function into the workflow and the record
Part 6: Paying for it and proving its value

If you run a health system or a clinical program, start by finding out where functional data sits in your record today, who captures it, and whether anyone can trend it. The answer usually shows how far there is to go.

Disclosure: I'm VP of Operations at BioMech Health, which works in objective movement measurement. This series reflects my own views and isn't about any specific product.

Daniel Salzberg is a Richmond, Virginia healthcare technology executive and VP of Operations at BioMech Health. More at danielsalzberg.com.